Andrew Chen Archives

Subscribe · Featured · Recent · The Cold Start Problem 📘
Dear readers, I have moved to Substack and I will be writing here from now on:
In the meantime, I will leave up for posterity. Enjoy!

Author Archive

Why investors don’t fund dating

without comments


I’ve been listening to the excellent Season 2 of the podcast Startup, which gives an inside look at YCombinator startup The Dating Ring (NYT coverage here). The episodes are all great. They talk about many important topics, but I had some specific comments on fundraising for dating products.

Here’s a simple fact: It’s super hard to get a dating product funded by mainstream Silicon Valley investors, even though it’s a favorite startup category from 20-something entrepreneurs. There’s a large swath of angels/funds who categorically refuse to invest in the dating category in the same way that many refuse to invest in games, hardware, gambling, etc. Perhaps they’d make an exception for a breakout like CoffeeMeetsBagel (I’m an advisor) or Tinder, but in the main, it’s an uphill battle for dating apps to attract interest. Here’s some data on the few dating cos that have raised.

Obviously, anyone starting a new company in dating should try to understand investor biases in this sector. This essay also compliments a previous one on operating, from HowAboutWe co-founder Aaron Schildkrout, now at Uber, who also wrote about his experiences.

Here are the reasons usually given for why investors don’t do dating:

  • Built-in churn
  • Dating has a shelf-life
  • Paid acquisition channels are expensive
  • City-by-city expansion sucks
  • Hard to exit
  • Demographic mismatch with investors

Let’s break it down.

Built-in churn
Churn sucks, and the better your dating product works, the more your customers will churn*. Every churned customer is a new customer you’ll have to acquire just to get back to even. When you look at a successful subscription service like Netflix or Hulu, you might find a churn rate of 2-5% per month, and you can calculate the annual churn through the following:

Annual Churn = 1-(1-churn_rate)^12
2% monthly churn = 1-(1-0.02)^12 = 21% annual churn
10% monthly churn = 1-(1-0.1)^12 = 70% annual churn

If you have an 70% annual churn rate, you have to have a strategy to replace almost your entire customer base each year, plus a bunch of percentage points to drive topline growth. You can imagine why successful public SaaS companies try to keep their monthly churn under 2%.

So what do the churn rates look like for a dating product? I’ve heard numbers as high as 20-30% monthly. Let’s calculate that:

20% monthly churn = 1-(1-0.2)^12 = 93% annual churn

You read that right. And that means at 20% monthly churn, it gets very hard to retain what you have, much less fill the top-of-funnel with enough new customers to grow the business. Scary.

With most subscription products, the more you improve your product, the lower your churn. With dating products, the better you are at delivering dates and matches, the more they churn! As you might imagine, that creates the wrong incentives. A product focused on casual dating, like Tinder, might escape this dilemma, but dating products generally have built-in churn that’s unavoidable.

Dating is niche and has a shelf-life
All this churn is especially complicated by the fact that the dating market at any given time is pretty niche. Similar to buying a car, refinancing your student loans, or moving into a new house, the reality is that being “in the market” as a single person looking to meet others has a limited time window. Another way to say this is the dating has “intent” the same way that shopping might, especially when you are talking about a paid subscription service. This limits the market size as well as restricting the types of marketing channels you can use to read those consumers.

A similar challenge is that these products aren’t “social” in the same way that Skype or Facebook might be. Although the stigma is quickly passing, it’s not like consumers want to sign up for a dating site and then invite their friends+family to join them on the site. In that way, it’s more similar to a financial or health product, where some privacy is required.

Again, one of the ways that the new generation of mobile dating products solve this is that they are free plus focus more on casual dating. Both factors open up the market to a wider audience, reduce churn, and create opportunities for viral growth.

Paid acquisition channels are expensive
Dating products have historically depended on paid acquisition channels to build their customer base, and other subscription products have generally done the same. In order to make the ROI work, you have to calculate your customer acquisition cost (CAC) versus your lifetime value (LTV) and make sure you are making enough money to support both the marketing as well as operations. In SaaS, you’d try to get a 3X ratio for CAC:LTV but that’s building in some profit for the company – a dating startup might be able to run it closer to the metal to get their initial growth.

Here’s a couple scenarios for products that buy their customers:

  • Make a ton of money all at once (example: car/insurance/loan/mortgage leadgen)
  • Make a little bit of money over a long period of time (storage, streaming music, etc.)
  • Make a little money at first, then grow the revenue over a long period of time (SaaS)

Here’s a visualization of this:



When you start to fill in this chart, you can see a couple things:


First, you’ll observe that of course the “ideal” case might look like a super low churn business that also generates a ton of revenue from each customer. However, the market size might be much smaller than the others. Christoph Janz, a venture capitalist and initial investor in Zendesk wrote a great essay on this topic, called Five ways to build a $100M business that talks about market size as an issue for this.

But back to dating- where does it go? The trouble is, it has some of the same economics for consumer subscription products priced at <$50/month, but at the same time, super high churn that looks like a one-time product. It’s hard to build a high LTV off of that, and so paid channels turn tricky.

Again, this is an area where the new mobile dating apps excel. They have the ability to tap into organic viral/word-of-mouth installs, are super sticky due to their messaging features, and free installs mean infinity return on ad spend (ROAS). At the same time, their focus on casual dating lowers churn and they can monetize via microtransactions. It’s a different model that’s more attractive.

City-by-city expansion sucks
Dating products inherently rely on a local marketplace, and bootstrapping a series of marketplaces is very hard, and expensive. People are willing to travel to meet each other, but only so much. And there needs to be the right mix of male/female participants (or whatever permutation makes sense). To make this work, each city needs to get spun up the same way that on-demand services are spun up, which is one of the reasons why local expansion has remained expensive and unscalable.

This is why we often see dating products continually hold a series of unscalable events/parties/etc in a city to get things going. Until there’s word-of-mouth, and enough people to generate a quality experience, the marketplace will suck. But it’s slow going.

Demographic mismatch with older, married investors
Dating solves a problem that’s most universal and acute for unmarried 18-35yos. Most investors who can write checks (as opposed to associates) are older, married, with kids. Oftentimes they haven’t had to date anyone in decades, unless you’re talking about Ashley Madison. Given this demographic mismatch, it’s a lot harder to get investors to put in the time to really understand the nuances of how one dating product is superior to another. This isn’t just a problem for dating, but also for women’s fashion or startups targeting international markets. It’s tricky, and investors would often rather sit back and wait for traction rather than investing on the merits of a product, something they’re willing to do in other categories. (Thanks to my friend Jason Crawford for adding this point)

Hard to exit
In the end, we’ve seen that dating products often end up being owned by IAC. They own Match, OKCupid, Tinder, HowAboutWe, and others. They have deep experience in local and dating, and the deep pockets to squeeze profit from the category. In contrast, we’ve seen recent dating cos like Zoosk withdraw their IPO plans. I don’t have any inside info, but I’m sure it’s because the churn was high, the channels were degraded, and it was hard to replace lost customers.

In the end, the lack of exits might be more the result, not the cause, of investor disinterest in dating. After all, given the challenges above, it’s very hard to maintain a steady customer base much less grow it consistently year-after-year.

A final note on this: Investor pattern-matching is lazy and often sucks. Google wasn’t the 1st search engine and it wasn’t clear the category was a good investment. Lots of failures. Same with Facebook or Whatsapp. But the investors who won were able to look at the specific characteristics of some of these new companies, rather than looking at the category, and bet that they’d figure it out. Very possible within dating as well, and I wish everyone who’s working in the category that they will be the ones to figure it out.

UPDATE: Fixed some math, thanks Hacker News commenters.


Written by Andrew Chen

May 26th, 2015 at 11:19 am

Posted in Uncategorized

Why we should aim to build a forever company, not just a unicorn

without comments


“Unicorn company.” It’s the latest bit of jargon that’s infected our conversations here in the Bay Area, to the point where both WSJ and Fortune have clever infographics and lists of the top companies. In pitches, entrepreneurs are asked to explain how their new company will become the next unicorn startup, and the tech press routinely debates if a hot new team will build the next unicorn. And yet, this term could not be a more meaningless goal for entrepreneurs.

After all, what’s the definition of a unicorn startup? Just one that reaches $1B in valuation? Who cares? I wish we’d just go back to saying “billion dollar startup” rather than unicorn, to reflect the real nature of the term, not one with a cutesy veneer.

It’s the ultimate vanity metric, because $1B of shareholder value is merely the lagging indicator that we’ve created something useful for the world. This should never, in itself, be the goal of starting up a company. So let’s all stop talking about unicorns. I’m calling peak unicorn. Let’s focus on the inputs for building impactful, lasting companies, where wealth creation is a side effect of doing a great job.

Instead, let’s talk about how to build our forever companies.

Low-attention spans in tech
When I first started out, as a young techie with a low attention-span living in Seattle, I had an irrational admiration for self-described “serial entrepreneurs,” the ones who build and sell a bunch of startups in their careers, even when they are quick flips. The variety of starting up multiple companies seems dreadfully exciting, especially when you are young and lack purpose. However, the more time I spend in the industry, the more my admiration shifts to those who start and run their companies for years, decades, and perhaps their whole lifetimes. Warren Buffett, Richard Branson, Jeff Bezos, Mark Zuckerberg, and others all fall into this camp.

These folks have started and built their forever companies. These companies also happen to be incredibly successful, but more importantly, as entrepreneurs they’ve found their life’s work.

After all, many of us in tech idolize Steve Jobs for his sense for design, and his vision. Some even emulate his fashion. But you know what’s hard to emulate? The fact that he started working on hardware/software products as a teenager, and built on those ideas for the next 40 years of his life, until he ran out of time. How many of us can profess a lifetime of dedication towards our work like that?

The forever company is an entrepreneur-focused counterbalance to the financially-motivated goal of becoming a unicorn. Hopefully we can build both! Of course we all want our companies to be valuable, and make a big impact, but while a unicorn concerns itself with the output of entrepreneurship, the goal of a forever company starts with the inputs and the right intentions.

This is different than a lifestyle company. Bezos runs Amazon as his forever company, but it’s certainly not just to support his lifestyle, it’s to make a much bigger impact than that. And Amazon took millions in venture capital money on their way to becoming a public company, to fully capture the opportunity. There’s a different kind of problem when the desire for a lifestyle interferes with the “forever” part of the goal. Those who underinvest in their products create the danger for a smarter/bigger/funded competitor to put them out of business, which is a lifestyle company that doesn’t last forever! This distinction is subtle, but important.

I didn’t coin the term. It’s the kind of idea that could only come out of a deep, late-night conversation with my sister and bro-in-law Ada and Sachin, who also work in tech. They mentioned it in passing as a worth goal for themselves, one day, and the term really resonated with me. It’s stuck like few things have, and I hope it sticks with all my readers too.

When forever companies scale, and when they don’t
Sometimes forever companies scale and become a multi-billion dollar company. In many cases, a forever company and a unicorn are the same, when the market is big, the team is talented, and there’s some good luck. These are the companies we all want to start and want to fund in Silicon Valley. These companies are easy to embrace.

But sometimes, a forever company just implies a lifetime of dedication towards something that may never get big. There is great honor in that as well, and I can’t help but admire those who pursue this goal. By now, we’ve all seen Jiro Dreams of Sushi, and his passion and skill for sushi is incredible. Closer to tech might be someone like David Kelley of IDEO, who founded his firm decades ago, and while they’ll never be a unicorn, I imagine he must be very proud of the work they’ve done over the past 25+ years.

Finally, I want to leave you with a great interview with Jiro I saw recently. He talks about feeling like a master only after reaching 50 years. The discussion on handmade versus automation fascinating, as well as the work ethic of younger generations. Everyone who’s working in design or engineering in software will relate, I’m sure.

I’ve embedded it below but if it doesn’t show, here’s the link to Vimeo. Enjoy.

Written by Andrew Chen

May 4th, 2015 at 12:51 pm

Posted in Uncategorized

Ten classic books that define tech

without comments


Today, I was asked for the definitive list of books that I’d recommend as the classics for tech products and business. It’s hard to pare things down to such a short list, since there’s so much great stuff that’s been written.

In addition, a lot of new exciting books have been published in recent years, but they haven’t stood the test of time. My lists solely consists of books that skew oldish, but have aged well and continue to provide value today. This leaves out amazing contributions such as The Hard Thing About Hard Things, Zero to One, and The Lean Startup, which will undoubtedly make a list like this in future years.

With that in mind, here’s my list. I’m sure you’ve read many of them, but hopefully you find a gem or two that you haven’t already read. In no particular order:

There’s also a number of great books that just tell the narrative of one company, such as The New New Thing, Startup, eBoys, Only the Paranoid Survive, Breaking Windows, but I’ve left those off the list even though they are very fun to read.

I’m sure I’ve missed some great ones! If you want suggest a replacement and/or share your own list, tweet me at @andrewchen.


Written by Andrew Chen

April 30th, 2015 at 2:50 pm

Posted in Uncategorized

How I first met Eric Ries and also why I’ve ordered his new Kickstarter-exclusive book The Leader’s Guide

without comments

Taken a few weeks ago at dinner, at Mission Rock in Dogpatch

It’s the last week to order Eric Ries’s new book, called The Leader’s Guide. In a very innovative experiment, it’s being published exclusively on Kickstarter. It’s the only way to buy a copy. I’ve already ordered a signed version and encourage you to support his work too. Here’s the link.

Making entrepreneurship mainstream
Like many of you, I’m a huge fan of Eric – he’s created a compelling, cohesive framework for thinking iteratively and entrepreneurially about products. The ideas are so powerful that the ideas in the book – such as “pivoting” and “MVP” – are now part of industry jargon and have even been featured on HBO’s Silicon Valley.  (Also, isn’t it inevitable that he makes a cameo at some point?) The ideas in Lean Startup are amazingly powerful, and I continue to reference them all the time.

How I first met Eric
Last month in March 2015, I hosted a dinner at Mission Rock in Dogpatch where he was a special guest, and we talked about how we first met back in 2008. Eric and I had our first coffee before The Lean Startup was a real thing, when he was between jobs and hanging out at Kleiner Perkins. (PS. if you’re interested in attending events like this in the future, subscribe to my newsletter and you’ll get email updates if I do this again sometime)

Anyway, here’s how I first got to know Eric- at that point, I was writing a niche, mostly unread blog about tech and products, at the end of my stint as an Entrepreneur-in-Residence at Mohr Davidow Ventures. I had maybe 100 readers total. It was thrilling to find another niche, mostly unread blog full of content I was interested in :) Looking at my referer traffic, I noticed I was getting a tiny bit of traffic from a blog called “Startup Lessons Learned.”

I clicked through, and my mind was blown…

First off, the blog was anonymously written. There were a ton of essays across a bunch of meaty topics – picking products, continuous deployment, landing pages, and metrics. It was obviously from someone who was on the front lines. I couldn’t tell who was writing it, but whoever it was, the content was amazing. I bookmarked it and added it to my Google Reader, and would diligently check for updates every week.

I just couldn’t believe that someone was writing this much great stuff without taking credit for it :) Eventually I found a cryptic email address from the blog, decided to write in to figure out who the hell was writing this amazing content. Soon after, I got a quick reply, and that’s how we first met.

A few months later, Eric told me a few months later he was going to leave Kleiner to work on a book. I was very confused :) Turns out that book he was working on was The Lean Startup, and it ended up doing pretty well!

Fast forward
The success of The Lean Startup has changed the culture of entrepreneurship across the world. And many new ideas, frameworks, and refinements have been built on top of the ideas presented in the original book. There’s a ton of lessons learned from trying to implement these ideas across a wide variety of industries – for example, at the aforementioned dinner event we heard from Eric on how people have tried to apply the ideas to non-tech industries like manufacturing, aerospace, as well as some of the nuances of applying MVPs in a design-centric world of consumer mobile apps.

Because of this, I was thrilled to hear that a lot of these case studies and ideas have been collected into a new book, The Leader’s Guide. I ordered my copy immediately upon hearing about it.

Here’s the link if you want to check it out too:
The Leader’s Guide, by Eric Ries.

Written by Andrew Chen

April 7th, 2015 at 4:01 pm

Posted in Uncategorized

This is what free, ad-supported Uber rides might look like. Mockups, economics, and analysis.

without comments


Cheaper and cheaper rides
Free, ad-supported Uber rides are inevitable, and if Uber doesn’t do them, a different competitor – perhaps Google! – will do it. It would be the next step in the industry’s trajectory towards lowering prices. Uber started in the high-end of the market, as “everyone’s private driver” which emphasized quality, but the early pricing was out of range for most. Dropping prices will increase demand, grow the market, and nothing will do that faster than going free.

Starting with UberX and now UberPool, each tier of service has dropped the price, tapping into more and more latent demand. Today, the figures are staggering – in San Francisco, Uber’s revenue is 3X bigger than the city’s entire taxi industry. Worldwide, the company now does 1 million rides per day and is adding 50,000 new drivers per month. Not bad at all.  I wrote a tweet about this concept a few weeks ago and wanted to expand more on this topic.

The next step is free. Lowering prices drive all sorts of goodness- more rides, busier drivers, shorter wait times – all which powers the company’s network effects. So how would you push the prices even lower than UberPool?

Well, here’s a crazy idea:

UberZero. A free, ad-supported tier of service that’s cheaper than UberPool and UberX.

That’s right, ads. I think it can be done well. And no, I don’t mean the cheesy in-taxi TVs that play ads nonstop, because those are broken for a variety of reasons I’ll cover. Instead, my proposal would be to put the ad units on your smartphone, in the dead time while waiting for your car and when you’re on your ride. And then to connect these ads with the big existing markets for app installs, lead generation, video ads, and local ads.

After all, advertisers are already comfortable paying a couple bucks for a variety of direct response actions from users. Individually, or together, you could imagine the above budgets making a big dent into the price of a single ride. In some cases, it might just be a discount of $1 or $2 off. In other cases, you might make the ride completely free.

Mockups for four potential ad units
Here’s four potential ad units that might work. These mockups were done in collaboration with the talented Chris Liu (@machinehuman), who’s done UX at Mercedes Benz, Alexa/A9, and IMVU. It was fun to throw together a couple designs.

Without further adieu, some of the concepts:

1) App Installs
The first and most obvious opportunity would be to tap into the ecosystem of paid app installs. Most of the $10B+ market for mobile ads is from driving app installs.

The user experience would be something like this- today, right after you book a ride, you often just sit there, staring at your phone, wondering how many minutes it takes for an Uber to travel a few blocks in San Francisco. Instead of just twiddling your thumbs waiting, how about getting a buck or two off your ride just from downloading and playing with an app?

Or similarly, imagine your current in-car experience. I often get into an Uber, pull out my phone, and do a little email during the ride. It would be easy to simply redirect my attention to an app, or a video ad, or some other activity that creates value for both myself and an advertiser.

It might look something like this:


As mentioned earlier, app developers are often paying $3-10 per install, especially in the games and commerce categories. This is a meaningful % when UberPool currently lets you go anywhere in San Francisco for $7 flat fee.

The biggest problem with this is that this is an incentivized install, which performs worse and Apple doesn’t like it. The plus side is that the folks who are really spending money in this area direct marketing-oriented and will back out the proper Cost-Per-Install to bid given all of that. As for Apple’s dislike of incentivized installs, you could certainly combine it with other ad units, and make the install a completely optional thing, which brings us to the next idea…

2) Video ads
One of the most compelling new ad formats for mobile is video, which are commanding high rates, especially when combined with a call to action to install an app. This is one of the many reasons why ad-supported companies like Twitter and Facebook are doubling down on in-stream videos.

So while you wait for your car to arrive, or once you get into the car, just watch a short clip that does a full-screen takeover of your phone. Then after you watch, the ad presents an optional action to download the app, which is a double dip on value. This packages both a brand ad with some a direct response ad unit, making for a powerful combination.

In Snapchat’s early tests on video advertising, Adweek and others have reported they are asking $750k, equivalent to a $100 CPM to participate. There’s no reason why Uber wouldn’t be able to command similar rates, especially for endemic advertisers in travel, shopping, and local.

It might look something like this:


The key problem here would be to provide context and targeting. A classic TV ad against a show like, say, Oprah, provides both context, targeting, and a brand halo effect. An ad to an anonymous passenger isn’t compelling until you know a little bit more about them.

This begs the question, what do we know about a passenger? What would make sense from a targeting standpoint? Targeting is interesting because it’s an anonymous segment of people, which you could make off a number of interesting values:

  • Name
  • Contact info (email/phone)
  • Location
  • Where they’re coming from and where they’re going
  • Work or personal credit card

Of course from a targeting perspective, you’d never pass on the personally-identifiable information to an advertiser, unless there was opt-in (more on that later!). But how compelling would it be to for example target people in a segment like the following:

  • Passengers heading to a movie theater
  • Passengers coming from SFO airport, heading to SOMA
  • Passengers with email addresses like and
  • Passengers traveling in a group of 2+ going out on a Friday night
  • Passengers heading home from work, or vice versa
  • Passengers heading to the Dreamforce conference
  • … and many more combinations

Now ask yourself, who else is able to deliver targeting ad inventory like this? There’s a very short list of companies that could even contemplate it. Of course it’ll be up to Uber as well as passengers to decide whether or not this kind of targeting criteria can be aggregated anonymously for ad targeting, but history has shown that ad targeting options tend to increase rather than decrease over time.

If you found the examples above compelling, imagine an instant rebate on your ride that relied on opt-in from the passenger to share this contact information directly. There’s a whole world of leadgen that’s willing to pay big bucks for accurate contact information for people who are in the market for targeted services and products.

3) Lead gen, especially for SaaS/B2B
I find leadgen an especially intriguing possibility because it’s commonplace for advertisers to pay something like $10 for an email address. For a SaaS/B2B advertiser, that number is even higher, as much as $30-50 in some cases. And if you’re talking about getting a credit card alongside an app install, that might be as high as $200 or more.

Of course you could never pass along a passenger’s contact info + credit card number without their consent. But what if you made it frictionless to consent? Then that previous list of attributes could all be sent in a series of confirmations: Name, contact info, payment information, etc.

At the heart of this is the fact that segmentation/targeting makes some audiences much more valuable than another. Today, a service like Uber charges a C-level executive the same price as the Average Joe. Imagine if you could have a leadgen campaign where an advertiser  decides that someone with a “” address is pretty valuable given that they’re a Google employee. Or perhaps it’s interesting to know that they are coming from a Salesforce conference.

You might provide a call to action like this:


This would make obvious sense for the high stakes world of SaaS/B2B leadgen, but it might also make sense to target a wide consumer base, especially around commerce. For example, targeting travelers who are arriving from the airport, to target them with highly personalized hotel/tour offers. Or targeting hardcore movie or concert-goers for their next night out.

Similar to the video ad scenario, the targeting capabilities would be key. Targeting based on location and context is key. And as I think about this, the most compelling part of the whole thing is that Uber/Lyft have data and context that is unique in the industry. They know where you are going, and where you are coming from, and enable you to go from Point A to Point B.

You know who else is like this and has created the most profitable and targeted advertising ever? Google, of course.

Google’s superpower is the Power of Redirection. When you search for something, Google can understand that intent and gently redirect you to the right place. Sometimes this is a destination that they deem relevant based on their algorithms, but sometimes, they just take you somewhere that’s paying them a bunch of money. This superpower is very valuable.

4) The Power of Redirection for local ads
Uber also has the redirection superpower, albeit in a different flavor. Similar to Google, when you plot a destination, there’s intent built into that. All the examples I’ve used previously indicate this- heading to Facebook’s HQ means something. Leaving from SFO means something. Heading to the 49ers stadium on a Sunday means something. And it would be easy for Uber to take you to your destination, but maybe they can suggest somewhere else to go, or perhaps they can suggest a complimentary product/service.

And using the analogy of organic versus paid search, perhaps there are organic destinations and there’s also paid destinations. In the case where a ride can be redirected to a paid destination, then Uber can literally, actually drive foot traffic. Pretty amazing!

For a trivial example, you might imagine something like this, that redirects you to a nearby Starbucks as a detour on your ride:



(The second mockup that bundles in restaurant reservations was sent in by Marc Köhlbrugge, founder of BetaList – thanks!)

You could imagine a clear application for travelers of course. Heading to the W Hotel? Why don’t you try an Aloft hotel and get $X off your stay. Think about how compelling this is compared to other forms of getting local foot traffic. Billboards? Radio? Well, in this case you can literally get someone to your footstep.

I think this could be fundamentally a new market, although it would make sense to tie this back to the $4.5B mobile/local ad market. Or you could tie it back to the unit economics of something like pay-per-call, where local vendors are willing to buy a live phone call with a customer for $10. Nevertheless, still super compelling when you are talking about tens of billions of local ad dollars moving to mobile over the next decade.

Lots of practical challenges
But wait, could you really make all rides free all the time? Maybe not. After all, there’s aren’t infinite drivers. Won’t this make the ride UX horrible? Maybe, which is why you’d have to be thoughtful with the design. Another issue is that advertisers won’t pay for passengers to download the same apps over and over again – they’ll want frequency caps. And Apple doesn’t like incentivized installs, or maybe passing PII to advertisers will be frowned upon by the press. All very real issues that will have to be worked out over time.

And in the end, maybe it’ll turn into free* with an asterisk, limited to 1 per day within your city and only in the morning. Or just another way to decrease fares by another 25%. Whichever way it’s done, I’m convinced it will be done, because the benefits are too great.

Isn’t this perfect for Google?
If Uber doesn’t do this, perhaps Google will. There’s been rumors that they’re working on some kind of Uber-competitor, and they already have all the advertisers to make the above ad units work.

Providing free, ad-supported rides would certainly be a unique way to enter the market. Even more so, there’s network effects. Ad networks are fundamentally marketplaces, and they have tremendous network effects at scale. You could see the following virtuous cycle:

  • Free rides = way, way more rides
  • More rides = more advertising reach
  • More reach = more interesting targeting options, and drivers
  • More advertisers = more profitable yet free rides
  • Which leads to more drivers, customers, and more

At scale, a network like this would be very hard to attack, especially when combined with existing ad businesses like the type that Google possesses. Thus, if we see Google launch an Uber-competitor, I think it’d be a matter of time before they experimented with this.

Why old in-taxi advertising systems suck
The final point I want to make is to contrast all of these ideas with the old in-taxi ad systems we used to see. If you need your memory jogged, they used to look something like this:


These systems suck. And they can’t generate the kind of economics to power the subsidy we’d want to see to make a dent on the actually fare price. The point isn’t to actually make money via advertising- the point is to drive down the cost of a ride so that the market is more efficient and liquid, creating network effects.

These old advertising displays are archaic:

  • Passive advertising content
  • Doesn’t know anything about you or your trip
  • Can’t get you to download an app
  • Limited internet connectivity

These systems are ineffective because ultimately, this display belongs to the taxi and not you. As a result, it lacks some of the key ingredients that an advertiser finds valuable: an easy way to get your contact info, or to get you to install an app on your phone. It’s good for display untargeted video ads and not much else.

That’s why Uber sits on a special opportunity to build an ad platform that’s never existed before. It’d be an ad platform that has unique access to context, intent, and built on your personal device. If it’s done well, advertisers will love it and consumers will be grateful to have free/discounted rides.

Yet another reason to be bullish about the company, in addition to all the great momentum they already have.

(Special thanks to Chris Liu collaborating on these great mockups that really make the discussion in this essay pop!)

Written by Andrew Chen

March 30th, 2015 at 9:56 am

Posted in Uncategorized

Personal update: I’ve moved to Oakland! Here’s why.

without comments

My new neighborhood, in Jack London Square, Oakland, CA.

Where’d you move?
I moved to a tiny neighborhood called Jack London Square in Oakland. Yes, it’s named for that guy that wrote about the gold rush. I’ve only been here 3 months but I really like it so far. Previously, I lived in Palo Alto for 5 years, then about 2 years in the Lower Pac Heights neighborhood of San Francisco, but had never really spent much time in the East Bay. I had sort of heard that people were moving from SF to Oakland, but didn’t really have a reason to check out the neighborhood until a few people I know moved here.

Here were some of the articles I read while doing research:

PS. if you call Oakland “the next Brooklyn” to people who’ve lived here for a long time, they don’t like it :)

I live near there too! / How do I find out more about it?
If there’s interest, I’ll host a tech get-together or two.

Sign up here to get updates on an upcoming brunch/drinks/dimsum/whatever in Oakland.

There aren’t too many tech people here, so it’d be fun to get the small community that is out here together.

Where is it relative to San Francisco? How’s the commute?
I travel to the city pretty much every day. I usually take the BART, and sometimes the ferry (it has wifi!).

There’s a couple ways to get to the city:

  • BART (10min walk + 20min BART)
  • Ferry ride (25min ride + walk from Ferry building)
  • Car (30min without traffic, 60min+ with traffic)

I used to live in the Mission, and going from 24th+Mission to SOMA is about comparable to my current commute. However, I occasionally do have the morbid fear that there’ll be an earthquake while I’m underwater in the train.

Screen Shot 2015-02-24 at 2.07.42 PM

Where’s all the good food?
Right now, the Uptown neighborhood is opening the most new/amazing restaurants, where you can eat before you go to the Fox Theater or the Paramount for a show. Jack London Square has great food as well – there’s an eclectic mix of fancy pizza shops, vegan, and southern. A quick walk into Chinatown provides an endless supply of cheap eats, and the Oakland Chinatown is huge – about 2x the size of the city’s, without the tourist stuff.

A quick map search shows you where all the food is- pretty much in the Broadway/Telegraph corridor, but Rockridge, Temescal, and Grand Lake do well too.

Doesn’t Oakland have a ton of crime?
Crime was one of my top concerns moving to Oakland, but if the spectrum in San Francisco is aggressive/disturbed people in the Tenderloin to the nicest part of Presidio Heights, I think Oakland is about the same. You wouldn’t want to walk around Market St at 3am and you wouldn’t want to do that on Broadway in Oakland either. (Palo Alto / Menlo Park / Atherton are on a whole other planet, of course)

The biggest lesson I’ve learned from exploring the long list of East Bay neighborhoods is that Oakland is very diverse, and while the crime factor is a big one, it’s an acute problem for some neighborhoods and less of a problem for others. So, it all depends (just like SF, btw).

Houses in the Oakland Hills look like the kind of fancy houses you’d see while driving on 280 in the peninsula. Some neighborhoods like Rockridge, Grand Lake, and Adams Point are small and upscale, not unlike University Ave in Palo Alto. Uptown/Downtown feels like Market Street in San Francisco, but inexplicably cleaner. My new neighborhood, Jack London Square, feels a bit like South Beach in SOMA.

On the other hand, neighborhoods in deep East Oakland don’t feel very safe. That’s where you can find the car sideshows on YouTube.

Is it cheaper to live there?
For now, buying or renting seems to be about 50-75% the cost of San Francisco. Maybe as low as 30% if you are adventurous.

Is Oakland really warmer than San Francisco?
Yep. Sort of like the peninsula, up to 10 degrees warmer. Sometimes I miss the fog.

Here’s a typical day on the waterfront in Jack London Square.


Where do you go for coffee?
The headquarters for Blue Bottle Coffee is here. Yes, there’s usually a line.

But there’s also a bunch of other coffee places too:

Interestingly enough, the density of tech in Oakland is still relatively low. Cafes aren’t full of tech bros with terminal open. Coworking spaces are more likely to be nonprofits, writers, and sales, rather than unpronounceable names of startups. I’m sure a bit of this might change over time, and there’s been rumors of one of the big cos taking over the old Sears building in downtown.


How do I dress when I visit Oakland?
Like this video.

What’s the best way to visit Oakland?
The first step is to come out here by car/BART/ferry and check it out. I’d encourage you to do it, I think you’ll be surprised by how nice it is. And as I said above, if you’re interested in attending a casual get-together in the new neighborhood, or if you already live around here, just sign up on this mailing list and I’ll post some future updates.

Written by Andrew Chen

February 24th, 2015 at 3:42 pm

Posted in Uncategorized

The most common mistake when forecasting growth for new products (and how to fix it)

without comments

Forecasting weather is hard, and so is forecasting product growth.

Startups are about growth
Paul Graham’s essay in 2012 called “Startup = Growth” makes a big point in the first paragraph:

A startup is a company designed to grow fast. Being newly founded does not in itself make a company a startup. Nor is it necessary for a startup to work on technology, or take venture funding, or have some sort of “exit.” The only essential thing is growth. Everything else we associate with startups follows from growth.

The other important reason for new products to focus on growth is simple: You’re starting from zero. Without growth, you have nothing, and the status quo is death. Combine that with the fact that investors just want to see traction, and it’s even more important to get to interesting numbers. In fact, later in the essay, pg talks about how important it is to hit “5-7% per week.”

Getting to this number while trying to show a hockey stick leads to a bad forecast. Here’s why.

The bad forecast
The most common mistake I see in product growth forecasts looks something like this:

Image 2015-02-18 at 7.16.38 PM

In this example, the number of active users is a lagging indicator, and if you multiply this lagging indicator of a growth curve, it’s a truism that the growth will go up and to the right. If you do that, the whole thing is just a vanity exercise for how traction magically appears out of nowhere.

And of course these growth curves look the same: They all look like smooth, unadulterated hockey sticks. The problem is, it’s never that easy or smooth. In reality, you’re upgrading from one channel to another, and in the early days, you do PR but eventually that doesn’t scale. Then you’ll switch to a different channel, which takes some time but also eventually caps out. Eventually you’ll have to pick one of the very few growth models that scale to a massive level.

The point is, incrementing each month with a fixed percentage hides the details of the machinery required to generate the growth in the first place. This disconnects the actions required to be successful with the output of those actions. It disassociates the inputs from the outputs.

In other words, this type of forecast just isn’t very useful. Worse, it lulls you into a false sense of security, since “assume success” becomes the foundation of the whole model, when entrepreneurs should assume the opposite.

Create a better forecast by focusing on inputs, not outputs.

How to fix this forecast
A more complete model would start with a different foundation.

It would:

  • Focus on leading indicators that are specific to your product/business – not cookie cutter metrics like MAU, total registered, etc.
  • Start with inputs not lagging vanity metrics
  • It’d show a series of steps that show how these inputs result in outputs
  • And, how the inputs to the model would need to scale, in order to scale the output

In other words, rather than assuming a growth rate, the focus should be deriving the growth rate.

If you plan to 2X your revenue for your SaaS product, which is done by doubling the # of leads in your sales pipeline, and those leads come from content marketing – well, then I want to know how you’ll scale your content marketing. And how much content needs to be published, and whether that means new people have to be hired.

That also means that if you want to 2X your installs/day, and plan to do it with invites, I want to understand the plan to double your invites or their conversion rates.

Or better yet, say all of this in reverse, starting with the inputs and then resulting in the outputs.

Inputs are what you actually control
Focus on the inputs because that’s what you can actually control. The outputs are just what happens when everything happens according to plan.

One helpful part of this analysis is that it helps identify key bottlenecks. If your plan to generate 2x in revenue requires you to 5X sales team headcount when it’s been hard to find even one or two good people, you know it’s not realistic. If your SEO-driven leadgen model assumes that Google is going to index your fresh content faster and with higher rank than it’s ever done, then that’s a red flag.

In the end, it’s also true what they say:

No plan survives contact with the enemy.
smart prussian army guy

Keep that in mind while you fiddle around with Excel formulas, and you’ll be in good shape.

Written by Andrew Chen

February 23rd, 2015 at 10:30 am

Posted in Uncategorized

The race for Apple Watch’s killer app

without comments


The upcoming race
As the release of the Apple Watch draws near, we’re seeing press coverage hit a frenzied pace – covering both the product, the watch’s designers, sales forecasts, and the retail displays. That’ll be fun for us as consumers. But for those of us who are in the business of building new products, the bigger news is that we have a big new platform for play with!

The launch of the Apple Watch will create an opportunity to build the first “watch-first” killer app, and if successful, it could create a new generation of apps and startups.

Why new platforms matter – the Law of Shitty Clickthroughs
Regular readers will know that I’m endlessly fascinated by new platforms. The reason is because of The Law of Shitty Clickthroughs, which claims that the aggregate performance of any channel will always go down over time, driven by competition, spam, and customer fatigue.

When you have a big new platform, you avoid all of this. So it’s not surprising that every new platform often leads to a batch of multi-billion dollar companies being minted. With mobile, it was Uber, Whatsapp, Snapchat, etc. With the Facebook platform, we saw the rise of social gaming companies like Zynga. With the web, we had the dot com bubble. It’s very possible that wearables, led by the Apple Watch, could be that big too.

With the Apple Watch, we have fresh snow:

  • Right after the launch, there’s a period of experimentation and novelty, where people are excited to try out new apps, no matter how trivial
  • A barrage of excitement from the tech and mainstream press, which will publicize all the big apps adding integration
  • A device built around interacting with notifications and “glances” which, along with the novelty effects, will cause engagement rates to be ridiculously high
  • The app store which will promote apps that integrate with the Watch in clever ways
  • Unique APIs and scenarios in health, payments, news, etc., leading to creative new apps in these categories

At the same time, there will be less competition:

  • Many apps will take a “wait and see” approach to the platform
  • Some teams won’t try at all Apple Watch, since it won’t be easy to jam their app’s value into a wearables format – for example, you can’t just cram any game on there
  • The best practices around onboarding, growth, engagement still have to be discovered – so there’s a higher chance someone new will figure it out

The above dynamics mean that the Watch launch will lead to some exciting results. Apple has been thoughtful and extraordinarily picky about bringing out new products, so with the Watch, we know they’ll put real effort and marketing prowess behind it. Combine that with the rumored ramp up to millions of units per month, and you can imagine a critical mass of high-value users forming quickly.

What kinds of apps will succeed? It’s hard to answer this question without looking at what you can do with the platform.

The Human Interface Guidelines is worth a skim
Beyond the ubiquitous buzz stories that have been released, it’s hard to have a nuanced discussion about the Apple Watch until you really dig into the details. Here to save us are two documents:

Both documents offer some tantalizing clues for the main uses for the Watch, as well as the APIs offered by Apple for developers to take advantage of. The HIG document is particularly enlightening. Going through the screenshots, here are the apps that are shown via screenshot:

  • Visual messaging
  • Weather
  • Stock ticker
  • Step counter
  • Calendar
  • Photo gallery
  • Maps
  • Time, of course :)


For the most part, this is exactly what you’d expect. These are all apps that have existed on the phone, and the Watch serves as an extra screen. I’m sure this will only be the start.

The more interesting question is what the new Watch APIs will uniquely allow.

Apple Watch will supercharge notifications
One of the biggest takeaways in reading through the HIG is the prominence of the notifications UI. Although you might find yourself idly swiping through the Glances UI to see what’s going on, it seems most likely that one of the most common interactions is to get a notification, check it on your watch, and then take action from there. This will be the core of many engagement loops.

For that reason, Apple has designed two flavors of notifications – a “short look” that is a summary of the new notification, and a “long look” that’s actually interactive with up to 4 action buttons. Here’s a long look notification:


Because it’s so easy to check your watch for notifications, and you’ll have your watch out all the time, I think we’ll see Apple Watch notifications perform much better than push notifications ever have. Combine this with the novelty period around the launch, and I think we’ll see reports of much higher retention, engagement, and usage for apps that have integrated Watch, and these case studies will drive more developers to adopt.

Waiting for the Watch-first killer app
Succeeding as a Watch-first app remains a compelling thought experiment. We saw that after a few years of smartphones, the question “Why does this app uniquely work for mobile?” is an important question.

Apps that were basically ports of a pre-existing website ended up duds – crammed with features and presenting a worse experience than just using the website. Contrast that to the breakthrough mobile apps that take advantage of the built-in camera, always-on internet, location, or other APIs available. Said another way, many flavors of “Uber for X” have failed because it’s unique to calling a taxi to constantly need to consume the service in new/unknown locations, and with high enough frequency for this consumption. Not every web app should be a mobile app. In the same analogy, the majority of apps in the initial release of the Watch may take it to simply be a fancier way to show annoying push notifications, and drive usage of the pre-existing iPhone app.

The more tantalizing question is what apps will cause high engagement on the Watch by itself, with minimal iPhone app interaction? That’s what a Watch-first killer app will will look like. I’m waiting with a lot of excitement for the industry to figure this out.

Good luck!
For everyone working on Watch-integrated apps, good luck, and I salute you for working to avoid the Law of Shitty Clickthroughs. If you’re working on something cool and want to show me, don’t hesitate to reach out at @andrewchen.

Written by Andrew Chen

February 17th, 2015 at 2:48 pm

Posted in Uncategorized

My top essays in 2014 about mobile, growth, and tech

without comments

Hello readers,

Happy 2015 and hope everyone had a relaxing holiday. As always, more essays are coming soon for the next year. In the meantime, I wanted to share some of my essays published here over the last year. If you want to stay up to date, just make sure to subscribe for updates.

Also, thanks to the SumoMe team (specifically Noah Kagan!) for supporting my hosting costs. I’ve used SumoMe since the product was in alpha and find it immensely helpful in building an audience.


The essays
Without further ado, here’s a selection of essays published in 2014:

Make content creation easy: Short-form, ephemeral, mobile, and now, anonymous

How to design successful social products with 3 habit-forming feedback loops

How to solve the cold-start problem for social products

Why consumer product metrics are all terrible

There’s only a few ways to scale user growth, and here’s the list

Why aren’t App Constellations working?

New data shows up to 60% of users opt-out of push notifications

Early Traction: How to go from zero to 150,000 email subscribers

Why Android desperately needs a billion dollar success story: The best new apps are all going iPhone-first

New data on push notifications show up to 40% CTRs, the best perform 4X better than the worst

Mobile retention benchmarks for 2014 vs 2013 show a 50% drop in D1 retention

Why messaging apps are so addictive

If you want more, here’s the link to subscribe to future updates:

Written by Andrew Chen

January 6th, 2015 at 12:30 pm

Posted in Uncategorized

Why messaging apps are so addictive (Guest Post)

without comments

[Andrew: This guest post is written by my friend and former Palo Alto running partner, Nir Eyal. Messaging apps have been a fascinating area within mobile, and the big reason for it is that the metrics – especially engagement – have been amazingly strong. I asked Nir to write a bit about why this might be the case, using the habit model. He’s in a unique position to comment on this: After spending time at Stanford and in the startup world, recently he’s been writing about psychology – particularly habit-building – and applying that to the realm of technology and products. Check out his new book Hooked: How to Build Habit-Forming Products and Nir blogs about the psychology of products at]

Nir Eyal, Author of Hooked:

Today, there’s an app for just about everything. With all the amazing things our smartphones can do, there is one thing that hasn’t changed since the phone was first developed. No matter how advanced phones become, they are still communication devices — they connect people together.

Though I can’t remember the last time I actually talked to another person live on the phone, I text, email, Tweet, Skype and video message throughout my day. The “job-to-be-done” hasn’t changed — the phone still helps us communicate with people we care about — rather, the interface has evolved to provide options for sending the right message in the right format at the right time.

Clearly, we’re a social species and these tech solutions help us re-create the tribal connection we seek.  However, there are other more hidden reasons why messaging services keep us checking, pecking, and duckface posing.

The Hook

In my book, Hooked: How to Build Habit-Forming Products, I detail a pattern found in products we can’t seem to put down. Though the pattern is found in all sorts of products, successful messaging services are particularly good at deploying the four steps I call, “the Hook,” to keep users coming back.

The Hook is composed of a trigger, action, variable reward, and investment.  By understanding these four basic steps, businesses can build better products and services, and consumers can understand the hidden psychology behind our daily technology habits.


Nir Eyal’s Hook Model


A trigger is what cues a habit. Whether in the form of an external trigger that tells users what to do next (such as a “click here” button) or an internal trigger (such as an emotion or routine), a trigger must be present for a habitual behavior to occur.

Over time, users form associations with internal triggers so that no external prompting is needed — they come back on their own out of habit. For example, when we’re lonely, we check Facebook. When we fear losing a moment, we capture it with Instagram. These situations and emotions don’t provide any explicit information for what solution solves our needs, rather we eventually form strong connections with products that scratch our emotional itch.

By passing through the four steps of the Hook, users form associations with internal triggers. However, before the habit is formed, companies use external prompts to get users to act. For messaging services, the external trigger is clear. Whenever a friend sends a message via WhatsApp, for example, you see a notification telling you to open the app to check the message.


WhatsApp’s External Trigger


Notifications prompt users to act, in this case tapping the app. The action phase of the Hook is defined as the simplest behavior done in anticipation of a reward. Simply clicking on the app icon opens the messaging app and the message is read.

When the habit forms, users will take this simple action spontaneously to alleviate a feeling, such as the pang of boredom or missing someone special. Opening the app gives the user what they came for — a bit of relief obtained in the easiest way possible.

Variable Reward

The next step of the Hook is the variable rewards phase. This is when users get what they came for and yet are left wanting more.

This phase of the Hook utilizes the classic work of BF Skinner who published his research on intermittent reinforcement. Skinner found that when rewards were given variably, the action preceding the reward occurred more frequently. When forming a new habit, products that incorporate a bit of mystery have an easier time getting us hooked.

For example, Snapchat, the massively popular messaging app that 77% of American college students say they use every day, incorporates all sorts of variable rewards that spike curiosity and interest. The ease of sending selfies that the sender believes will self-destruct makes sending more, shall we say, “interesting,” pics possible. The payoff of opening the app is seeing what’s been sent. As is the case with many successful communication services, the variability is in the message itself — novelty keeps us tapping.


You never know what you’ll see when you open Snapchat


The final phase of the Hook prompts the user to put something into the service to increase the likelihood of using the service in the future. For example, when users add friends, set preferences, or create content they want to save, they are storing value in the platform. Storing value in a service increases its worth the more users engage with it, making it better with use.

Investments also increase the likelihood of users returning by getting them to load the next trigger. For example, sending a message prompts someone else to reply. Once you get the reply, a notification appears and you’ll likely click through the Hook again.


Growing a “buddylist” on Snapchat is an investment in the platform

Through frequent passes through the Hook, user preferences are shaped, tastes are formed, and habits take hold. Messaging services are here to stay and we’ll most likely see many more iterations on the theme as technological solutions find new ways to bring people together. By understanding the deeper psychology of what makes us click by knowing what makes us tick, we can build better products and ultimately live better lives.

Written by Andrew Chen

November 4th, 2014 at 10:30 am

Posted in Uncategorized

Why Android desperately needs a billion dollar success story: The best new apps are all going iPhone-first

without comments


Why startups are all going iPhone-first
There’s been a number of articles over the last year that reiterate a simple fact: The best new apps are all going iPhone-first. Here’s three popular articles on this topic over the last few months:

At this point, going iPhone-first is a widely held best practice. It’s our generation’s version of “Nobody ever got fired for buying IBM.” While there’s some debate on the margins on how quickly you should follow up with an Android app, certainly no one is arguing for Android-first (meaning, don’t do iPhone). There’s a number of reasons for this consensus, which the above articles thoroughly explain- here’s the superset of the reasons they give.

  • Device fragmentation – both OS versions, carrier/handset add-ons, and hardware itself
  • Less advanced, less stable tools and documentation
  • Larger install base, but smaller addressable market (Feldman states that a 50% market share really translates to 12% if you support the most recent versions of Android, versus 30% for iPhone)
  • Less valuable audiences on Android, losing ground in the US in key demos
  • Cheaper iPhones may steal marketshare from Android in the future
  • Higher cost of development for Android (2-3X claims Steve Cheney)
  • $800k-$1.2M seed rounds leading to a “all-in on one platform” strategy

Note, I’m not saying I agree with above, just summarizing what’s been said.

In addition to this, I’d also like to add a couple more human aspects of the decision:

  • Many/Most startup founders and employees are Steve Jobs fanboys, carry iPhones, and want to design for themselves
  • Their friends carry iPhones, and they want to make something for their friends
  • There’s more hipster designery mobile developers who build for iPhone, and that’s the supply-side of talent in SF Bay Area
  • Investors (other than Bubba Murarka) usually carry iPhones, so it’s easier to pitch to them
  • More tech press outlets want to cover iPhone-specific news

Whether you agree if the above is sane or not, the reality is, there’s a lot of friction to going against the norms. In order for a whole class of developers to move en masse to the Android platform is going to require a big carrot. I’m going to argue that this big carrot is going to be that the best developers, the ones who are investing $1M+ into a single app, need to feel like there’s such a huge opportunity in Android that they can’t miss out.

What Android can learn from Microsoft Windows
This consensus towards iPhone-first is happening at a critical time for Android. In many ways, the platform has been a huge success, and many who lived through the Windows vs. Mac years could make some interesting comparisons.

Here’s a stab at it- here’s some of the key reasons why Microsoft Windows won:

  • Cheaper
  • Ubiquitous
  • More open
  • Better penetration into the workplace
  • Lots of applications that were exclusive to the platform

I’d argue that on almost all the point above, Android has achieved the same success as Windows. It’s cheaper, there’s more devices sold, it’s more open. But a critical component, of having more apps, isn’t there. Remember how there was always some key games that’d run on Windows that wouldn’t exist on Mac? Or how there were a bunch of business applications that would only run on Windows? That meant that the Windows platform had the virtuous cycle between developers and users to drive total domination.

But Android is not Windows. When you look at the current mobile ecosystem, iPhone has more apps. It has better apps. It gets the designery, well-funded startups to build iPhone-first.

Consumers and developers, together, will continue to choose the iPhone until that network effect is broken.

Today, Android is merely playing catchup – every time there’s a proprietary iPhone app, soon thereafter, they’ve done a good job convincing developers that they also need to release an Android app. Yet, to play to win, Android needs to convince many, many developers to create apps exclusively for their platform, just like Windows did a generation ago.

How does Google get there?
To me, the biggest thing that Google is lacking is a billion dollar tech startup story that’s exclusively about how Android is a better platform for developers. That story doesn’t exist, and as a result, people have focused more on the friction in going Android-first, rather than the opportunity.

IMHO, Android gets there by rewarding startups that are exclusively choosing its platform. I’m talking my book here, as I’m involved with a few Android-first products, but it’s also nevertheless from direct knowledge. Google should extensively feature apps that aren’t merely clones of iOS apps – it’s not enough to play catchup. Instead, Android should seek to really showcase apps that take advantage of very differentiated features and APIs. There should be a billion user success story around Android launchers and lock screens in the US, rather than acquihires/flips of Aviate, Emu, Cover, and others. These were solid companies led by good teams that wanted to go Android-first, but there was a missed opportunity in supporting them.

App store discovery for iOS is a glaring weakness. It’s editorially driven, and doesn’t give great new apps the chance to be successful, which is why four companies own 70% of the top apps – Google, Facebook, Yahoo, and Apple. They’ve created a winner-take-all environment, especially as Facebook has shown how to effectively use App Constellations to drive traffic between apps. Instead, Google could create a much more fluid ecosystem, which would reward and boost apps the way that search has driven traffic to millions of long-tail websites. Everyone does SEO because they know that yes, you can create a public company like Yelp, or a fast-growing startup like Genius, on the Google Search platform. There’s no story like this in mobile.

And yes, shifting installs from the “head” developers into the long tail might make it less attractive for some, but the biggest developers will always develop for both – they don’t have a choice. The battleground is for the hearts, minds, and product roadmaps of new, innovative apps that will drive adoption for their parent platform.

Android is an important platform, and it’s built more closely to the open nature of the internet, and so for that reason I’m rooting for it. But to make it the first choice for developers, it needs to do more than it is.

Do you work at Google?
Finally, if anyone at Google is reading this, email me: voodoo [at] gmail. Like I said, I’m happy to talk my book :)

Written by Andrew Chen

August 27th, 2014 at 10:27 am

Posted in Uncategorized

Why aren’t App Constellations working? (Guest Post)

without comments

[Recently, I read this roundup of perspectives on “App Constellations” on the new social/professional news app Quibb. This is an emerging product strategy in mobile embraced by Facebook, Linkedin, Foursquare, Twitter, and others, and found it fascinating. Thanks to the authors below for sharing their opinions on this new approach. -Andrew]


Fred Wilson recently coined the term ‘App Constellations‘ to describe what he was seeing in the mobile app ecosystem with respect to distribution. Over the past few months, mobile companies have continued with this strategy, the idea being that standalone apps are better for mobile, versus the all encompassing platforms that dominate desktop experiences – and it’s a better position to own and control several of these key apps on any person’s device.

It has been interesting to watch as Dropbox, Facebook, and Foursquare experiment with this approach – but it doesn’t seem to be working, at least not yet. CB Insights shared some discouraging data last month, and things have only gotten worse over the past few weeks for the most recent apps to be put out by these big players:

Screen Shot 2014-07-11 at 2.12.32 PM

Screen Shot 2014-07-11 at 2.13.47 PM

Screen Shot 2014-07-11 at 2.19.31 PM

Screen Shot 2014-07-11 at 5.01.00 PM

It’s important to note that none of the companies using this strategy have promoted their unbundled apps aggressively – beyond Facebook Messenger, which is seemingly the only app where this App Constellation approach is paying off:

Screen Shot 2014-07-11 at 2.16.50 PM

Toufeeq Hussain (Senior Product Manager at Storm8)

The App Constellation strategy works when you have a core resource which can be shared across multiple apps. Slingshot and Poke are attempting to create a new resource (reply-to-view-images for Slingshot and disappearing images for Poke) and hence isn’t really leveraging whats core to Facebook (social graph and shared photos). In the case of Slingshot, even the social graph had to recreated from scratch. So even though these apps get huge media attention when they launch, they slowly slide down the charts as there is nothing holding them up.

One company that has done a great job of using core resources and creating a “basket of apps” is Evernote. It currently has Evernote Hello, Evernote Food, Skitch and Penultimate. Each of these are focused around helping users create more notes and thereby getting more usage of the core Evernote product. Of course, not all of Evernote’s apps are in the top-50 lists but they are targeted mainly at new users who are looking for more specific solutions to note taking. Carousel is on the same lines but my feeling is that Dropbox needs more apps that read data stored in Dropbox than contributing to it. If the goal is to contribute to the Dropbox then it needs to be something more than just syncing photo/videos as the default Dropbox app already performs that.

The Asian messenger apps use the contact list as the core resource across any apps integrated into LINE or WeChat. Invitations are sent through the core messaging app but specific functions are performed in the corresponding apps. Games, photo sharing, stickers all have specific apps but use the core LINE or Wechat identity and social graph to seamlessly work across multiple apps.

In conclusion, though its very valuable for a Facebook or Dropbox to shoot for “stars” and build constellations, what we have seen from companies like Evernote and the asian messenger apps (LINE, WeChat) is that a “basket of apps” approach that leverages a common core resource between other apps might actually be a more scalable strategy. Its very hard engineering “stars” – a lot of things need to fall in place to be a top-50 free app, a better strategy for Facebook might be to play to its strengths than alienating new apps from core FB resources.

Bubba Murarka (Managing Director at Draper Fisher Jurvetson)
App constellations are being deployed because the problem of distribution on mobile is “solved” in the sense that large incumbent app owner, and mobile marketers with sufficient resources, can predictably drive installs of their other apps via well known set of steps that includes cross promotion, cross linking, merchandising on mobile web & of course paid distribution. One of the best examples of this is how Facebook has managed to drive massive number of installs of the standalone Messenger app via the main Facebook integration. I would go so far as to say that if an unbundled app from a large incumbent does not to have a massive install base it is intentional to enable the app to mature before investing in driving distribution. FWIW, I am not saying distribution for younger companies without existing massive install bases, well known brand names or meaningful financial resources is “solved” yet
because, well it is not.

As for the value of this strategy it is easy to say it doesn’t work but that is not nuanced enough analysis of why companies pursue this approach. The multi app strategy allows more experimentation, different release cycles and tailored experiences to drive deeper engagement. This allows faster iterations for nascent product lines which is critical for finding product market fit (traditionally the key advantage startups have over large incumbents). Using Facebook Messenger, as an example again, FB only focused on driving installs 3 years after they had initially released the standalone app. I’d hypothesize that once Dropbox wants to drive installations of their app constellation they will have no problems – “Install Carousel and get 500 free MBs of storage” or “To sync your photos to Dropbox you need to install Carousel”. In summary, I’d suggest the best way to assess the value of this strategy is not to look at installs as the only measure of success.

Casey Winters (Growth Marketing Manager at Pinterest)
App unbundling or constellations are a nuanced strategy that I think needs a few, rare conditions to be effective. One is that your main app needs to be a top app already, with little room to improve. Then, it is advantageous to create a new app to see if you can take another top spot on the App Store/Google Play, and another icon on a user’s phone (see my blog post for more details on that strategy).

When you have this scenario, it’s little risk to create a new app, but it means that a very successful company needs to launch a new app and have it be an immediate success. You don’t have the advantage of iterating and starting small and working your way up to popularity like most apps do. It’s very tough to launch new apps this way and be successful because a ton of people try the app the moment it’s launched before you have any market feedback. Typically, people try it, discard it, and you don’t get another shot. So what ends up happening is that popular apps buy newly popular apps instead.

The only way I have seen the constellation strategy work is if the new app was already a core feature of the main app and is then unbundled. Facebook Messenger is an example of this. Since the feature is already popular inside Facebook, and the new app is now where that functionality lives (and that functionality hasn’t meaningfully changed), the new app is successful. Where foursquare erred is the the check-in was declining in popularity in their app, and when they unbundled it they changed the functionality meaningfully to upset those core users.

Alex Schiff (CEO and Co-Founder of Fetchnotes)
Unbundling into “app constellations” is understandably a compelling strategy. More real estate, more targeted products, and more mind share — hooray!

The thing is, outside the tech community most people just don’t download that many apps. Statista put out a report late last year that on average, US smartphone holders have installed 26 apps. To put that in perspective, that’s just over a page of apps on an iPhone 5 screen. Not only are most people not reading about or searching for apps, but when they do hear about one from a company they know, the default behavior is not, “Wowee! I already have Facebook – I should start using their new app!”

Putting the apathy of mainstream consumers aside, there’s a much deeper problem with the whole unbundling strategy. It only works if it’s a fundamentally distinct behavior being segmented into a stand-alone application. I think Facebook Messenger is a great example of unbundling that worked — messaging is very different from browsing stories and stalking people. Separating the two made both my messaging and social voyeurism (let’s be real, that’s what Facebook is for) experience better. Moreover, Facebook wanted to be your go-to messaging utility. That couldn’t happen, I believe, unless it was its own application.

Today, over 200M people use Messenger.

Now compare that with the launch of Paper. Paper, for all intents and purposes, has the same core features as Facebook proper – you browse stories, accept friend requests, view notifications, etc. More recently, they even added back in some of the features they left out, like birthdays and events. The major difference is UI (it’s beautiful) and a philosophical focus on stories over people.

In other words, it’s just a different approach to Facebook. Most users I’ve talked to use Paper instead of Facebook, not alongside it. Since Paper offers pretty much the same functionality as Facebook proper, most people just aren’t that motivated to try it out. As of June 11, Paper has only 119,000 MAUs. Frankly, I wouldn’t be surprised if Paper was just a test for a new approach they’ll be bringing to the core Facebook app. Unless Facebook strips Paper down to be stories and stories alone, I don’t see it surviving long-term as a stand-alone app.

There are certainly more granular product reasons Slingshot, Paper, Carousel, etc. haven’t taken off. However, across any app constellation effort, the products need to compliment — not cannibalize — each other.

Messenger does exactly that. Paper doesn’t.

Adam Sigel (Product Manager at Aereo)
App constellations, unbundling, whatever you want to call it, will ultimately yield a better experience for mobile users and better business practice for the companies making apps, but we’re going to have to work through some pain in the short-term. The trouble for now is that app constellations are ahead of the rest of the mobile experience.

It starts with app discovery. The App Store (for iOS especially) is largely leaderboard driven, and it’s hard to crack into the top ranks, especially for non-gaming apps. One “north star” app in a constellation makes it much easier to build satellites around, and we’re seeing that with Facebook, Dropbox, Google, LinkedIn, and Amazon. As Fred Wilson wrote, this creates a “rich get richer” scenario and creates enormous challenges for newcomers.

Having different apps optimized for different use cases is great, but managing all those apps is a pain. As I mentioned in a blog post about invisible apps, the mobile OSes need a way to have apps that exist outside the homescreen. “Winning the homescreen” just doesn’t make sense for lots of apps (finance and travel, to name a few). iOS is the worst offender here compared to Android and Windows Phone, but this could be addressed a number of ways including design changes, gesture controls, or anticipatory computing.

Even though more apps are building in deep linking capabilities—a very good thing for the mobile experience overall—app switching still stinks. It’s visually jarring to a lot of users, and app management is still a power user skill. To go with the typical early majority example, my parents don’t double-tap the home button to switch apps, nor do they put very much thought into which apps go on which screen or in which folders.

These are temporary imbalances, and recent announcements from Google and Apple suggest directional improvement. As app discovery improves, mobile OSes continue to evolve, and the market matures, we’ll get new, more sophisticated and seamless mobile experiences.

There’s even more discussion in the comments, on Quibb.

Written by Andrew Chen

August 4th, 2014 at 10:00 am

Posted in Uncategorized

There’s only a few ways to scale user growth, and here’s the list

without comments


Scaling growth is hard – there’s only a few ways to do it
When you study the most successful mobile/web products, you start to see a pattern on how they grow. Turns out, there’s not too many ways to reach 100s of millions of users or revenue. Instead, products mostly have one or two major growth channels, which they optimize into perfection. These methods are commonplace and predictable.

Here are the major channels that successful products use to drive traction – think of them as the moonshots.

  1. Paid acquisition. If your users give you money, then you can buy users directly through ads. Usually companies try to maintain a 3:1 CLV:CAC ratio to keep their margins reasonable after other costs. (eBay, Match, Fab, etc.)
  2. Virality. If your users love your product, then you can get major “word of mouth” virality driven by a high Net Promoter Score. If you can get your product to spread as a result of users engaging with the product, you can further optimize the viral loops using A/B tests to generate even more virality. People often measure “viral factor” to see how effectively existing users attract new users, and of course, you want your viral factor to exceed 1.0. (Facebook, Instagram, Twitter)
  3. SEO. If your product creates a ton of unique content, in the form of Q&A, articles, long-form reviews, etc., you might end up with millions of unique pages that can in turn attract hundreds of millions of new users who are searching for content via search engines. (Yelp, Rap Genius, Stack Overflow, etc.)
  4. Sales. For startups targeting SMBs or the enterprise, you’ll end up fielding a large sales org to handle both inbound and outbound. This is especially true for companies targeting local SMBs, where telesales becomes the only option. Of course, to make this work, you’ll need to generate a multiple in revenue of what you pay them.
  5. Other. There’s the odd partnership, like Yahoo/Google, that can help make or break a startup – but these are rare and situational. But sometimes it happens!

These channels work and scale, because of two reasons:

  • They’re feedback loops. Each of these channels creates exponential growth because when you make money from customers, you can use that money to buy more customers, which give you more money. Or in the virality scenario, a cohort of new users will invite even more users, who then invite even more on top of that.
  • They have a high ceiling on saturation. Part of why paid acquisition will always be around is because people like free products, which cause these products to monetize using ads. As long as people will love free products (which they will, forever), there will be advertising to buy. The biggest ad networks reach a billion users or more. Similarly, SEO works because almost everyone uses Google, so as long as you’re dealing with a high-volume base of searches (like music lyrics, or products) then you’ll be able to reach hundreds of millions of users.

It might seem like it’s best to crack one of these channels right away, and then ride then into glory. But that almost never happens, and instead startups have to work towards them – but it takes time. To figure out if your CLV and CAC match up, you need to buy some users, then wait 6 months to see how well they monetize. If you want to see if your product is viral, you need to build your app, then wait to see if you have the retention and frequency to support a strong viral loop. SEO is hard because after the content is built, Google has to index it and you have to build PageRank. This can take months and years.

New products often only have months, or a year, to live, so these strategies are often not a real option.

High-risk, high-reward
Attacking one of these scalable channels is high risk but also high reward. Every startup has to make sure they are able to slot themselves into one of channels in order to scale their business, but in the meantime, how do you show enough traction to not run out of money?

This essay by Paul Graham gives us a clue, as he writes about Startups = Growth:

A good growth rate during YC is 5-7% a week. If you can hit 10% a week you’re doing exceptionally well. If you can only manage 1%, it’s a sign you haven’t yet figured out what you’re doing.

Another way to say this is, growth is measured through a percentage and so early on, small things can drive a high % growth when the base is small. When you’re starting, there’s a whole list of other tools you can use which don’t scale at all but are nevertheless low risk.

Here are some low-risk, unscalable ways to get users:

  1. Getting your friends+family to use the product
  2. Emailing/posting among your local community, whether that’s college or an alumni mailing list or whatever
  3. Guest writing on niche blogs – you often see this with mommy blogs, etc.
  4. Cold e-mailing potential users and influencers
  5. Engaging with potential users over Twitter, Reddit, forums, and other communities
  6. Contests and giveaways, partnering with a blogger/YouTuber or something
  7. Getting covered in niche press outlets, like the tech press
  8. … etc., etc.

All of the above require hustle, but are low-risk and fairly high-percentage. And when a contest can generate a few thousand signups, on a small base that’s not bad at all. The other added benefit is that these methods put you in direct/close contact with your users. So in the early phase, when you are still working on product/market fit, this can be an important way to learn if you have the right product.

However, none of these methods scale well, which is OK, if you know when you need to move on. Even getting covered in the mainstream press, like NYT level, maybe only garners a few hundred thousand signups max. Getting featured by Google or Apple is about the same thing. That’s better than nothing, of course, but it’s still far below what you need to get on a rocketship trajectory. For the rocketship, you’d need to perfect one of the 4 main channels I listed earlier.

So ultimately, how do you balance these? Let’s talk about the barbell strategy.

The barbell
To answer the question of how to balance these growth projects, let’s talk about the barbell strategy. The barbell strategy is a way that investors can split their holdings between some high-risk/high-return investments as well as low-risk/low-return conservative investments. Investopedia describes it:

Put your eggs in two baskets. One basket holds extremely safe investments, while the other holds nothing but leverage and speculation.

In the context of these growth channels, the key is to balance a series of progressively more scalable growth projects, while keeping track of the big growth channels that will help you shoot the moon.

Do the methods that don’t scale
During the early days, by all means, sign up friends and family. And get those blog mentions, and do all the content marketing you can handle. That’ll help create a base of engaged users, while you hit product/market fit. At each point, as what works caps out, go after the next marketing channel that can drive incrementally more users. In the early days, perhaps a contest partnership with a niche blog would do, but after a while, maybe you’d hire a small team to author long-term content marketing pieces to circulate.

Invest in moonshots
The other end of the barbell, the high-risk/high-reward projects, should be taken with deliberate projects and analysis. If you need your userbase to generate a lot more unique content for SEO, start fiddling around with features that reward long-form content. And start tracking what % of users write great content. And start making the small changes needed for Google to index your site. After a few months of this, you can start to understand what it would take to create enough pieces of unique content to make an SEO strategy work. You can usually work this kind of thing out on a spreadsheet.

Balancing between the two
It’s important to balance these short-term and long-term efforts. If all you do is work on nonscalable marketing methods, then inevitably the channels will tap out and your growth will slow. When you see the startups that are highly dependent on press hits for their traction, but seem anemic otherwise, this is exactly what’s happening.

The barbell strategy helps products make progress on long-term goals while still creating short-term momentum – you’ll need momentum to attract investor interest, but you’ll need the long-term scalable growth channels to really build your business.

Good luck. And if you have a product that’s working well, has a nice base of traction, and now the only things that can move the needle are scalable methods, don’t hesitate to email me for advice: voodoo at gmail.


Written by Andrew Chen

July 8th, 2014 at 3:27 pm

Posted in Uncategorized

Why consumer product metrics are all terrible

without comments


The reality of consumer products
I’ve never met an entrepreneur who’s happy with their metrics.

Whether you’re talking about sign up rates, retention rates, or how often your users create content – on face value, the metrics always seem terrible. The secret is, almost everyone’s consumer product metrics are horrible, so once you start to compare them with everyone else’s terrible metrics – then at least we’re all in the same leaky boat together!

Other than the exceptional cases, consumers are impatient and disinterested in your product. Even the ones who sign up to try it out, only a small % are willing to stick around to use it more. As we discuss later, a typical product might see 90% refuse to sign up to a product. And then of the ones who do sign up, over 90% of users disengage and become inactive over time. These metrics are terrible, but they’re normal.

The purpose of this discussion isn’t to excuse mediocre engagement or retention, but rather, to have an honest discussion of what most companies are seeing in the market. This will help us plan better, give us more options for our Plan B, versus being total newbs on the issue.

This essay breaks down a few different metrics and the uphill slog we all face as consumer-focused entrepreneurs:

  • Signup rates
  • Retention and frequency
  • Social graph density

And before we start, it’s worth mentioning that every product is different. Mobile apps often have better engagement metrics, but have lower upfront conversion rates. SEO products have the lowest signup rates. But the intention of this essay is to add to the discussion for the kinds of social apps being built right now – social consumer products on web and mobile – and give everyone a baseline for discussion.

Signup rates as low as 1%
Average signup rates are surprisingly low. On homepages, it’s not so bad- sometimes 10 or 20% signup rates are possible. But they can be as low as 1%, or even lower, when you’re talking about non-homepage pages where people are coming in from SEO. In the extreme, when a user arrives on a content-filled landing page after typing in a query like “what is this growth I have on my back?” their primary interest is the content, not the product you’ve created. Similarly, there’s always pressure from Google’s robots to present as much content as possible, rather than hiding it behind a registration wall.

Thats why for SEO-driven products like Stackexchange, Yelp, and others, the conversion to a signed up user is extraordinary low on these content pages, sometimes much less than 1%. This leads to a pretty depressing metric in an era where most social products measure and report their Monthly Active Users, which consist of activity from users who have signed up. Unique users per month seems so 1998 :(

What if you want to raise the signup rates on these detail pages? Of course you can choose to raise this number by gating the content, as Quora has does, but perhaps at the expense of UX:

Screen Shot 2014-05-12 at 1.13.22 PM


But these are just the content pages. If we’re talking about the homepage, we’d expect signup rates to be much higher. The reason is that this traffic is usually based on word of mouth, which leads to sign up rates that are 10% or higher, since people are looking for your product in order to try it. Even better, you can send them to a minimal homepage that generates signup rates closer to 20% or 30%.

Over 95% of your users are inactive on any given day
Another metric that’s easily depressing is retention, where it’s common to see that the vast majority of your users, often over 90%, aren’t engaged on a daily basis. Instead, they’ve churned or are only active a few days per month.

The reasoning for this is simple. It’s become common to look at retention/frequency metrics in the form of D1 versus D7 versus D30 retention. Naturally, D1 means, “the number of users active on the day after signing up.” And usually retention curves look something like the below, where there’s fall-off pretty quickly with eventually stabilization around a mediocre number – often a single digit percentage:

Screen Shot 2014-05-12 at 3.11.52 PM

Usually there’s a very steep drop-off over the first week or two, and then it starts to stabilize. But you lose a ton of active users in the meantime, which is a result from multiple factors. This curve combines a few different aspects of your product:

  • First, how many users sign up and actually try out your product (onboarding)
  • And also, within a month, how many days are they active? (frequency)
  • Finally, how useful is your product over time (long-term retention)

Given that frequency is often low – 3 or 4 active days per month isn’t uncommon – when you pair that with crappy onboarding or retention, then very quickly you’ll see that getting 10% of your users to come back every day is an amazing feat. Anything more than 10% of your total users coming back every day is a success case! More often it’s 5%, or even lower.

So what if your metrics aren’t at this level? Sadly, this isn’t something that’s easily fixable with something superficial, like more email or push notifications. As I’ve noted before, a lot of these engagement metrics are more nature than nuture, and getting high usage every day has as much to do with the product category you’re building for as anything else. I’ve yet to see a product with horrible DAU/MAU get fixed using cosmetic changes. If your engagement or frequency sucks, figure out how to tie it into someone’s pre-existing behaviors, rather than asking them to do something new.

Changing engagement metrics might be the hardest thing to do with products, though. You can make your onboarding better, or get people to invite incrementally more friends. But getting them to come back over time, that’s not something that’s easy to solve using optimization techniques.

50% of your users are forever alone
So let’s say you build a new social product, whether it’s a new form of microblogging  or a new messaging app. Of course, the ideal is to have a nice feed full of personalized content. But it turns out, most products are very, very far away from that. How many people have, effectively, zero friends? You’d be surprised to know that often 50% or more of your users don’t know anyone else in the service, meaning that you need to backfill their feed with a bunch of content just from one person, or worse yet, impersonal content.

In fact, one of the most explosively viral products in recently history had a full 65% of their users disconnected from anyone else: Instagram.

Here’s a pie-chart by RJMetrics of how many Instagram users followed, while the product was in their first year:



And later in the article, this is what they say about it:

Interestingly, over half of Instagram’s users are following exactly one other user, with another 13% not following anyone. We checked into that, and it looks like the vast majority of users who follow only one other user are following the “Instagram Team” account, which was likely automatically added to their list at signup.

This means that 65% of users effectively follow no one.

(Emphasis added.) This is amazing, and ultimately didn’t stop them from becoming a very functional social network alternative to Facebook itself.

When you combine the fact that getting social graphs to fill is very hard, and the fact that only a few percentages of users will author content (known by the 1% rule), then you can imagine why creating a healthy, dynamic news feed is so hard.

Ultimately, density can be solved by more growth. More users mean a denser social graph. But also, a key component of getting people to follow more people is to connect their Facebook accounts, their email addressbooks, and other pre-existing graphs which help them bootstrap their relationships. Or do what Twitter does, in forcing users to follow before creating an account.

Mediocre metrics aren’t an excuse
The point of this essay isn’t to provide an excuse for mediocre metrics, but rather, to point out the harsh reality of the situation. There’s just a stark contrast between how much we as consumer entrepreneurs care about our products, versus our target audience who really doesn’t give a shit about how much effort we put in. As a result, people aren’t signing up, and if they do, they don’t use the product nor have a good experience. It’s very hard.

But even as the average product’s metrics suck, as an industry we’re looking for the unicorns. So just as I say that a 30% DAU/MAU is good, when you compare that to Whatsapp’s 70%, you can see the gulf between good versus great. We all want great, because the tech industry is all about building great products.

On the plus side, even though these percentages all seem small, great businesses have been built from a few percentage points here or there. How many paid subscription services monetize by convincing 2-3% of users to pay? Tons of them. Or who build billion dollar ad-supported businesses just based on getting a few % of users to click on ads? That’s everyone in the ads business. So it can work, but until it’s scaled and is growing faster, these metrics can look like a mess. Until then, keep at it.

Written by Andrew Chen

May 13th, 2014 at 10:26 am

Posted in Uncategorized

How to solve the cold-start problem for social products

without comments


Social products need mass before scaling growth
I often write on the topic of how social products can scale growth, resulting in inbound emails to the effect of “how do I get my product to go viral?” The problem is, until you have a strong baseline of engagement, it’s nearly impossible to have a metrics-oriented discussion on growth and virality. So you have to get that first, before you can talk about the next step.

The focus should be on creating that baseline – a small-to-medium sized network of highly engaged users in a big market, that’s growing. Maybe this is 10,000+ active users organically gaining hundreds per day, at a 20%+ DAU/MAU. If you can hit that, then it’s much easier to talk about how to scale it up. I’ve written Zero-to-Product/Market Fit in the past to talk about some of the steps you might take to reach this stage. Similarly, I have some slides for this topic. (And if you’re at this point, don’t hesitate to email me)

There’s a unique aspect to social products in getting to this baseline, which is how can you solve the dreaded cold-start problem? If your product is inherently social, but you don’t have a critical mass of users, then it’ll naturally fail. How do you get beyond that? This is different than productivity or SaaS products because you don’t just have to get the product right- you have to get your initial user network to be large enough and active enough too.

Here’s a few ways I’ve collected over time on how to approach that problem:

Single user utility
This is one of the most common ways to approach the cold-start problem. Give people a value-proposition that gets them creating/curating content within your network, and as a by product, it’ll help bootstrap the network around the user. I think of Pinterest as the quintessential example here, where you can use it as a tool to collect/gather/organize content around a particular project you’re pursuing- decorating a new apartment, planning a wedding, or switching to a new diet. As you’re doing this, then you use the common mechanisms around finding friends, Facebook sign in, etc., to build a network around the user. If you can get this to grow fast enough, and build the right social feedback loops, then users will find themselves blending a single user value prop with a network value prop over time.

Linkedin is another classic example here, where initially they could market themselves as a way to put your resume online. But of course, once you go through their onboarding flows, you’ll quickly find out that people are connecting and reaching out to you via your profile, thus cementing the network value proposition.

A blog network like Tumblr is another great example. People like making their own websites, and you can use Tumblr for that – plus you get the themes, tools, and domain for free. But once you’re set up, it becomes easy to get reblogged and followed and all of a sudden you’re part of the network product.

The trickiest part of using this strategy is that you’re asking users to switch their mindset from one value proposition into the other. Managing that transition isn’t easy. You may find that users actually want their single user value prop to be private, and nonsocial by nature. Or you might find that if you don’t get the social feedback loops right, you may not be able to convert your one-off users into network users fast enough, and it feels like you are maintaining two separate products. And it might feel like your product isn’t really working if the majority of your users aren’t involved in the network.

Publishing into an pre-existing network
A variation of the single user utility is one where the primary functionality of your product is to share into a pre-existing network. The classic example of this is Instagram, which provided the initial value prop of photo filters and sharing to Facebook, which can be used even if none of your friends are using the service. However, it spread virally over time, which brought more people to Instagram, and this was then used to bootstrap a separate network based on following celebrities rather than the bidirectional Facebook friend model.

(I published a guest post Social Products with with utility, not invites, as a longer exploration of this idea)

The main challenge with this model is twofold: One is the “two value proposition” problem as stated before: Initially, a large % of your users might view Instagram as “that app I use to post to Facebook” rather than a destination in itself. The second challenge is that you can have a platform dependency that may not end well if your “host platform” decides to cut you off.

Small network requirements
Not every product has to have a single user value proposition, and in fact, it can complicate things to feel like you have to design for multiple use cases at the same time. Instead, a different approach would be to focus on building a product that has a small critical mass requirement. In fact, you could look at the following categories and assess their critical mass requirements:

  • Skype: 2+ people
  • Group mailing list: 5+ people
  • Social network: 10+? 50+?
  • Social+mobile+location based: Lots :)

So one strategy is, how can your product be useful for just a small handful of people? That way, if you have a big launch, you can get lots of active pairs of users, like families or couples, and you can hold on to your audience. But if your product requires a very large critical mass of users, then maybe it will be very hard to get there.

Local network saturation
For products that require a large critical mass to get started, I’m already skeptical. But if you must, going after some kind of hyper-connected vertical is a good way to start. Rather than getting 1,000 users randomly and who don’t know each other, instead you focus on getting 1,000 users who are densely connected already. That way, you can saturate the network and hold onto that group of 1,000, and then go from there.

Although the following companies also often had attributes such as single user value prop and small network requirements, it’s useful to think about starting within a niche: Yammer did this within a company. Yelp did this within San Francisco. Facebook within Harvard, and Twitter within the tech community at SXSW. Snapchat within SoCal high schools.

Perhaps there’s a niche hyper-connected pre-existing network that matches with your product, and if you can retain those users, then you can build a much larger network from there.

Why big unfocused launches often fail
The above provides a clue on why big social product launches on Techcrunch or DEMO or whatever often fail. The problem comes down to the fact that for social products, you often need hit some metric of connection density to succeed. A “minimum network density” metric, if you will. And when you think about it like that, you’d much rather have 100,000 users with a density of 30 connections/person, than 1,000,000 who have a density of 2 connections. Because ultimately, those million users will churn out because they won’t have the content and feedback loops necessary to stay engaged.

Big launches fail because they might pump up the total users number, but don’t help much with the network density number. If anything, they might lower the average. So if you want to go with the big launch, make sure that it either targets a network that’s hyper-connected, and who will onboard nicely into your product. Or make sure there’s a very strong single user value prop, and even if they can’t find anyone they know in the product, that’s OK.

OK, good luck my friends!

Written by Andrew Chen

March 27th, 2014 at 11:34 am

Posted in Uncategorized

How to design successful social products with 3 habit-forming feedback loops

without comments


Social products share a common ancestry and set of problems
It’s been a decade after Friendster popularized the notion of the social network, and we’ve seen hundreds of flavors of social products. Many of them are very different from each other, showing that success can come from many variations. I’ve come to believe there’s 3 main feedback loops that drive the success of these social product designs – here’s the trifecta:

  1. A feedback loop that rewards content posters when they push new content into the network
  2. A feedback loop that rewards passive content consumers with relevant and valuable content
  3. A feedback loop that rewards (and culls) connections within the network

It’s great when all three feedback loops act in harmony. As users act within each feedback loop, everyone’s happy, and the players in the ecosystem produce and consume valuable content for the network. When this happens on a daily or hourly basis, it creates habitual usage within your product- driving engagement and retention.

On the other hand, when even one feedback loop starts to fail, reverse Metcalfe’s Law goes into effect, leading to stagnation and ultimately, network collapse.

As an industry we’ve often talked about the distribution of content creators, curators, and consumers – it’s often known as the 1/9/90 principle. But that’s about the distribution of these different kinds of users, and not about fundamental motivations behind their actions. The feedback loops for social product aims to think in terms of why these feedback loops are able to create happy emotions and build up habits. Furthermore, by looking at each loop in isolation, it becomes more obvious where one could innovate- by adding a twist in content creation, consumption, or how people are networked. I’d argue that anonymity, constrained media types, algorithmic news, and other innovations all fit into these feedback loops in different ways.

Content posters that crave feedback (or utility)
First and foremost, let’s talk about the folks who post content – these are the 1% and 9% part of the 1/9/90. These users might post content by creating it in a textarea or uploading a photo, or it might be more curation oriented- simply retweeting a funny link or sharing a link. Either way, they take an action that writes new info into your network that impacts the content consumption experience. The feedback loop that’s important here is to reward content posters with social feedback. You publish content to your audience and then social feedback trickles in over time, drawing you back to the product. If content creation is easy enough, and the social feedback is compelling enough, then you do more. And so the loop continues.

It turns out that what type of content people post is important: Social products ultimately have some kind of content in the middle of it (sometimes called the social object), that determines the posting/consumption behavior of the content. This might be a tweet, a photo, a musical playlist, a restaurant review, or even a commerce page. It would be a mistake to assume that it’s as simple as wanting this content to be as simple as possible to create, because you also need to make it a frequent behavior. You also need the resulting content to be compelling as well – it’s these constrains that make this system tricky.

First let’s talk about what it means to make the posting “easy” – it’s not just that the tools are simple, but also:

  • You’re already creating it, so it’s not a new behavior (for example, almost everyone sends links, photos, etc.)
  • You can create it in seconds (sometimes via an artificial constraint)
  • You do it all the time, and over a long period of time
  • You don’t feel self-conscious publishing it
  • You can use new technology that lowers the bar (location sensors, camera, etc.)

A lot of the recent innovations in social products have focused on making this easier. One important tool is the use of constrained media types, where a tweet of 140 characters ensures a level playing ground for content so everyone can write a tweet in a few seconds. The 6-second Snapchat lowers the mental effort in taking the perfect photo. Foursquare uses our smartphones to make it easy to publish our location, whereas years ago, the effort on a feature phone would have been much higher. Similarly, the new trend of anonymity is another way to lower our inhibitions towards content creation. (I’m excited about the trend towards wearable and ubiquitous computing because they’ll be tools for all sorts of easy content creation.)

The tricky part of content creation is that the output has to be compelling to consumers, and over a long period of time. If your content is novelty (for example an avatar creator), then it may thrive for a period of time but ultimately the loop will weaken and stop. That’s fine for an ad campaign but not a product.

On the other hand, sometimes content can be very high cost but still be really compelling, for example long-form writing or high-production video production. You end up with a small % of creators who can actually author the content, but the end result is compelling enough that the whole thing keeps going. Yelp reviews, Stackoverflow, and others operate like this, with a push from SEO which help both creators and consumers find the site again over time.

Ultimately, the balancing act between content creation cost, the frequency/retention of it, and how compelling the output is – well that’s the magic of a new product design.

The health of the feedback loop around content consumption versus social feedback is based on a number of key variables, all of which are interrelated with each other:

  • What % of users create content
  • How much content is created (ease, frequency, retention)
  • Who this content is shown to
  • How compelling the content is
  • What % of consumers give feedback to the content creators
  • How compelling that feedback is
  • Whether the feedback brings back content creators to make more

The tricky part to the above is that many key variables oppose one another. You can increase how often content is shown to people just by blasting out content indiscriminately, but that decreases the relevance of the content. You can make it really easy to give user feedback, but at the cost of making the feedback less compelling. All of these tradeoffs ultimately manifest themselves in the design of a social product, hopefully in the right dosage and combination.

One footnote is that content posters can also be compelled by providing a single user utility, which produces compelling content as a byproduct. The classic example of this is bookmarking- Pinterest and Delicious help you organize content as your single user utility, but once the content is in the network, other folks can interact with it. This ultimately bootstraps the network as positive social feedback flows in, ultimately replacing the “organize stuff” value proposition with a “people tell you how much they love your stuff” benefit.

Content consumers want relevant content, updated frequently
Now lets think about the viewing experience. When it comes to content consumption, I think about the things that people want to look at every day. There’s not too many of them. News about their friends/family, news about the world. News about work. That’s one big chunk. Entertainment, which these days might look like YouTube videos, but even easy-to-create memes. For some demographics, maybe they want to see commerce content – shopping is always fun. And if you have hobbies, maybe you want to see a bunch of vertical content about that kind of thing – whether it’s about the arts, cooking, or programming.

The feedback loop for content consumers is simple: Every time they open your app or website, they see compelling content. That builds a habit for them to check in every morning, every time they’re standing in line, and every time they’re bored at work.

Yet the loop is easily broken – here’s the usual failure states:

  • Feeds that lack content
  • Feeds with stale content
  • Feeds with too much content
  • Feeds with irrelevant content

Lack of content and stale content comes from using a friending/following method of connecting content posters and consumers – but often, the network is underdeveloped or isn’t growing fast enough. Or maybe there’s not enough friend density to drive a full feed. Or even if there is a lot of users using the product, there isn’t the “right” users – for instance, an adult user stumbling into a website mostly filled by teens. These are some of the common reasons why it can be difficult to evaluate new social products – even if the mechanics and loops are well setup, if you don’t have the right users it’s hard to see the magic.

But once there’s a nice balance of new content coming into a feed at about the rate that content consumers want to see it, something great happens. Then the engagement can lead to people giving social feedback to the folks who posted it in the first place – via likes, comments, re-shares – and that stimulates the production of more content.

Connecting content posters and consumers to drive relevance
The way that content consumers participate in the feedback loop is that they give feedback to content creators. But before they do that, they need to have a method of picking what content is relevant to them on their home screens:

  • Picking people (Facebook, Twitter)
  • Picking topics (Quora, Stackexchange)
  • Leaderboards (Reddit, Hacker News)
  • Editorial curation (Medium)
  • Algorithmic curation (Flipboard, Prismatic)
  • Location (Foursquare, Highlight)
  • Anonymously matched (Secret)
  • … and more to be invented!

All of the above work, with different tradeoffs. Allowing people to customize their content consumption based on people and topics is the most scalable, but the hardest to get started. It’s a classic cold-start problem. To get to that, you need a critical mass of content creators who are making the kind of content that might attract a passive audience. Given that content creators are also consumers, that’s why oftentimes it’s the easiest to get started with a group of content creators.

The feedback loop about generating meaningful connections needs to reward the network when authentic connections are made. When you pick a new topic, or a new person, does that expose you to new content that then gives you new opportunities for people to follow? Do you have plenty of opportunities to unfollow or otherwise clean out your feed of irrelevant information? And are new people joining the product all the time, driving notifications, re-engagement, and ultimately new content into the network?

Editorial curation and leaderboards (like Hacker News) are easier to start, but have the drawback that they don’t scale well. Editorial requires you hire lots of people. Leaderboards create a single public space where it’s difficult to create a “one size fits all” experience that makes everyone happy.

It’s also difficult to mix the two. If you combine user generated content with editorial, within the same feed, then inevitably editorial content will “steal” the feedback from the UGC. That’ll weaken the loop. Instead, to really make sure that enough social feedback is being given, the goal is to make a feed with compelling assortment of content, and a lot of easy ways for consumers to interact with the content creators.

Another interesting issue on social feedback is the issue of quality. If you upload a video to YouTube, and then get 1000s of incomprehensible comments from teenagers, is that better than a smaller number of comments from thoughtful people? I suppose it depends on your own tastes, but over time, I’ve personally come to value the feedback of a small group of people I respect rather than trying to maximize the levels of pageviews or comments that I get.

This can be a difficult challenge because startups obviously face the pressure to grow, and one of the easiest ways to do that is to get your users to invite and add lots of meaningless connections. At the same time, if they follow too many users, or topics, then their feed will get busy and the product will lose relevance. So ideally, you have a system in place where users can add (and remove) connections to other users easily, and the system is able to suggest more relevant connections. This should also provide a better and more personalized experience around content consumption.

Building a checklist to ensure your loops are healthy
I leave you with a checklist for those of you who are designing social products, but find that your feedback loops aren’t quite working. The question I’d ask is, zoom into each of the feedback loops, starting with the folks who are posting content. Ask yourself, are they getting feedback on every action they take? Is it high quality feedback that makes them feel good? Are they making enough content to be interesting? If not, the feedback loop is broken and needs to be fixed.

For content consumers, are people getting high value, meaningful feeds? Or is it a random mishmash of popular content in your product? And if the feedback loops aren’t working, consider creating a small network where it all works, and grow that out, rather than forcing bad feeds on everyone that visits.

Or alternatively, consider taking a small part of another products’ feedback loops, and tweaking it a little. There’s many innovative products yet to be invented.

In a decade of social product design, we’ve seen many significant innovations around many components of these feedback loops. Facebook innovated with real names and a privacy model which helped drive closer-knit social feedback. They also invented the feed, a new way for posters and consumers to more efficiently transact on content. Twitter pioneered the follow model, which is yet another way to connect people. Instagram took advantage of much easier content creation methods on your smartphone, combined with plugging into existing networks, to bring something new to the model. And recently, anonymity apps like Secret are connecting people in yet another new way.

When I first arrived in Silicon Valley back in 2007, I remember a very smart B2B investor asked me, “Does the world need another social app?” implying that the category had been fully exploited. I think we see that in fact, given the years of solid innovation since then, there’s many new social products yet to come.

Written by Andrew Chen

March 17th, 2014 at 1:45 pm

Posted in Uncategorized

Congrats to my sis Ada Chen, who’s joining SurveyMonkey as VP Marketing

without comments

I’m happy to congratulate my kid sister Ada Chen (@adachen) who joined SurveyMonkey as their VP Marketing this week. Awesome.

Ada is exceedingly modest for a successful entrepreneur. She has two successful exits and a deep background in marketing (yes, it runs in the family!). After attending UPenn and joining Microsoft, she soon moved from Seattle to San Francisco to join her fiance Sachin Rekhi (now husband). In the process, Ada met with a ton of different startups, ultimately joining as one of the first dozen employees at Mochi Media, an Accel-backed games+ads startup. Soon after, it was grown to over 100M uu/month and acquired for $80M by Shanda Games, a public games co in China.

Soon after, she started Connected with her husband Sachin, backed by Trinity and 500 startups, which aimed to manage all your professional relationships in one place. The technology behind the product was amazing, and was quickly picked by Linkedin to form their new Contacts product – the announcement here. And now, she’s moved on to one of the so-called “unicorn” billion dollar companies in Silicon Valley at SurveyMonkey.

It’s great to see friends do well, and even better to see my sister do well. I’m happy to share this good news.

And finally, a photo of us in our bowl cut days:


Written by Andrew Chen

February 26th, 2014 at 12:18 pm

Posted in Uncategorized

How to make content creation easy: Short-form, ephemeral, mobile, and now, anonymous

without comments

Image 2014-02-10 at 3.58.44 PM

Like many folks over the last couple days, I downloaded Secret and started playing with it over the weekend. The NYT even wrote an article about this category of apps. A few thoughts went through my head- is this a gimmick? Are the push notifications too much? Will I have secret fatigue? Yet after a couple days of using the app, I’ve come to believe that anonymity is a powerful new innovation that dramatically lowers the cost of creating content.

This is important because content creation is the heart of any social app. If it’s easy to create compelling content, then that content is quickly shared to other networks, driving viral growth back to its source. A fresh stream of compelling content brings the bulk of any social product’s primary audience – a large group of passive consumers who just want to flip through all the cool photos, videos, tweets, and more, maybe commenting or liking a few they really feel strongly about.

However, content creation is hard. That’s why, famously, only 1% of users will do it. In the last few years, social products have started to innovate around making it easier to create and curate content. I wrote about this previously before in the essay, Constrained media: How disappearing photos, 6 second videos, and 140 characters are conquering the world.

But to summarize, here are some of the ways content creation is being made easier:

Short-form content forces everyone on the platform to keep things casual. For example, Vine’s 6 seconds or Twitter’s 140 characters. This reduces competition and the feeling that you have to do something really long, well thought out, and intense.

Ephemeral content makes everything throwaway, like Snapchat’s photos. If you can’t save it, and look at it later, then it’s no big deal if you send an ugly/uncomposed photo, especially if it’s an ugly duck face selfie.

Mobile content takes advantage of the fact we’re now carrying a bunch of sensors with us at all times. Foursquare makes it easy to share a location by using your phone’s GPS – this is something that’s unique and wouldn’t have been easy 10 years ago. Photo-sharing social apps have exploded now that we all carry cameras 24/7. Wearable computing will lower the barriers even more, and enable ambient/ubiquitous content creation – Runkeeper maps are just the beginning.

Curated content can be great because you don’t have to create the content yourself- you just have to share the link. So whether it’s sharing shopping links on Pinterest or memes on Reddit, all you need to know is copy and paste.

… and now, it’s clear that Anonymous content may be it’s own thing:

Anonymous content makes it easier to share what you really think, without worrying about what other people will think of you. For the last few years, the trend has been towards real identities – but with that paper trail, it’s easy to self-censor. Anonymity removes the desire to self-censor, at least for certain kinds of content.

I think we’ll find that anonymity, just like the other above options, is just a design choice that has its own limitations and tradeoffs. The power of real identity is that your content becomes an extension of yourself. So as long as the content is positive – an accomplishment, or a fancy vacation – then you’re likely to want to share it publicly, attached to your friends, with your real name. But there’s also a world of content we’re all censoring from other people – our insecurities, authentic (sometimes negative) opinions, and so on. – and anonymity is powerful there.

It’s impressive to see entrepreneurs continue to reinvent social paradigms even 10 years after Facebook got started. I remember a few years ago, a smart VC asked me, “why does the world need another social network?” when in fact, it looks like we’ll see many new successful social products over the next few years. Excited to see what else people come up with.

Written by Andrew Chen

February 10th, 2014 at 6:17 pm

Posted in Uncategorized

My 2013 essays on mobile, startups, and tech

without comments

Happy new year! Here’s a quick list of some of the essays I wrote (and guest posted) for 2013.

Long form stuff
Zero to Product/Market Fit (Presentation)
Here’s a presentation I gave to a group of entrepreneurs talking about the issues starting out and getting to product/market fit. Given that most startups fail before hitting that, I think it’s the right thing for most folks to focus on, versus growth or company culture or whatever.

Rational Growth (PDF): An intro to growing user signups via data and analytical thinking
An ebook on how to think systematically about growing user signups. Covers the idea of taking a user flow, modeling it out on a spreadsheet, then tweaking numbers to guess effects, and then A/B testing to validate.

Essays that I wrote in 2013 about startups and tech

Mobile traction is getting harder, not easier. Here’s why.
The mobile ecosystem is evolving, and a lot of the marketing tactics that worked a few years ago don’t work anymore. Launches, getting featured, are all helpful but aren’t as powerful as before.

When a great product hits the funding crunch
An analysis of Everpix and how they failed to raise their next raise, even given their high-quality product and effective freemium numbers.

Constrained media: How disappearing photos, 6 second videos, and 140 characters are conquering the world
A key bottleneck in social products is the participation rate of content creators – adding constraints like 140 characters and 6 second videos helps make content creation more casual, and thus more people do it.

Confessions of a Startup Seagull
It’s easy to criticize new startups.

Why it’s hard to evaluate new social products
Social products create an experience using both the “bits” combined with the other people who are using the service. It’s hard to evaluate a new social product if you don’t have a critical mass of friends using it too.

Books I’m Reading
I read some random books.

Ignore PR and buzz, use Google Trends to assess traction instead
It’s easy to confuse PR buzz and traction. I use Google Trends to figure out how many people are searching for a brand, which IMHO has been one the best ways to assess if something’s really working.

I’m a Google Glass skeptic and think it’ll be the next Apple Newton
Wrote this before I even used a Google Glass, and now that I have, I definitely haven’t changed my mind. It has to be a lot better than the less geeky alternatives (smartphones, smart watches), in order to get people to wear it 24/7.

Minimize your Time to Product/Market Fit
Given that most startups fail before hitting P/M fit, here’s a couple thoughts on how to increase the odds of getting there.

Why are we so bad at predicting startup success?
Humans suck at probability, and VCs/entrepreneurs are no different. It’s hard to draw conclusions by looking at the 10-20 successful startups generated each year, the dataset is too small.

How this blog grows: Evergreen content, Social whales, and “Don’t get bored”
Build a feedback loop using evergreen content that cross-sells to social media channels.

Why developers are leaving the Facebook platform
The Facebook growth opportunity is 100% over. Move to mobile.

The death of RSS in a single graph
I turned off RSS subscriptions on my blog this year, and am moving to email. Here’s one of the main reasons why.

Linkedin, Facebook, Google, Twitter, eBay, YouTube, Wikipedia, Amazon, Hotmail, Blogger, Apple: How they used to look
Humble beginnings.

New college grads: Don’t sell your time for a living
I wrote this as part of a Linkedin feature for new college grads. Most people sell their time for a living, and as a result, they lack understanding on how they create value in the world plus it’s a shitty personal business model since you’re always running out of time.

Also, I had some wonderful guest essays this year too – I won’t attempt to summarize them:

Why you can’t find a technical co-founder (Elizabeth Yin at Launchbit)

The critical metrics for each stage of your SaaS business (Lars Lofgren of KISSmetrics)

9 ways a billion dollar new mobile company might be created (Bubba Murarka at DFJ)

The highest ROI way to increase signups: Make a minimal homepage (Mattan Griffel)

Use this spreadsheet for churn, MRR, and cohort analysis (Christoph Janz, Point Nine Capital)

3 common email marketing failures (Elizabeth Yin at Launchbit)

Social products win with utility, not invites (Sangeet Choudary)

How to grow your app revenue with DuPont analysis (Kenton Kivestu)

Written by Andrew Chen

January 2nd, 2014 at 1:39 pm

Posted in Uncategorized

When a great product hits the funding crunch

without comments


Building a great product is not enough
Today I read a well-done article by The Verge on the shutdown of Everpix, a photo startup that’s gained a small but loyal following. It’s a great read, and I’d encourage you to check it out. There’s a lot of things to comment on, but the Everpix story is a common one these days- a lot of startups have built great initial products, and even shown some strong engagement, but ultimately not enough traction to gain a Series A.

The essay on Everpix drove home a lot of recent trends in startups that have gained momentum for the last year or two. Let’s examine a couple of these trends.

Funding goalposts continue to move
The first thing we’ll talk about is the company metrics. One of the best things about this article was that they did a good job of covering some of Everpix’s stats on engagement, conversion rate to premium, etc.

Everpix stats

  • 55,000 total signups and 6,800 paid users
  • Freemium biz model of $4.99/month or $49/year
  • Free-to-paid conversion rate of 13%
  • 4.5 star rating with 1,000+ reviews
  • MAU/signups of 60%
  • WAU/signups of 50%
  • Raised $1.8M and then a seed extension of $500k
  • Ex-Apple founders with 6 FTEs

You can see that other than the top-line metric of total signups, the other metrics are quite solid. If this company were started just a few years ago, I’m convinced they would have had no problem raising their Series A. These days though, it’s gotten a lot harder.

The reason for that is the “moving goalposts” on what you’re expected to do with your funding.

It’s been widely noted that investing milestones have evolved quickly over time:

  • In 1998, you’d raise $5M Series A with an idea and not much else. The capital would be spent to build the product, and hopefully you’d have some customers at the end of it, but it wasn’t required. You had to do crazy stuff like put machines into a datacenter, at this point. Then you’d raise a Series B to scale the marketing. The qualitative bar for the team, idea, and market was high.
  • In 2004, you’d raise $500k with just an idea. Then you’d build the product and spend $5M to market it. At this point, you could use a free Open Source stack which would accelerate development. You didn’t need to build a datacenter either.
  • In 2013, these days, you are expected to have a product coded up and ready before you raise your first substantial angel round. Maybe the product won’t be launched, but people will want to play with a demo at least. Then you raise $1-2M to get traction on your product. Then if you have millions of signups, then you get to raise your Series A of $5-10M.

In fact, it’s been famously written by Chris Dixon, now a partner at Andreessen Horowitz, that 10 million users is the new 1 million users. I’ve previously written that Mobile Startups are Failing Like It’s 1999, due to the long launch cycles that the Apple Store encourages. I’ve also written about mobile getting harder and not easier over time.

There’s a couple things going on: The sheer proliferation of seed-funded startups, combined with investors who want to invest post-traction, post-product/market fit. Combine this with 1999-style launches for mobile apps, and you have a big mismatch in the supply and demand for funding. Series A venture capitalists are often acting like growth investors now, where they want the entire equation de-risked before they put in much capital, and it’s reasonable to expect this given the technology stack and massive distribution channels.

My question is, in 2016, will the bar be even higher? Maybe angel investors will expect a working product, reasonable traction, and product/market fit all before they put in the first $1M? How much can market-risk be proved out before any professional money is raised?

Monetization won’t save you if it’s not combined with growth
The Everpix story also shows that having a business model isn’t enough- after all, a 12% conversion rate to premium is stellar, which you can compare to Evernote’s 6%, as they mention. The problem is, if you have monetization in place, investors also want to see a lot of growth. Or you need enough growth and scale to be profitable without outside funding.

Work backwards on the latter to see what that looks like:

  • 6 FTEs plus operations costs about $100k/month
  • At $5/month, you need 20k paid subscribers to break even
  • At a 12% free-to-paid rate, you need 160k signups

Turns out, 160k users is a lot, especially if you have a short runway. It’s well outside the boundary of a list of friends and family, or a Techcrunch article, or a big week of promotion from Apple . If you combine this with the rest of your schedule, like 6 months to raise VC, another 6-12 months to build the product, etc., then you don’t have much time to hit your traction milestones.

In contrast to the option to hit profitability, VCs don’t care that much about small scale monetization. They understand that a freemium service can get 1-5% conversion rates, and the question is if you have enough top-of-funnel signups to make the revenue numbers big. In fact, too much focus on monetization too early can lead a red flag, since it’ll mean maybe the entrepreneur is thinking small rather than focusing on winning the market.

A modern startup’s costs are all people costs
The final thing that’s worth pointing out in the article is the cost structure of the company and where the money went:

  • $565k consulting and legal fees
  • $128k office space
  • $360k operating costs
  • $1.4 total personnel costs

In other words, 80% of the costs went towards the employees and contractor/consultants/legal. It’s basically all people costs. You could argue that the office space is really just a function of the people too. Really, only ~15% of the capital went towards actually running the service.

If anything, this trend will only continue. San Francisco housing costs continue the rise, while computing infrastructure only gets cheaper and more flexible.

The nice thing about these costs, of course, is that you can always scale them down by scaling down your team. It’s complicated to do this, of course, since the value in this acquihires incent you to keep a large group of people going up until the end. But if you are convinced to work on the business for the long term, you can always scale things down to a few core folks, though it can be painful.

This is another reason why increasing your cost structure can be tricky if your product isn’t working in the market already. You end up in a case where just a year or two down the road, you have to make the tough decisions to keep going, or to shut the product down. So if you are working on something that you’re really passionate about – or as they say, amazing founder/market fit – then you may want to delay the team buildout so that you don’t end up creating that situation in the first place.

“Milestone awareness” and clear product roadmaps
Ultimately, this flavor of startup shutdown will continue to happen. Products that hit immense traction are the exception, not the norm, for a reason. Given that, what can you do? Ultimately, every founder needs a strong sense of “milestone awareness.” What I mean by that is the ability to understand what you need to accomplish before the next round of funding, and then to work backwards on that until you can put together a reasonable roadmap to get there. You might have to cut costs if the plan doesn’t seem to work. And you’ll have to revisit this plan on a regular basis to understand how it fits together.

The problem with hyper product-oriented entrepreneurs is that they often have one tool in their pocket: Making a great product. That’s both admirable, and dangerous. Once the initial product is working, the team has to quickly transition into marketing and user growth, which requires a different set of skills. It has to be more about metrics rather than product design: running experiments, optimizing signup flows, arbitraging LTVs and CACs, etc. It’s best when this is built on the firm foundation of user engagement that’s already been set up. In contrast, an entrepreneur that’s too product oriented will just continue polishing features or possibly introducing “big new ideas” that ultimately screw the product up. Or keep doing the same thing unaware of the milestone cliff in front of them. Scary.

Any startups that are at the “just add water stage” should email me and I’ll connect you with the resources and people to grow.

It’s funny that people take the lesson away from Apple that you should just focus on product. That’s only half the story, I think, because when you dig into why Apple is so secretive, it’s because the company is really focused on advertising and product launches. The secrecy that’s so deeply embedded in the organization facilitates their distribution strategy- can you imagine building your company culture around your marketing strategy? That’s what Apple’s done, though it’s not often talked about.

Good luck, guys
Finally, I want to wish the Everpix team good luck- they put together something that thousands of people enjoyed. That’s very hard, and more than most people can say. And they took away some very useful lessons that will only make them better entrepreneurs.

It’s never an easy thing to shut down something you’ve worked on for years, but I was insanely happy to see such a high-quality post mortem from The Verge. Thanks for writing this up, guys!

Written by Andrew Chen

November 5th, 2013 at 2:23 pm

Posted in Uncategorized

A clever way to buy Facebook ads based on what your users like (Guest post)

without comments

My friend Gagan Biyani wrote up a great piece how to analyze what your Facebook audience is interested in, and using that to buy ads. He’s generously shared it, below. Gagan is CEO and co-founder at Sprig, and before that was at Lyft and started up Udemy. You can follow him on Twitter at @gaganbiyani and he has a new Medium account here. -Andrew

Gagan Biyani, Sprig:
Building Target Groups for Facebook Ads

Facebook advertising is tricky and there are multiple facets to it. By far the most important value of Facebook is being able to target based on demographic information. In this post, I’ll show you exactly how to use the “affinity ratio” to figure out what Facebook likes to target and dramatically increase the performance of your Facebook advertising.

As mentioned above, you have to have Facebook Connect on your app and you must grab the likes of your users. A sample of your users is OK so even if Facebook Connect is merely one option amongst many, that’s fine.

(Credit: This method was created by my co-founder at Udemy, Eren Bali. We tested dozens of other forms of targeting and nothing came close. I’m sure there are many other companies that have come up with this on their own.)

Note: this only works if you have Facebook like data from your user base

Step 1: Figure out what Facebook pages your users like

Here’s the trick: Download a CSV of all of the unique facebook pages your users like and the COUNT() of the number of users who like each page. Example:


You’ll notice these numbers don’t make a whole lot of sense. That’s because I made them all up!

Notice a few things. First, this list is sorted by greatest # of likes. That list is already a bit useful — and probably something you’ve looked at before. The problem is it doesn’t make this data useful enough.Everyone and their mom’s likes Michelle Obama, so you can’t target your advertising that way. From here, you have to figure out which one of these pages is actually useful to you.

Second, if you have any reasonable-sized user base, you’ll probably have 1000’s of results on the left column. That’s fine but we’ll use a small list for this example. You may have to have some sort of COUNT() limit to make this easier (aka only pages with over 1,000 likes make the cut).

Step 2: Add in the “Global Likes” of those Facebook Pages

Now you need to figure out how many globlal likes each of the pages on your list has. Use the Facebook API or do it manually if you don’t have access to dev resources.

In our B.S. data set above, I went ahead and did it manually. Here’s the data:


In case you’re wondering, I came up with this list by checking Facebook’s page recommendations. They were right on some things (who doesn’t like a little Dr. Oz in the mornings?) and wrong on others (phh, I don’t care about animals).It is probably starting to make sense now. Its not just about the total count of your users who like a given page, its actually about the relativecount.

Step 3: Create a ratio of [Count(users)]/[Global Likes]

From here, your goal is to create an “affinity score” (name created by Dinesh Thiru, who runs marketing at Udemy)


To make these numbers easier to read, I multipled the affinity score by 10,000. Depending on the number of users you have in Column Count(users), you may multiply by a smaller factor of 10.

Now, you have a relative score that allows you to compare different pages. I sorted this list by affinity score. Its interesting to see pages like “Michelle Obama” and “Food Network” to go from the top to the near-bottom of our list! Of course this is make-believe data, but when you have real data you’ll see similar results.

Step 4: Group your high affinity pages

Once you have a list of affinity scores, you need to group them into categories. This is important because otherwise, you wouldn’t have good Facebook targeting groups. Targeting users who like BothSidesoftheTable and the Golden State Warriors will make it hard to write ad copy and create cohesive campaigns.

1-AbAqc1TX5RrSKeZco-5WsgNatural groups will form when you start looking at your data.Two things matter when you are in the final stages of this:

  1. Sample size. The larger the size of your group, the more people you can target with your ads. You don’t want too large a size, though, because then you are paying crazy CPM’s and competing with a larger breadth of advertisers.
  2. Grouping. This is based entirely on your judgement. The natural groups are always ones where you think there’s a lot of overlap amongst those users. Its fairly obvious that people who like TechCrunch and BothSidesoftheTable overlap. In situations like the Monterey Bay Aquarium and In Defense of Animals group, its just a judgement call. Go with your gut.
  3. Expand your targeting using groups. As you create groups, it will be easy to start finding more users to target. So if you have tech blogs like TechCrunch on your list, you can add other ones such as PandoDaily, BusinessInsider and even CNet. Be careful though: there may be a reason your users don’t already like those pages. At Lyft and Udemy, we would use separate ad campaigns for “related” groups and monitor performance accordingly.

That’s a wrap folks. If you have questions, please feel free to ask and I’ll try to get to them.

P.S. This is why we started the Growth Hackers Conference and why I regularly read blogs like Andrew Chen’s or Sean Ellis’s. If you like this, I’ll also try to blog more to help share this kind of information. Tips like this used to be locked up in people’s heads — so poor entrepreneurs like me could never learn them. Now, you can pay $300 (with coupon code “FBadv”) and save months of time and thousands of dollars by going to conferences and reading blog posts about growth hacking. You might also find your next opportunity, meet a great candidate or connect with an industry insider who mentors you.

Written by Andrew Chen

October 28th, 2013 at 10:04 am

Posted in Uncategorized

Use this spreadsheet for churn, MRR, and cohort analysis (Guest Post)

without comments

[Andrew: Christoph Janz has written some of the best essays on SaaS metrics and cohort analyses, and he was kind enough share the latest with us below. A bit about the author: Christoph is co-founder and Managing Partner at Point Nine Capital, an early-stage venture capital fund with a strong focus on SaaS investments. Their investments include Zendesk, FreeAgent, Clio, Geckoboard, Contactually and Unbounce. Christoph blogs here and is the creator of a popular SaaS metrics dashboard]

Christoph Janz, Point Nine Capital
Cohort Analysis Spreadsheet

If you’re a long-time reader of my blog (or if you know me personally) you’ll know that cohort analyses are one of my favorite tools for getting a deeper understanding of a product’s usage. Cohort analyses are also essential if you operate a SaaS business and want to know how you’re doing in terms of churn, customer lifetime and customer lifetime value. I’ve blogged about it before and have included “Ignore your cohorts” in my “9 Worst Practices in SaaS Metrics” slides.

My feeling is that over the last 12 months the awareness for the importance of cohort analyses has grown among startup founders. One reason may be that thought leaders like David Skok have been writing about the topic, another reason are web analytic tools like MixPanel and KissMetrics that make it simple to create cohort analyses.

And yet, many founders are still having difficulties with cohort analyses, be it with the collection of the data or the interpretation of the results. With that in mind I wanted to create a simple cohort analysis template for early-stage SaaS startups.

Download the Excel file here.

The idea is that you have to enter only a small amount of data and everything else is calculated automatically. Specifically, what you’ll have to type in (or import from a data source) is the basic cohort data: How many customers did you acquire in each month and how many of them were retained in each subsequent month. If you also want to see your churn on an MRR basis and get a sense for your CLTV, you’ll also have to enter the corresponding revenue numbers.

💌 Are you up to date?
Get updates to this essay, plus more on cohort analysis and SaaS growth.

If you’re not sure how to read a cohort analysis, here’s a quick explanation:


Here are some brief notes on each of the arrays in the sheet:

A1: This is where you enter the raw data. Start with January 2013 and enter the number of new customers that you’ve acquired in that month. Then move to the right and enter how many of those January 2013 customers were still customers in February, March, April and so on. Then move on to the next row. If your data goes further back than January 2013, extend the table accordingly.

A2 and A3: A2 takes the data from A1 and shows it in “left-aligned mode”, making it easier to compare different cohorts. As you can see the columns have changed from specific months to “lifetime months”. A3 shows the number of churned customers as opposed to the number of retained customers. Both A2 and A3 aren’t particularly insightful to look at per se, but the data is necessary for the calculations in B1, B2 and B3.

B1: Shows the percentage of retained customers, making it easy to see how retention develops over time as well as to compare different cohorts with each other. What you’ll want to see is that younger cohorts are getting better than older cohorts.


B2. This is kind of like the  “inverse” of B1, showing the percentage of churned customers as opposed to the percentage of retained customers. In any given row, the sum of the percentages of churned customers plus the percentage of retained customers equals 100%.

B3: B3 is similar to B2, but the difference is that churn isn’t calculated relative to the original number of customers of the cohort but relative to the number of the cohort’s customers in the previous month. Let’s say you have a cohort with 100 customers and after 6 months the cohort has been reduced to 50 customers. If you lose 5 customers in month 7, this represents 5/100=5% churn in B2 but 5/50=10% churn in B3.

So what’s the correct number? There’s no right or wrong here, it depends on the question that you want to ask. If you want to know e.g. “How many customers do I lose within the first six months?”, B2 (in conjunction with B1) gives you the right answer. But if you want to know what percentage of customers you’re losing per month (important when you look at data across multiple cohorts and for lifetime estimates), take a look at B3.

What you’ll want to see in this table is that after a usually relatively high churn rate in the first lifetime months churn starts to stabilize (because the people who never really adopted the product in the first place are now gone).

C1-C3: Same as A1-A3, just for MRR instead of customer numbers.

D1-D3: Same as B1-B3, just for MRR instead of customer numbers. What you’ll want to see is that your MRR churn is lower than your customer churn due to account expansions.


E1 and E2: If you enter the CACs for each cohort, these tables show you when each cohort breaks even.

Also take a look at the second tab in the Excel sheet, which calculates/estimates customer lifetime and customer lifetime value on a cohort basis. Note that the data is highly speculative for younger cohorts for which there isn’t much data yet.

Further notes are included in the Excel sheets.

If you have any questions or comments, please feel free to reach out!

Written by Andrew Chen

October 24th, 2013 at 10:30 am

Posted in Uncategorized

Zero to Product/Market Fit (Presentation)

without comments


Starting from a blank canvas is the absolute hardest thing to do. It’s insanely hard. And yet the brave entrepreneurs in the tech industry do it, time and time again, and every year, a few are successful. It’s an awesome thing when it works though.

Anyone who’s working on a new product has question: How do you get to product/market fit? If you’ve been wandering the desert and you can’t retain your users, is there anything you can do about that? Is there a way to add structure to something that seems awfully random? How do you even know you’re there?

Below is a short and sweet deck that I created and presented a few years back to Stanford students. It’s focused on a simple idea: The path from zero to product/market fit can be a straight line, or windy, depending on what kind of idea you choose. I present the tradeoffs involved in creating a product that’s a bit more incremental, versus something that’s more breakthrough.

Although this deck is mostly focused on consumer and consumery enterprise products, it’s also fairly general, drawing ideas from the book Blue Ocean Strategy and others.

Hope you enjoy!

San Francisco, CA

📥 Get this deck as a PDF, plus new updates and essays in the future:

Above: It’s difficult to product/market fit, but today we’ll talk about some of the tradeoffs you can make for it to be easier!

Traction is everything – it’s what investors ask for. It’s what prospective employees and your team is looking for. But getting traction is really a reflection of your product “working” – where users are engaged and retaining, and there’s some organic acquisition. Product/market fit is when people who know they want your product are happy with what you’re offering – more on the topic here at the a16z blog.

Above: Important, once you get to product/market fit, you can shift your focus from the initial product to distribution and win the market. It’s important to do it in this order! In the haste to get to market, there are teams that start spending too much time on partnerships, paid marketing, etc. – all in preparation for a big launch. But the problem is that if you have a leaky bucket, then all those users at the top of the funnel won’t help.

Above: So what does product/market fit look like from a metrics standpoint? Really the best metric to think about is probably retention, which you’d analyze using cohorts. That means it’s sticking, and you need to figure out acquisition. Once it’s all figured out though, then you get the hockey stick. That’s what we’re all looking for, right?

Hitting the inflection point means that you can focus on scaling growth. The activities here are many of the things that I write about – funnel optimization, new user activation, creating notifications and email drip campaigns, etc., etc.

But it’s important to think of these are distinct activities.

If your product isn’t working, you won’t solve that by doing growth activities.

There’s a certain lightning-in-a-bottle aspect to building new consumer companies for that reason.

This is why 20-something year olds often build awesome new companies in Silicon Valley- they make lots of stuff, hit product/market fit, and the capital/talent in the Bay Area comes to help them scale

Entrepreneurs who are young are less encumbered by the notion of “this is how things have always been done” or “this kind of thing has never worked.” They can try, fail, iterate, and make their way towards something pretty new.

Above: Here are some specific ways that I might quickly judge if a consumer product has p/m fit.

You want to see DAU/MAU at >25%. A world-class leading DAU/MAU would be over 50%.

There’s a certain minimum for organic acquisition. You want to see hundreds if not thousands of signups per day, and a D1 of at least 30%. >70% would be world-class there.

And ultimately there’s some scale to show it works – probably 100,000 DAU.

For a SaaS product, you’d like at other metrics that might focus on monetization rather than engagement. Although some products can be evaluated using both sets of metrics.

Above: If you think you have that, send me an email :)

If you compare the metrics above, most startups don’t have product/market fit. It’s usually not working.

So what do you do?


Above: These days, most startups fail because of lack of P/M fit, not technology risk. So if there’s one thing that will kill you, it’ll be the product never quite working, and thus, all the subsequent problems that come with that: Lack of investor interest, employees leaving, cofounder fighting, etc. It’s a stressful time when it’s not working

I find that product/market fit is usually reached either right away, or you have to be incredibly thoughtful about your iterations if you’re almost there but not quite.

So how do you get there?

Above: I want to present to you one framework to think about this, and what kinds of tradeoffs you can make.

Above: Here’s a coffee cup. Product/market fit is actually easy to get, for a coffee cup. It has to hold coffee. That’s about it. There’s not too many variables in designing a cup, and so there’s not a lot to mess up.

Above: Contrast that to a digital product, where there are a million variables. It’s so much more complex than a coffee cup! It’s easy to mess something up.

(The above is a Facebook experimental project, called Paper, that never quite worked out)

It’s easy to mess up because the software medium is so rich. You can build or create anything, and you often end up with something that users can’t quite understand. Worst yet, once it’s not working, adding features won’t help! Sometimes the foundation is just not strong.

Keep these examples in your head, as two extremes on a spectrum. The coffee cup. The messy random digital product.

Above: The more I’ve thought about this problem, the more I’ve come to realize there’s really a spectrum between new versus pre-existing product categories.

The new product category carries a lot more risk, which we’ll discuss.

The pre-existing category carries less risk, but you face more competition, more expectations, etc.

What you want is a point in between those two extremes.

First, over time I’ve come to believe that you actually do want your product to be in a pre-existing product category. Google was not the first search engine, Facebook was not the first social network, and Microsoft was not the first OS.

However, those companies came to dominate those markets because they came in early, when the dynamics were still developing. And the markets grew and grew. But it was also clear what they offered, how they developed something killer.

Above: I want to be more precise than to say that the market must be big – more specifically than that, there must be a large number of users who are pulling on this category in the market. By pulling, I mean folks are actively searching for the product. You can see it in Google Trends. They are clicking on links that mention products in the category. There’s already spend.

Let’s contrast this to the big fake market sizes that entrepreneurs often trick themselves into. Yes there are a lot of college students, and maybe that’s your target market. But how many of them actually want this? Already own a precursor version of your product, or are googling or searching or talking about the category? That’s the real assessment of demand.

Above: Related to the above point, the best markets already have competitors. And maybe they are pretty decent. It’s a good sign when folks can make a decent living building products in the space, but you just have to find a wedge to get in. Or maybe you have a clear thesis on why and how the market will get a lot bigger quickly.

The best case is where the market is large, but fragmented – perhaps because the competitors are not that competent or there’s something structural, like regulatory issues, that keep things small.

Above: If you are in a pre-existing market with competitors and pre-existing demand, then you need a wedge. You need to analyze the market, understand the segments, and figure out how to break in.

My favorite book on this topic is Different: Escaping the Competitive Herd. Worth reading.

Above: There’s a huge advantage to building for yourself because the intuitive advantage it gives you in iterating quickly is fantastic. This isn’t always possible, but it can be a good thing.

Just as we discussed with the coffee cup, one way to get to product/market fit quickly is to just copy something.

There are a ton of problems with this strategy though – if you’re the second player with the same product, you’ll always remain the second player. Your team will hate this strategy. Investors won’t get it.

Sometimes the market is big enough to make this work, but it’s not often.

Above: The better approach is to do a twist. On the spectrum of pre-existing market to new market, pick a point in the middle. Build something that consumers fundamentally understand, but with a clear innovation that you can market around.

A lot of getting to product/market fit is being thoughtful about the category, the axis of differentiation, etc.

I want to run through some common mistakes that people make.

Above: We love these “X for Y” type companies. And it can be a way to generate cool ideas, but it doesn’t say much about the underlying market size and existence of pull in the market.

For instance, “Pinterest for dogs” and “Pinterest for businesses” are both X for Y ideas, but only dogs are a segment of Pinterest. A business-targeted product is more a description of the mechanics of the product, not a smaller set of the broader Pinterest market. And as we mentioned, identifying a wedge in a pre-existing market is key.

📥 Get this deck as a PDF, plus new updates and essays in the future:

As a result, you can think about existing versus new markets using The Substitution Test. If I use this new product you’re building, does it take away engagement or $ from something else. If so, then it’s truly a segment and a heads-on competitor. It’s counter-intuitive but I consider that a good thing, because they you are truly competing in a pre-existing framework- you just need the right functionality to win.

Above: Every existing market has a baseline of product quality, functionality, etc. We’d all love to build the minimum viable product, but sometimes it doesn’t work because the category has evolved sufficiently that you need more than the bare bones.

Sometimes people love a certain kind of tech. Today it might be crypto, tomorrow it might be something else. And they are looking for some kind of product and category to apply to, rather than fundamentally solving peoples’ core needs. It’s an easy mistake to make for technologists.

As you know, I hate new consumer behavior :) Sometimes this can work but it’s rare. At the very least, products have to tap into the same fundamental human motivations that have driven us for 100,000s of years.

Above: This is a San Francisco thing, but entrepreneurs are often building things that sound cool only to each other. No one else cares, but it’s fun and captures some tech trend that’s confined to the SOMA bubble. I call this “art for artists”

Above: This is common. You can execute everything correctly – iterate quickly, great tech, hire well, etc. – but ultimately still take huge risks on your product and market in such a way to not actually be successful.

Above: So let’s say you get something that has P/M fit, and it starts to take off! It’s a good problem to have. So what do you do?

Above: Luckily, you can summon the power of the Bay Area to help you solve this.

There’s a large body of work and experts on scaling companies. You can raise money, hire operators who’ve scaled, and you’re on your way.

I don’t mean to downplay this, except to say that the zero to product/market fit part is so hard that this next stage will be much easier than the first. Ultimately, there’s a few growth channels that work, and it’s much more of an optimization problem.

Above: For the paid ads channel, you’re ultimately looking at it as a LTV versus CAC problem. With some blended/organic traction too.

The ecommerce companies, OTAs, and marketplaces often grow like this.

Or you want to build a viral loop – if your product is social or collaborative in nature – and you can optimize for invites or content sharing or some other loop. This is a deep topic that could be a multi-part series of its own things.

This channel is used by the social networks, video sharing products, but also even productivity/collaboration products in the workplace.

If your growth loop involves a lot of user-generated content, then you can build on Google. For products that are really about reviews on local businesses, companies/products, or even real estate, we see them use this loop.

Above: If you’re at product/market fit, I want to hear from you! Send me a note anytime. Thank you :) (For future reference, you can also download the PDF here).

Written by Andrew Chen

October 14th, 2013 at 10:50 am

Posted in Uncategorized

The Rise of Fat Venture Capital

without comments


Reinventing the VC industry
In 2007, before YCombinator and AngelList had changed the industry, I worked in a nondescript office park in the heart of the venture capital industry off Sand Hill Road. Amid the leafy sprawl of buildings next to 280 and Stanford University, billions of dollars were and are invested out of the fancy offices of VC/PE firms you’ve never heard of. The whole industry has been shrouded in opaqueness since it was created decades ago, built on relationships from business schools, professional networks, and investor referrals. In 2007, I worked at a big firm as an Entrepreneur-in-Residence, and did my best to make sense of this world.

The first thing I noticed was that the venture capital industry seemed so opaque as to be self-defeating. Entrepreneurs didn’t know who to talk to, and most can’t name more than 2 or 3 firms. While venture capitalists would pontificate about the importance of differentiation, all of their websites consisted of a bunch of blue-shirted dudes that all invested in the same stuff. This always confused me.

I remember at one point I had lunch with a General Partner of a Sand Hill firm and we talked about PR:

Me: So why don’t VCs do more PR?

General Partner: Great entrepreneurs know to come to us. We don’t have to go to them. And in fact, having great investment returns and being out there in the public don’t really correlate, so we don’t do it.

It sorta made sense to me then. This approach had been true for the first few decades of the venture capital industry. Getting ahold of a GP through an entrepreneur’s network seemed like a basic test of competence.

Thus, other than raising the money, you didn’t need much overhead to run a VC. You just needed a couple guys in a nice office, 1 admin per partner, and a website. If you wanted to get fancy, you could have a venture partner or associate or two. Maybe 6 professionals plus their admin staff.

But just a few years later, in 2013, we now see VC firms with over 80 professionals. Not just one, but multiple firms are doing this. And even small funds are experimenting with events, content marketing, software infrastructure, etc.

This is the new era of “Fat Venture Capital.”

How did this happen?

Lots of trends driving differentiation of capital
There’s a lot of  trends driving VCs to differentiate. The obvious stuff: It’s cheaper to start a company. There’s more seed money, in the form of both accelerators writing $20k checks and seed funds writing $500k checks. There’s a lot more information out there, for both what firms to pitch and what terms to expect. There’s been a lot of great material covering the trends affecting startups so I won’t elaborate more here.

What’s less frequently discussed has been the recent trends on the venture capital side. A lot of firms decided that the high market risk, low tech risk nature of digital bets meant that they needed to employ a “barbell” strategy. Lots of seed stage investments, plus lots of late-stage investments, and not much in-between. Many early firms that traditionally would lead Series As and then hand off the larger “late stage venture capital firms” now see themselves playing at every stage. But if you wait to do later stage bets, then the traction is more obvious, and there’s more head-to-head competition between firms as they chase obvious deals. More head-to-head competition means that differentiation matters.

Similarly, the lower capital requirements of startups means that there’s been a lot of new firms started recently. It’s a lot easier for some successful startup execs to start a new VC fund by raising $20M or $50M, rather than a traditional VC that raises $200M or more. This has brought a lot of entrepreneurial energy to a sector that often behaves like a sleepy money management industry, rather than the dynamic startups in which they invest.

This entrepreneurial energy is especially important in an industry where the guys who invented the industry are now long retired. When you read books or watch documentaries about the early VC industry, you can see that the early guys who knocked on the doors of insurance companies in the midwest were truly entrepreneurs. Years later, many firms are led by professional investors who are two generations removed from these early innovators. While the successful-entrepreneur-turned-VC is still widely admired, plenty of firms are stocked with MBAs-turned-associates-turned-VCs, who are prone to view VC as a career rather than an competitive and entrepreneurial endeavour in itself.

And of course, many large firms are simply reacting to the competitive pressure from Andreessen Horowitz. They popularized this services-based approach and this WSJ interview is worth reading about their approach. The CAA analogy is particularly insightful.

The unbundling of the General Partner
The impact for entrepreneurs is straightforward. It used to be that an investment from a General Partner of a VC was a bunch of things bundled together:

  • Money
  • Expertise
  • Oversight
  • Professional network

These days, we’re moving to a model where this all gets unbundled. Money comes from the VC, but also from a long list of investors sourced from crowd funding platforms. The professional network for a firm is supported by a large services team. Advice can come from a long list of advisors and operational folks that are easy to track down on LinkedIn. Functions like executive recruiting, technical recruiting, PR, etc. don’t come from the particular GP who wrote the check, but rather, the services team that works for them. Instead of the GP sending you random links they’ve read, instead there’s dashboards and link-sharing services that are private amongst the other portfolio companies of the investor.

Furthermore, we’ll see the internal functions of a VC firm, like marketing, evolve to be pointed towards entrepreneurs rather than investor management. Instead of press releases on PEHub, we’ll see firms act more like SaaS marketers: Content marketing, professional events, even paid advertising. Why shouldn’t a VC firm be buying Facebook ads targeted at the next crop of Stanford CS graduates?

Ultimately great VCs will continue to do well. Many won’t embrace the Fat Venture Capital model, but will do fine, because their judgement, expertise or network is just that good. But for the rest of the industry, moving to differentiate will be the norm.

Entrepreneurs will benefit. More transparency and competition will mean that the “the good ones will come to us” attitude will be a thing of the past. Instead, great VCs will chase the great entrepreneurs, because in a world where the supply of great entrepreneurs is smaller than the supply of plain ol’ money, that’s the way it should be.

Written by Andrew Chen

October 9th, 2013 at 10:30 am

Posted in Uncategorized

How Google and Zynga set & achieve meaningful OKRs (Guest Post)

without comments

[The topic of setting goals, especially quantitative ones, is a really important topic for any team and any startup. A common format has been the OKR, or “Objectives and Key results” which is in use at Google, Zynga, and many other companies. It consists of setting an objective, and also a series of measurable results aligned with that objective. My good friend Kenton Kivestu, (now Flurry but previously Zynga and Google) has experienced this framework first hand and had a lot of insightful stuff to say about it. You can follow Kenton on Twitter and read his blog here. -Andrew]

Kenton Kivestu, Flurry:
How to set & achieve meaningful OKRs

In 2011, when I joined Zynga to work on the mobile poker franchise, we were getting trounced by a competitor named Texas Poker. Their UX was better. They were smoking us on appstore top grossing rankings. And they had 5x the features (not to mention a “premium” pro version of the game they’d just launched). For a company who unequivocally dominated the poker space on FB, our position in the mobile poker market was borderline comical.

The team set an OKR to take the throne: become the #1 top grossing iOS poker game. And then something incredible happened, we did (~6 months later). In my career, I’ve seen many an OKR go haywire (both at Google and Zynga) so this post is my attempt to distill & isolate the common traits I’ve seen in good implementations of an OKR.

First, what is an OKR and why bother?
Google has used OKRs since 1999 at the urging of KPCB partner John Doerr and Rick Klau (former Googler, now Google Ventures partner) has a good post on what is an OKR is (and it’s history). And you can read up on them at Quora too. But the TL;DR is an OKR is a stated goal, known to the whole company and has a pre-defined rubric to measure your success in achieving it.

Whether your at a start-up or big co, the only thing in endless supply is constraints. Time, developer resources, energy and the list goes on. And in a resource constrained world, the best plan of attack is to marshall resources and focus efforts on the best leverage point.

Now the question is, how do you set good OKRs? Like all things, there are many ways to successfully skin a cat but the 3 most common traits I’ve seen in teams that have set and achieved awesome results with OKRs are these:

Contrary to popular belief, this doesn’t mean the the OKR needs to be explicitly quantitative (eg drive X% increase in sign-ups or Y% year over year growth in recurring revenues), although it’s fine if that is the case. But it definitely needs to be measurable in the sense that the team can unequivocally evaluate their progress at the end of the OKR period.

For us, the iOS app store Top Grossing rankings were our measuring stick (for better or worse). There was no fudging the numbers and it was dead simple to measure, we were either ahead or behind.

OKRs are like money. Mo’ money, mo’ problems. The surest way to negate any positive impact from a good OKR is to set 10 good OKRs. It can seem alluring at first – “we’ll accomplish so many things this quarter/year!” – but it will backfire. There is little doubt that 10 good OKRs is worse than nothing. Your team will be torn between competing priorities – should John the engineer work on X or Y this sprint? One will get us closer to OKR A and the other toward OKR B. Inevitably this leads to prioritizing OKRs – this might work if you have 1 or 2 but anymore than that is going to be a recipe for a wasted quarter.

For us, we had 1. If a feature/chunk of work didn’t directly contribute to us climbing the iOS top grossing rankings, it was de-prioritized. As a result, we got a late start on expanding to other platforms (Android tablets, Kindle Fires). That was painful but the focus brought the right end result, since the iOS Top Grossing “pot o’ gold” was orders of magnitude more rewarding than early adoption of Fire or And tablets.

Worth doing
Again, this seems obvious but is worth stating. A good sanity test is ask yourself, “If the company were a person, would it put the successful completion of this OKR on its resume?” If the answer is no, boot it. All too often well-intentioned but “empty” OKRs end up dictating resource allocation. These culprit OKRs typically rise out of a discussion around some product tactics (eg let’s reduce friction in the sign-up flow by X%).

Ok, but OKRs stunt our creativity
An oft argued counterpoint is that OKRs will stunt creativity and the team’s ability to tinker and meander into some great discovery. Defenders of this theory highlight examples of accidental discoveries leading to huge innovations – but this is missing the point. OKRs don’t preclude accidental discoveries, they simply make sure that in absence of a brilliant accident the team is on track to do something else meaningful.

If you’re finding your team sets quarterly OKRs only to trash them each quarter given a brilliant mid-quarter discovery, you’re either:

  • a 2-sigma team that repeatedly launches industry leading core features every quarter despite not initially planning them
  • setting OKRs not “worth doing” as evidenced by repeated willingness to ditch them
  • unfocused and need more operational discipline

In conclusion
OKRs are not a panacea. And they can lead a team astray if they don’t keep them focused, measurable and meaningful. But the flip side is that a well-set OKR can generate a 10x result. OKRs can drive focus and relentless execution, and in turn, those drive incredible results.


Written by Andrew Chen

October 1st, 2013 at 11:00 am

Posted in Uncategorized

Case studies from “Why you can’t find a technical co-founder”

without comments

This is a followup guest post by Elizabeth Yin on her popular essay, Why you can’t find a technical co-founder. Liz is the CEO and a co-founder of LaunchBit, an ad network for email newsletters.  Previously, she worked at startups and Google, and went to MIT for her MBA, and Stanford before that.

Elizabeth Yin:
Why you can’t find a technical co-founder (part II)

I received a lot of emails about my last post on Andrew’s blog: Why you can’t find a technical co-founder, which discussed:

  • Deal-breakers for non-technical entrepreneurs who are looking for technical co-founders (for example: location isn’t a deal breaker but idea validation is)

  • How to get traction on a product idea without a product

  • 3 examples of startups (AngelList, Yipit, and Beat the GMAT) who have become successful today but started without full-fledged products

So, I wanted to dive into this topic more.

Building a minimum viable product with a static website
It can be all too easy to want to build out a full-fledged minimum viable product in code.  But, as an entrepreneur, you only have so much runway.  One of the best ways to cut down on the time required to get to product/market fit is to skip the coding process entirely.  It’s more time-consuming to program in one direction only to realize that you need to build something in a completely different direction.  Not to mention, morale takes a hit when you have to scrap all your work.

So, one of the best ways to speed up time-to-market is to build a minimum viable product with a static website.  Disclaimer: this isn’t possible for every product, but a lot of web business ideas can be built initially with simple static websites.

Here are two case studies of companies that first used simple websites to test their respective business ideas before building out full-fledged dynamic sites.  It was these static websites that helped them find product-market fit very quickly, which enabled them to scale sooner.

Fandeavor is a company that offers game-day experiences.  Founders Tom Ellingson and Dean Curtis previously worked at Zappos, a company that would often sponsor sporting events.  Tom and a handful of Zappos employees would often get red carpet treatment at these sporting events, but these special experiences were limited only to high-end sponsors.  “But what if general consumers could purchase the same awesome sporting experiences too?” they wondered.  So before they left their full-time jobs, they decided to test whether this idea had any legs.

Tom called up a contact at the University of Nevada Las Vegas (UNLV) who ran sports tournaments.  Tom convinced him that the Fandeavor team could help him sell special sporting experiences around the upcoming tournaments.  These gameday packages would include things like signed basketballs, box suites, special sports gear, and even the opportunity to present the game ball at mid court.  In parallel, Dean set up a basic website that would provide information about these gameday experiences.


Fandeavor’s initial website

Through a cross-promotional partnership with UNLV, who promoted these experiences through their Facebook fanpage, these first experiences quickly sold-out.  It was from testing these first experiences that the Fandeavor team realized that doing cross-promotions with other organizations to promote their gameday packages helped them build their audience and also made their sales successful.

The duo continued testing their business idea by offering more experiences on their website.  They even created sections of their site that were empty but helped track demand of what teams and sports people were most interested in.  Just simply by tracking clicks on their relatively simple site, they could figure out what types of experiences to offer next.

Tom and Dean left their day jobs at Zappos to work on Fandeavor full-time.  But soon, they realized that the kind of experiences people were interested in were not necessarily box suites.  A lot of consumers wanted even simpler things — help with travel, hotels, and other logistics in building a great trip experience around tournaments and games.  These were experiences that the team could build with virtually no special business development deals.  So, they started curating these experiences themselves manually.  Eventually, they were able to develop a process around doing this efficiently.

From doing quick tests using a simple website, Fandeavor was able to figure out the mechanics behind how to get their supply (the right curatable gameday experiences) and the demand (promoting through cross-promotional partnerships).


Fandeavor’s website today

Once they established the processes for scaling both sides up, they were able to later build a much more sophisticated backend to curate all their experiences and build their team to repeat these processes.  Today, Fandeavor has raised $525k and is growing 50% month over month.

Moveline is a company that aggregates moving quotes from moving companies.  Founders Kelly Eidson and Fred Cook realized that the moving process was difficult for people for a whole variety of reasons and wanted to help make this process better.

Fred and Kelly collected leads on a simple website to start working with qualified consumers.  And soon, they started going into homes of people who were moving to talk with them about moving issues.


Moveline’s initial website

Kelly and Fred quickly learned that it was a headache for people to document every little thing they had to move and then submit all those items to different moving companies to get quotes.  Kelly and Fred realized that the crux to making the moving process better was to develop a better way of itemizing objects.

At first, the Moveline team was not really sure how to do this better.  They initially went into people’s homes and manually categorized items into spreadsheets.  They repeated this with dozens of people who were moving.  The breakthrough in building a process around this came when someone in a different city contacted them requesting their itemization-help.  Not able to physically go to that person’s home, Fred and Kelly asked to do the itemization over Facetime.  It turned out they were able to accurately and quickly itemize over the video conferencing software, which later became a part of Moveline’s core product.  It turned out that using video conferencing software gave them an edge — they could itemize just about any move from anywhere without requiring local on-the-ground teams.

It was from these initial conversations with dozens of consumers that Moveline was able to tease out a very specific problem to solve in moving.  And, once they were able to manually figure out how to use technology to solve this problem, they were ready for scale.  It was because they used a simple website that fed them leads, but still required the team to “be the product,” they were able to really understand and solve this particular problem in moving.

Moveline’s website today

Only once they had a clear idea of what needed to be built into the product, Fred started hiring a team of engineers and product people to build the first version of Moveline’s software.  Today, Moveline is an 18 person company, has raised $3 million in funding, and since launching nationally in March has added over 200 moving companies to its network.  They now serve customers in over 100 cities in the U.S. and internationally.

Both Moveline and Fandeavor primarily used static websites to collect leads and kickoff interactions with potential customers.  Although static sites don’t seem very sophisticated, through the use of simple input fields and forms, you can collect information to vet potential leads, and you can use them to collect customer insights.  Static websites are a great way to quickly test your product and get to product/market fit.

For more examples of companies who built minimum viable products without coding, see these awesome posts written by Ryan Hoover and Vin Vacanti:

P.S. If you want to learn how to build your first web prototype without coding, attend this workshop I’m hosting on Building a Mobile-friendly Websites Without Code on October 3.  Get 25% off with this discount code: “andrew-hustler”.


Written by Andrew Chen

September 25th, 2013 at 11:01 am

Posted in Uncategorized

Ignore PR and buzz, use Google Trends to assess traction instead

without comments

[Yelp shows a healthy navigational search graph – lots of people are continuing to search for its brand, and you can see some seasonality where it peaks every August, goes down during Q4, then starts coming back up in Q1.]

PR buzz is useless for assessing product traction
Traction is everything, and it’s easy to confuse press buzz with actually having product/market fit. Writers for blogs and newspapers love novel ideas that sound amazing on paper, especially when the new products are being introduced by credible entrepreneurs. However, when you’re in the business of making product and investment decisions, it’s important to understand what’s actually working and what’s not- having buzz isn’t enough.

This article is about one of the ways to answer the question, “Is X product really working?” The easiest, fastest, free way to assess the traction of a competitor or buzzy startup is to use Google Trends. It’s a great tool from Google that gives you a chart of how many search queries are being generated, which is a fantastic way to see if the consumer pull demand is increasing or decreasing.

Navigational queries are the best representation of consumer “pull” from the market
The reason why this works is that Navigational Queries are one of the three major kinds of queries that consumers plug into search engines. People search for a brand like “yelp” “facebook” “zynga” when they want to directly navigate to that domain, and as you might imagine, it’s a strong indicator of customer loyalty. These navigational searches represent the “pull” of the market, and if the graph of this pull is flat or declining, then you might have a problem on your hands.

Consumer demand is the leading indicator, uniques are the lagging indicator
The reason why a graph of navigational queries is so powerful is that it partially removes three major sources of traffic which are often inauthentic, unsustainable, or susceptible to artificial inflation:

  • Unsustainable paid ads, where spending outpaces lifetime value of the users acquired. Or, commonly in mobile, where a bunch of ads are purchased in an attempt to shoot an app up in the charts.
  • Content farming, where a lot of low-quality content pages are created. Although a lot of users may end up arriving at these pages as a result of searching and clicking on something, they don’t know (or care) that they are on these pages- these aren’t really long-term users or customers.
  • Drive-by traffic, which often looks like photo/video hosting, or IFRAME’d content, which people click into from Twitter, Facebook, or some other social channel. Similar to content farming, these are often “one click wonder” visits where people click in to view some content, but don’t actually engage or care about the underlying product.

Of course, doing something simple like graphic navigational queries isn’t strong enough to remove the entire effect of the above- instead, it just mutes it a little bit, and that’s the important thing in the long run. Ideally you want to see a long, smooth set of traffic that’s been built over months and years, not a huge spike driven by unsustainable means.

This consumer demand curve is really the leading indicator of traffic. Generally if I continue to see the graph going up and to the right, I will think that a company is pretty healthy. If it’s not, no matter how much is being written about them in the press, I’ll be skeptical.

Zooming in on traffic seasonality
One of the best parts about Google Trends is how granular it gets. You can ask to see the graphs on a 30-day rolling basis, as well as breaking it down by country. One the common patterns you’ll see is that there are two kinds of websites:

  • Time-saving, where you use them at work Monday-Friday and they help you do your job.
  • Time-wasting, where you use them at home in the evenings, and heavily on the weekends. They spike Saturday/Sunday and flatten out during the weekday.

Here’s an obvious example of a time-saving product, Linkedin, which shows huge gains during the week but then it gets depressed during the weekends- amazingly the weekdays are 100% higher than the weekends, according to this chart!

Screen Shot 2013-09-03 at 2.56.35 PM

Triangulate with AppAnnie, Facebook stats, Twittercounter, Twitter search, and Quantcast
Of course, there’s a ton of caveats to using such a simple tool. Searches will rise even when unsustainable methods are used, just because more consumers are being exposed to the brand. Similarly, the other big issue is that mobile apps don’t often benefit from search engine traffic the same way that websites do, since none of the content in the apps are being indexed.

Thus, remember that this analysis is strongest for web products, and to think of this as directional.

If you want a higher degree of confidence, consider using AppAnnie, Facebook stats, and Quantcast to figure out what’s going on. With AppAnnie, you can see the rankings of mobile apps and how they change over time. This can be useful in conjunction with Google Trends, since you can look up something like Angry Birds, which used to be the top grossing iOS app, which you can see below:

Screen Shot 2013-09-03 at 2.49.03 PM

When you combine the above drop in rankings to their Google Trends chart, which I’ve included below, you can see that Angry Birds is stalling quite a bit:

Screen Shot 2013-09-03 at 2.50.58 PM

You can also use Facebook stats services, for the products that allow for Facebook sign-in, to get a sense for how things are directionally going. Quantcast and Compete are also free services that let you look up uniques/month for web products, though they are often wildly in conflict with each other since they use the same kind of sampling that makes Alexa unreliable too. Back in 2006 I wrote about how Alexa works and all the flaws with their methodology, that I learned first-hand working with Nielsen/ComScore in my adtech days.

Qualitatively, doing a Twitter search on the brand is great too. Ideally you want to see a ton of authentic tweets about people actually talking about the product- again, the focus is on whether or not people really understand they are using a product and what they think of it. You can supplement this with a service like Twittercounter to see if their follower count is growing well over time. If a product claims a ton of uniques in a press release, but very few followers and very little Twitter conversation, it might be inauthentic traffic. (Thanks to Adam Besvinick for reminding me to mention Twitter searches)

How are Airbnb, Foursquare, Pinterest, Twitter, Quora, and Yahoo doing?
And finally, I thought it might be interesting to share the current graphs of a couple current (and previous) high-fliers to see how they’re doing on Google Trends. Some of the are universally considered to be doing well, and some are not. let’s see what Google Trends says, at least on a high-level.

Seems to be growing very nicely. You can see some seasonal peaks in August where people rush to go on vacation for the summer, and then it rapidly drops off from there. The growth seems very strong.


This is a tricky one- you both see that it’s flat, which is in line with what’s being discussed in the press. On the other hand, since it’s mostly mobile, it’s hard to say if this navigational query analysis is that useful. The silver lining on this also is that the traffic has held steady and hasn’t declined for a year now, which means there’s likely a strong core of retention holding everything together. Requires some more analysis to figure out what’s going on here.


Same kind of indicators as Foursquare. There was a huge run up in 2011/2012, due to a lot of Open Graph nonsense, but now that the traffic spigot has died down, there’s still a good base of activity. As I understand it, they’re over 100M uu/month and holding onto that traffic, which should be enough audience to build a solid company.



Alongside Facebook’s graph, both seem like they are plateau’ing and going flat. But could this just be users transitioning from web-centric usage to mobile? I guess we’ll know when their IPO S-1 documents come out- are they accelerating audience growth or is is starting to slow down?


Although Quora hasn’t been in the news much lately, and a lot of digerati seem to have abandoned ship, their core base of traffic looks pretty good. Growing slowly. Building high-quality content via SEO is really hard- but it looks like Quora is gradually succeeding. Also note the epic launch curve, and that years later the current demand graph is still 1/3 of the initial peak.


It’s fascinating to me that Yahoo was actually still growing strongly up until 2010 or so. Now it’s flat and maybe even declining a bit. Curious to see if these curves can actually reverse themselves- a more detailed analysis would probably pull up these curves for every single important property, from Mail to News to everything else, and see if any one property is dropping faster than the others.


Written by Andrew Chen

September 3rd, 2013 at 3:12 pm

Posted in Uncategorized

Books I’m reading (2013)

without comments

A friend asked me what I was reading this year, so I wanted to share the sorta tech/nerd related ones at least, along with a quick blurb about what they are:

  • The Little Kingdom. Michael Moritz, in his previous job as a journalist, covers the early Apple years. Late on he wrote a followup, which I haven’t read yet. Great complement to all the contemporary Steve Jobs adoration, since it’s written from the perspective of the early days.
  • Expert Political Judgement. UPenn professors analyze why people are so bad at predicting all sorts of things in geopolitics, whether it’s elections or which dictators get deposed. Talks about two styles of analysis- hedgehogs which have “a big idea” and start their analysis with that versus foxes that try to analyze lots of data.
  • Predictable Revenue. Ex-Salesforce sales head breaks down how they sold to B2B. Lots of great details on how to organize sales teams, generate leads, incentive compensation, etc.
  • Engineers of Victory. Detailed dives into specific WW2 engineering problems: Defeating the UBoats, resisting the blitzkrieg, etc. Talks about how the engineers played a role in winning the war.
  • The Better Angels of our Nature. Amazing book by Steven Pinker, which I originally found via this glowing review by Bill Gates. He calls it one of the most important books he’s ever read. Pinker tells a compelling story, via graphs, anecdotes, and academic studies, about how violence has fallen over the last several thousand years.
  • The Signal and the Noise. One of my favorite books I read this year, by Nate Silver. Talks through how people go about modeling different things, whether it’s elections, gambling or weather. Lots of important points made about model errors and how people suck at predicting.
  • Sports Gene. After reading Malcolm Gladwell’s Outliers, this book is a great followup that talks more about the “nature” part of the nature/nurture debate. Talks about Jamaican sprinters, Kenyan runners, high jumpers, and the variance in the 10,000 hour “rule.”
  • Antifragile. Loved the first 1/3 and last 1/3 of this book. Taleb talks about the idea of antifragility, where things benefit from disorder. (Not just robustness, which resists disorder). He starts with the idea from a financial concept, but cleverly applies it to his own personal health and weightlifting routine. Could probably be shorter and less boastful though.
  • Your First 1000 Copies. Short and sweet book on how to build a mailing list to launch a book. A friend sent it to me after he started noodling on writing a book. I found some of the mailing list ideas helpful for this blog.

So there you have it! If you have more recommendations for what to read this year or next, shoot me a tweet at @andrewchen.

Written by Andrew Chen

August 26th, 2013 at 11:23 am

Posted in Uncategorized

Constrained media: How disappearing photos, 6 second videos, and 140 characters are conquering the world

without comments


Constrained Media. It’s an innovative category of products that ask invite users to create content on a platform, but with arbitrary constraints- Twitter’s 140 character is perhaps the most famous example.

There’s now been a whole series of these apps, quite successful ones, such as:

  • write in 140 characters or less (Twitter)
  • compose a 6 second video (Vine)
  • upload maximum 400×300 thumbnails (Dribbble)
  • view a photo in 3 seconds before it disappears (Snapchat)

Why would it make a product more successful by forcing constraints on content creation? Isn’t more always better? Wouldn’t each of these products be better off by removing the constraints? And does every constraint work, or is it all arbitrary?

I’ll argue that the constraints are a fundamental part of what makes the products work. The higher engagement in constrained media products is based on their ability to break through the 1% barrier for content creation. This 1% rule is the famous rule of thumb for user generated content services like Wikipedia or YouTube that says 1% of your users will create content, 9% will edit and curate, and 90% will just sit back and view.

Of course, having only 1% of your users actively creating content sucks. So let’s talk about how to fix that.

Frictionless content creation
The obvious thing is that constrained media apps make it easy to create content. Anyone can type in 140 characters, take a photo, or hit a button to compose 6 second of looping video. Constrast this to a big blank textarea like traditional blogging or a sophisticated photo tool like Photoshop, which requires much more creative energy to use.

More interesting is how these constraints impact the simplicity of the product UI. These constraints mean that the product can support a smaller number of use cases, making it more toy-like, and easy to use. Often, you can power the entire interaction with one button, like Snapchat or Vine. Just hit a button to create content, and once you hit the limit, it’s all over- no worries about editing and rearranging the content.

Both the simplicity of the content, as well as the product UI, makes the whole experience much more directed and higher conversion.

Communication rather than publishing
Building on easy content creation, the next step is shift the context to communication, rather than publishing, which encourages a higher level of participation. The 1% rule sounds good on paper, but think about it in the context of communication products. What’s the content creation % for email, IM, Skype, or texting? I’m sure it’s a lot higher than 1%, perhaps even close to 100%. The point of communication is that all parties involved create content that’s directed at other people, and everyone participates.

Twitter has @mentions, Dribbble has rebounds, and Snapchat is all about communication. This invites people to participate, because the media can be directed at other people, and there’s a built-in context to talk to one another. This leads to email notifications based on healthy user-to-user engagement. This drives frequency, virality, and all sorts of other good stuff.

Replying is easier than creating
Creating content from scratch is hard. Similarly, being the first to communicate can be hard- anyone who’s introduced themselves to a stranger knows the feeling. However, replying is easy. If someone takes a picture of themselves making a funny face on Snapchat, then a natural response is to make a funny face back. Even more if you know that the picture was sent specifically to you, then you feel like you owe a response.

If anything, this increases the constraints- you have the constraint of knowing who you should reply to, and also the constraint of the kind of content that was sent to you. And surprisingly, these constraints make it easier to come up with something to send back.

Make casual content OK by reducing the variance in effort
Nobody likes a showoff. And in fact, a platform with too many showoffs lead to funny social norms, where people tend not to participate because they don’t want to compete with those who are more skilled or who have more time.

Instead, constrained content creation reduces the variance in output between the low-skilled and high-skilled users, which makes it so that everyone can have fun. The best analogy for this might be something like kickball versus professional baseball, where the former is more about everyone having participate by “dumbing down” the sport, not winning. Dribbble is a community of designers where posting your in-progress work in 400×300 “shots” is part of the norm- meaning more frequency and engagement. Constrast this to portfolio sites that you update once a year at most.

Discoverability of content is an important factor too. If you make it too easy to find the more effortful or highest skill content, this creates a kind of leaderboard that discourages content creation, although the content consumption experience might be improved. It’s a tradeoff. Snapchat’s private, ephemeral context means that it’s the only place where it’s safe to post crappy selfies of yourself.

What do you do with all that extra engagement?
All of the above translates to more frequent, more inclusive content creation. This powers traction. More frequency of use means there’s more opportunities to take users through viral loops, as well as firing organic user-to-user notifications that power retention. It becomes easy, for instance in Snapchat’s case, to ask the user to include a couple extra recipients of a photo after you’ve replied. Or after you’ve created a 6 second video, it’s easy to ask the user to share it onto a couple different social networks.

So the next time you’re designing a new social product, consider adding a constraint, but not any arbitrary one. Make it one that makes content creation easy, more communication-oriented, and produces the social norms you want. That’s the best way to beat the 1%.

Written by Andrew Chen

August 5th, 2013 at 11:28 am

Posted in Uncategorized

The highest ROI way to increase signups: Make a minimal homepage (Guest Post)

without comments

Mattan Griffel has written some great essays on user growth over at Growhack, and you can  follow him on Twitter at @mattangriffel. In particular I’m fond of his essay The Minimal Homepage, which states something that everyone who’s A/B tested their homepage knows: Keep it simple, and ask for your signup upfront. It’s one of the easiest and highest ROI ways to increase signups, because your visitors won’t find their way onto low conversion pages and bounce. Surprisingly, it’s still counterintuitive to many. I’ve referred people to this blog post before, and Mattan graciously offered for it to be reposted here. Enjoy! -Andrew

Mattan Griffel:
The Minimal Homepage

What do you notice about the homepages of the fastest growing companies in the world?

Dropbox_Homepage-1024x715 Quora_Homepage-1024x715 LinkedIn_Homepage-1024x715 Groupon_Homepage-1024x714 Pinterest_Homepage-1024x715

Here’s what I’ve noticed:

  • No access without signup. Most startups make the mistake of giving people who visit their site free access to content, whether it’s apartment booking or daily deals. This is often a bad idea. Contrary to popular belief, the more things a visitor can interact with on your site before they’re prompted to sign up, the lower your signup rate will be.
  • Navigation and hyperlinks are almost always absent. Over the years internet marketers have developed what they call the “Squeeze Page” with minimal content and a single clear call-to-action because they discovered that additional information could distract a visitor or cause them to click away to a different website. Notice that there’s nothing below the fold on any of these sites.
  • Focus on a single, clear value proposition. In almost every case, the product’s value proposition is boiled down to one clear statement: “Your best source for knowledge” or “Be great at what you do”. People almost never read more than one sentence on your site (and they won’t even read that one unless it’s big enough and strategically placed), so there’s no point in trying to figure out your top 3 “bulletpoints”. This also makes it much, much easier to test as a growth hacker. Just replace one sentence with another until it works.
  • Your product is not about sharing. I see this mistake all the time. Lots of startups start out thinking that people will use their product because it helps them “share” things more easily. Let me be clear here: most people do not share. And even those people who share things aren’t sharing things 90% of the time. Most of the time on the web is spend consuming, not producing. More than 50% of Twitter users almost never tweet. This is why Twitter has shifted their messaging from “the easiest way to share with your friends’ to “Find out what’s happening, right now, with the people and organizations you care about”. If you cater only to proactive people, you’ll be alienating most of your potential users.
  • Big images. Big images increase conversion rates. Just do it.
  • Embedded signup forms. Start your signup process on the homepage so people don’t have to click through to a new page for no reason. Generally speaking, the more clicks you have in your signup process, the more people will drop off along the way. Note that these signup forms are almost always on the right-hand side, above the fold. They also rarely ask for more than a name, email and password.

When I tell people these things they often complain: “But everyone knows Twitter and Facebook, so they don’t have to explain what their product is about. No one has ever heard of [my startup] so I actually need to explain it to people.”

You are wrong.

Maybe you and I already know what Twitter and Facebook are about, but we’re not the people they’re trying to get to sign up on their homepage. 2.4 billion people use the internet and more using it each day. Believe it or not, there are still people on earth who haven’t heard of Twitter or Facebook. Those are the people these homepages are trying to convert – not the luddites who refuse to sign up (trust me, Twitter and Facebook stopped caring about them long ago).

The same is true for your startup. Don’t be stubborn and don’t think that for some reason your startup is an exception. Making that kind of assumption because you’re scared to try something counter-intuitive is a sure way to make sure you never do anything innovative.

[UPDATE: I read a great comment on Facebook and wanted to share it below. -andrew]
Emmett Shear makes a great point in a comment on this essay, included below:

I dislike essays like because while part of his point is valid (look at all these companies who have decided to gate things behind “signup first” and have very simple front pages!) there are tons of counter examples. Just look at the Alexa top 10.

#3 YouTube — putting a giant “sign up first” wall in front of YouTube probably would have killed them.
#6 Amazon — Amazon is all about converting people into accounts AFTER they decide to buy, and you better believe they’ve a/b tested it.
#7 Wikipedia — Primarily a read-first experience
#8 QQ – Holy crap that is a lot of text

Now Google/Facebook/Baidu certainly follow the “simple homepage” design. But he’s overgeneralizing terribly and shows no indication he’s aware of it. The point of thinking about design is to be aware of tradeoffs, not to push the latest trend as “the smart way to do it”.

Said another way, increasing signups isn’t necessarily important for every company, and many successful companies don’t focus on it. So I would restate to “here’s how to increase signups” idea with “here’s how to increase signups once you’ve decided that signups are important to increase.” Great point Emmett!

Written by Andrew Chen

July 29th, 2013 at 11:07 am

Posted in Uncategorized

9 ways a billion dollar new mobile company might be created (Guest Post)

without comments

My good friend Bubba Murarka recently started blogging over at He’s now a Managing Director at DFJ and tweeting at @bubbam. Prior to DFJ, he headed up Facebook’s Android efforts, and is an expert on all things social and mobile. He wrote the blog post below on his blog, which I’ve cross-posted here. -Andrew

Bubba Murarka on Mobile:

We’ve been in “New Mobile” – a world of wireless broadband and mobile OS platforms enabling great end user experiences – for about 5 years. The improvement in the capabilities of devices has been astonishing. But in truth we are still in the first inning of New Mobile reshaping just about everything we do and everywhere we do it.

Since leaving Facebook, I’ve been asked more and more for my perspective on mobile ecosystem. Here are my current observations on why New Mobile is still in the earliest stages:

  1. The move from feature phones – mobile phones without robust browsers or a compelling application ecosystem – to always-connected touchscreen computers in our pockets still has a long way to go. Smartphones are barely the majority of total mobile phone sales in the U.S., let alone globally.
  2. The industry talks about smartphones and tablets as both being “mobile” devices instead of seeing them as two very different beasts. This is starting to change and I’m excited to see the wave of companies that are “tablet first” – but please don’t let that become a mindless mantra!
  3. It’s no longer about iOS vs. Android. Now the hard question is whichAndroid versions (Gingerbread vs. Jelly Bean) and flavors (e.g. Samsung, Amazon, etc.) you are targeting and why. Said another way, Android fragmentation, and dominance, has just begun.
  4. Completing transactions on mobile is still a big hassle (except for M-Pesa). App store and carrier billing fees are too expensive to be an option for anything other than high-margin digital goods. Whoever cracks this in a way that any 3rd party app can use is going to be very rich.
  5. Content creation on mobile devices is horrible. Much of the content we consume on mobile today requires the capabilities of a PC to produce, including the keyboard, mouse and purpose-built apps. Products likePaper and Vine have shown that there is considerable demand for creation via the touchscreen.
  6. True mobile multitasking hasn’t been invented yet. Smartphone screens are smaller and better suited to handle one app at a time with abstracted file access. But we’re used to working with multiple windows and applications our computers with a global file system. When will a new UX model emerge, especially on tablets, to enable multitasking?
  7. There’s no “mobile native” ad unit to allow publishers to monetize their audiences and thus focus on building richer and more engaging experiences. Instead, startups have to spend a ton of time on business model innovation, which is another really hard problem to tackle. My money is on Facebook cracking this nut (full disclosure: I am still heavy on the stock, so my money is literally on them) though I think Yahoo could be a surprise contender.
  8. Only two types of paid subscription services have gained traction on smartphones: Content licensing such as Rdio and Pandora One, and storage such as Evernote. What else are users willing to pay a subscription for on their smartphones?
  9. There have been some billion dollar exits like Instagram and Waze, but we haven’t had a stand-alone, New Mobile company go from garage to an enduring multibillion-dollar independent company in the Americas or Europe yet (it has happened in China though).

There is a lot to be unpacked and everything above is up for debate as we refine our collective thinking through discussion. The only thing I know for sure is that I’m excited to learn about, identify and nurture the best mobile-focused companies out there.


Written by Andrew Chen

July 17th, 2013 at 11:00 am

Posted in Uncategorized

Mobile traction is getting harder, not easier. Here’s why.

without comments


The “classic” growth formula for mobile is broken
Once upon a time, the formula for getting mobile traction was something like this:

  1. Build something insanely great
  2. Get Apple/Google to feature you, alongside a big PR launch!
  3. Watch your app hit the charts
  4. Buy some cheap installs to propel it even further
  5. Voila, hockey stick! (and hopefully not a shark fin)

This worked for a few memorable years, and things were good- especially for new startups and indie developers. But gradually this classic formula stopped working, with nothing equivalent to replace it. Getting initial traction on mobile has gotten a lot harder, even though you’d expect a richer and bigger mobile ecosystem to have emerged to increase the opportunities to achieve mobile growth. (ps. if you’re interested in developing new approaches to mobile growth, just email me.)

It was only a matter of time. As I argue in my essay The Law of Shitty Clickthroughs, all marketing strategies eventually result in shitty results over time. In marketing, first-movers trump (at least initially) – if you do something new, then you’ll see high response rates as people respond to the novel tactic, whether it’s a new kind of creative, a new acquisition channel, etc. Eventually though, as your tactics become industry-wide “best practices,” the response rates fall as your customers get used to the techniques.

History has repeated itself again, within the channels that drive mobile traction. Let’s discuss how the ecosystem has matured, including factors like: Increased app store competition, Higher CPI rates, editorial dynamics, and the overall investment trend.

Product differentiation is harder with a much bigger app store
Let’s cover the most obvious thing first- the number of apps has gotten a whole lot bigger. Whereas before a new app might be competing against non-consumption, in all the major mobile categories there’s been a huge increase in the total number of apps. Those that were successful in 2009-2010 are now facing 4-8X the competition, if you look at just the aggregate numbers.

Whether you’re building an app for photos, shopping, messaging, local, movies, or news, there are now 2-3 very high-quality competitors in each category. A new mobile developer is no longer competing against the first wave of amateur-built apps. These days, it’s much harder.

Below is a recent chart that shows the incredible growth in # of apps:



Cost Per Installs have gone up over time
Initially, buying an app install was relatively cheap. You had a lot of options- everything from mobile ad networks, incentivized install providers, “free app a day” services, and even more adventurous options. More importantly, not a ton of companies were doing it, so prices were low.

This Cost Per Install has skyrocketed though, both due to demand and a lack of supply. After only a short time, the supply of paid installs has contracted as Apple has banned some providers and warned others. Similarly, mobile games figured out the enormous monetization potential on iOS and Android – they’ve bid up the installs significantly, up to a few bucks per install.

Here’s a chart showing the increase in CPIs over the first half of 2012, though anecdotally I hear it’s much higher than this now:


Editorial teams further the platform’s own strategic goals
The editorial teams inside the Apple and Google stores can certainly help some apps, and they do. Yet they are skewed more towards the needs of the consumer, and to the goals of the platform.

For Apple, my impression is that they care more that the first 25 apps that a user installs are amazing experiences from well-known brands, rather than servicing the needs of the overall million apps that in the store. As a consumer, I surely appreciate this, but it doesn’t help new unknown developers break into the market.

For Google, any team that’s met with them in the last few quarters can tell you that they care a lot about tablet devices. While they are winning market share on phones, the numbers for iPad versus Android tablets show a different story. If you want to be featured in Google Play, they strongly encourage you build a tablet app even if the market for it is tiny. They also care a lot about Google+, but that’s another story.

Investment has dried up for experimental new consumer mobile apps
While investors still have an optimistic outlook for the overall mobile market, there doesn’t seem to be a lot of conviction to deploy their capital on risky new consumer mobile startups. My sense is that there’s a feeling the ‘great consumer mobile experiment of 2009-2012’ has been run, where a ton of seed capital went into a wide range of mobile companies, and now the motivations have changed.

Just look at how the composition of YCombinator Demo Day companies has changed- in the late-2011 event I attended, it was >50% consumer mobile. Now it’s SaaS, consumer hardware, marketplaces, etc. Mobile is often an aspect, but no longer the main focus.

The silver lining
Despite the difficulties outlined above, I’m still wildly optimistic about the future of mobile. It’s still the best platform upon which to build a new company, but we must choose to embrace and work around the new difficulties we’re facing in 2013. It’s not enough to simply repeat what worked in the past- otherwise we’ll have a new generation of mobile companies that fail like it’s 1999, as I’ve written about.

While it’s getting harder, the opportunities within mobile are still the largest since the beginning of the computer industry. We’re barely over majority smartphones within the US, as Nielsen reported last month (June 2013). While it’s impressive that some apps have reached 100M+ installs, in an overall market of billions, we’re just getting started. We have a lot to look forward to over the next 10 years.

Written by Andrew Chen

July 9th, 2013 at 3:36 pm

Posted in Uncategorized

Why you can’t find a technical co-founder (Guest Post)

without comments

This is a guest post by a friend of mine on email marketing. Elizabeth Yin is the CEO and a co-founder of LaunchBit, an ad network for email newsletters.  Previously, she worked at startups and Google, and went to MIT for her MBA, and Stanford before that. PS. I’m training growth hackers. Email me.


image credit: SilentMode

Hey, I’ve got this great idea for a startup…do you know any developers who might be interested in working with me?

I get asked this question a lot.  So, my co-founder Jennifer and I were curious and surveyed developers on what would compel them to team up with a non-technical co-founder.

The results were surprising.  This survey was not particularly scientific.  We received 104 submissions from developers, of which 35 were actively working on their own projects full-time and 69 were not.  We asked participants to rate how important a particular criteria was to them in deciding whether to join a non-technical person’s startup.  1 = Not important.  5 = very important.

Location is not a deal-breaker


I would’ve expected location to be a deal-breaker for just about everyone.  I would have expected all would-be technical founders to strongly prefer being in the same city as his/her non-technical counterpart.  But, only about half said location was a deal-breaker.

Idea validation is extremely important


In contrast, idea validation was extremely important to potential technical co-founders.  You, as a non-technical entrepreneur, are not selling a dream or the vision.  You are selling traction.  Some people who took our survey commented about their ideal proof of validation, “If they have $1M in sales and have shown that people are willing to buy this thing without it even existing.”  Another mentioned, “Validated early adopters/customers [are people] who said they’re going to pay for the product when their minimum viable product is out for their use.”

Prior relationship is not a deal-breaker


Also interestingly, I would’ve expected developers to overwhelmingly prefer to work with people they’ve already worked with or know from before.  And while, the data shows that many people would prefer to work with someone from their past, about 40% of technical folks don’t really care.

Pulling this all together, if you’re looking for a technical founder, the number one thing you should be doing is to get traction for your startup idea.  This means validating your idea, getting customers or users, and ideally getting revenue.

Getting traction without a product
So how do you get traction without a developer to build the product?  Very much in the spirit of the Lean Startup Methodology, there are a number of successful tech startups that got started without doing any programming.  Here are 3 companies that took off without writing any code in the beginning.

Fast-growing startup Yipit, a deals-aggregation company, got started in 2010 as a side project without any code.  The founders, Vin Vacanti and Jim Moran, wanted to just get Yipit out the door in a couple of days, so in the beginning they manually aggregated deals from major daily deal sites — Groupon, LivingSocial, et al — by hand.

Screen Shot 2013-06-17 at 2.30.03 PM

They put up a landing page to aggregate email addresses and collect category preferences.  Then, they manually categorized each of the deals they collected and emailed their subscribers based on indicated preferences.  In a the true spirit of hustling, they did not build a web crawler to aggregate the deals — they manually aggregated deals themselves at 3am everyday.  As Yipit started getting traction, it was getting more unwieldy to handle, but instead of hiring developers to build out web scrapers, they hired more people to continue manually aggregating, categorizing and emailing deals for 9 months, because the Yipit team wanted to continue to learn how to tweak the product quickly to make it better.  Since those early days, Yipit has since raised $7M+ in total.

Beat the GMAT
Beat the GMAT, a social networking site for prospective MBA students, started in 2005 as a side project — just as a blog.  The founder, Eric Bahn, used his blog to solve his own GMAT problems to help him practice for the exam.  His blog became so popular, readers started emailing him to ask for help on problems.  Although Eric would email people back, he was soon receiving 50+ emails per day from blog readers.  He realized he needed to scale himself.  So, he replaced his blog with forum software so that readers could help each other.

However, the number of visitors to his site was not large enough to make the forums particularly lively or helpful.  So, he continued personally answering a lot of people’s GMAT questions in the forums.  He took this a step further — he wanted to wow his visitors with quick responses, so he made sure each posted question in his forums received a response within 1 hour.  It turned out that a lot of prospective MBA students, however, lived in Asia, so he hired a contractor to call him whenever a prospective MBA student posted a question in his forums.  This often required Eric to jump out of bed in the middle of the night to answer forum questions.  But, a year into following this exhausting routine, Eric found that he had built up enough traffic in the forums, and other people were now responding to questions before he could even reach his computer.

Screen Shot 2013-06-20 at 7.50.50 PM

The MBA community started clamoring for more, so the Beat the GMAT team decided to transform their forums into a full-fledged social networking site.  Since the team wasn’t technical, they outsourced the development of their site to what it is today.  Beat the GMAT bootstrapped its way to $1M+ in annual revenue with just 4 full-time employees and was acquired by Hobson’s in 2012.

Founders Naval Ravikant and Babak Nivi had already been successful entrepreneurs by the time they started AngelList, a social networking site for angel investments.  They had the resources to build out a huge site for AngelList, and they initially did, but they quickly found they had overbuilt and did not have users.  So, they took a step back and dumped  everything.  They started again using a mailing list and Wufoo forms to hack together a community of entrepreneurs and angel investors.  They asked both sides to fill out forms with information about their companies and investments.  They manually brokered introductions between relevant entrepreneurs and investors.

Screen Shot 2013-06-20 at 7.52.00 PM

Only once they started getting interactions going did they decide to build out the product that we see as AngelList today.  “We always do it manually…until we know how it works and then we automate it,” explained Naval.  Today, over 1000 startups have been funded on AngelList, and the company is rumored to be raising a first round of funding at $150M valuation.

The truth is — traction matters.  And, if you’re a non-technical founder with just an idea, it’s probably tough to find a technical co-founder.  Having traction on that idea will make it a world easier to find technical talent.

P.S. But, I know that figuring out how to get traction isn’t easy.  So, I’m organizing a conference called Hustle Con (July 9 in Mountain View) to teach new entrepreneurs on how to get customers first.  We’ll be talking about topics such as how to go from 0 to $5M in revenue and how to build an audience before you have a product.  We have a great full line up of speakers including Scott Cook (founder of Intuit), Gagan Biyani (co-founder of Udemy), Jess Lee (CEO of Polyvore), and Arjun Arora (CEO of Retargeter) who will share how they acquired customers.  And, I’m giving away one free ticket on this blog.  All you have to do is tweet why you want a @hustlecon ticket before June 27th, and the best tweet will win.  For those who don’t win, get 25% off with this discount code: “andrew-hustler”

Written by Andrew Chen

June 25th, 2013 at 10:45 am

Posted in Uncategorized

How to grow your app revenue with DuPont analysis (Guest post)

without comments

About the author: Kenton wrote this fantastic piece about analyzing in-app revenue, drawing from his work experience at both Zynga, where he runs their mobile poker product, and before that, Google. You can follow him at @kivestu and his blog here. -Andrew


When I worked at Google, Eric Schmidt used to say “Revenue solves all known problems.” He was right.

And if you’re monetizing a mobile app today, there is a good chance that in-app purchases (IAP) are a critical component of your monetization (if not the sole pillar).* Yet we don’t have great tools for understanding the mechanics of revenue models driven by IAP. Financial analysts who wrestle with similar problems can shed some light.

Financial analysts often use a technique called DuPont Analysis – named after the famous chemical company that created it – to understand what components of a business are driving financial returns.

The DuPont Analysis equation looks like this:


This equation states that if you take a company’s profit margin, asset turnover and financial leverage, multiply them together, you’ll get Return on Equity (ROE) – a measure of how much profit a company generates per the amount invested into the company. It’s an insightful way to quickly understand what is the driving force behind returns (and also known to be one of Buffet’s favorite metrics).

The key insight from DuPont analysis is the principle of decomposing a common metric into the components that drive it.

To successfully monetize via IAP you need a deeper understanding of revenue drivers – top line revenue or even revenue / daily user (aka ARPU) is not enough. A more sophisticated understanding starts with the Transactions Payers Revenue (TPR**) equation:


Rev/DAU (aka ARPU) is a quick measure of how much revenue you’ll make for each user you have on a given day – it’s an overall indicator of your ability to monetize your users.

Payers/DAU measures how many of your users on any given day actually pay – meaning on that particular day X people actually transacted within your app.

Rev/txn measures how much each transaction was worth – this is particularly important for developers that have a large range of price points available in their app (as many games do, for example). Note: If you only sell a single item this metric will be a constant equal to the price of that item.

Txns/payer measures the number of transaction you got for each payer you had in the app in any given day (eg how many transactions did the avg. payer complete.)

Let’s run through a mechanical example. Let’s say two different mobile apps have a $0.10 rev/DAU. On the surface, it might seem like these apps are similar:


But if you dig a little deeper and collect the other key metrics we’ll need for TPR analysis, differences will manifest themselves:


And if you run the calculations, your picture now looks like this:


So what?

The chart above is critical to understand if you’re focusing on improving monetization because it tells you where your leverage points are. For example, if you’re the CEO of product B and you tell your VC that the critical way you’re going to grow revenue is by converting more payers, your VC ought to call BS. Why? Because 5% of daily users paying is already pretty damn good. You might be able to get 5.1% or even 5.5%, but that won’t move the bottom line much. A 10% improvement on an already stellar number, would equate to an extra $10 / day (10% improvement on payers means 5 incremental payers, each doing 5 txns / day, giving you $10+ bucks a day).

However, if you instead focused on increasing the revenue per transaction, you might find there is significant upside. After all, people shell out in excess of $0.99 for a single Coke in most places around the world, so it seems like your revenue per transaction has head room to grow. Maybe by highlighting volume discounts (or some other product tweak) you could get revenue per transaction up to $0.60. It’s not quite a Coke yet, but hey, its an improvement. That would be pretty great though – and it’d net you a $50 bottom line improvement, not bad!

What next?

The TPR equation, while helpful, is just the first step. Any thorough understanding of IAP revenue will require peeling back another layer of the data. For example, is payers / DAU being driven by active payers or new payer conversion? Or what about lapsed payers returning to the app? What about “red herrings”? In some cases, an increasing rev/DAU metric might actually point to long term problems acquiring and monetizing new payer (this can happen when new DAU starts declining, new payer conversion dips and rev/DAU looks healthy because committed, elder users of the app are pushing it up). More on these more nuanced layers in a followup post.

*The most recent Distimo data suggests that revenue from IAP no accounts for 70%+ of app store total revenues, up from ~50% in Jan. 2012.
**TPR stands for Transactions Payers Revenue. I didn’t want to be narcissistic and name if after myself and TPR sounds official, akin to those TPS reports (albeit hopefully more valuable).

Written by Andrew Chen

June 17th, 2013 at 11:00 am

Posted in Uncategorized

New college grads: Don’t sell your time for a living

without comments


If there’s one thing I could tell every graduating student, this is what I’d say:

Jobs suck. At least the traditional version of a job, in which you do something you sorta hate, from 9-5p, and are paid for your time to just grit your teeth and do it. Let’s call this the “sell your time” version of a personal business model: You sell your time to an employer, and they pay you for that time.

Turns out this personal business model sucks.

Even The Onion agrees, in their article Find The Thing You’re Most Passionate About, Then Do It On Nights And Weekends For The Rest Of Your Life.

There’s so much conflict stemming from the fact that this is the predominant mode of work in our society. All the hand-wringing about work/life balance, finding what you love, kids versus work, etc. – an important source of these anxieties come from the fact that a “sell your time” model of work means you’ve set your personal time (and goals) in direct conflict with the time you have to sell for work.

Stop selling your time
There’s a better way – though it might not be the easiest way. The key is to find a way to stop selling your time, and to find another business model instead. And the important aspect of this personal business model is that you’ll be able to make money even if you are sleeping.

1) Learn to make something. Anything.
First and foremost, I think it’s important to learn to make something. Anything. It could be an app, blog, table, YouTube channel, video tutorial, or anything else. Then study the people who have become successful enough to support themselves in this craft, and study them, copy them, stalk them, and meet them.

It always shocks me when people don’t really know how to make anything. Or haven’t ever tried. It’s something we’ve all done as kids – drawings, crafts, etc. – but somehow a very large number of professional workers find themselves in a state where they only know how to repackage other peoples’ work rather than doing anything themselves. Weird.

2) Create a feedback loop with your audience/customers
Remember that the end goal isn’t to make art, it’s to get out of selling your time for a living. So even while you’re learning to make stuff, you’ll want to learn how to make stuff that people actually want. This means you need to create a feedback loop between you and your customers, whoever they may be. This means you’ll want to constantly show people your work, no matter how bad it is. You’ll want to try and build an audience, or a customer base. Again, this is a skill in itself and may take years to figure out.

It’ll also be an opportunity to find small wins in what you do- whether that’s improvements in craftsmanship, or from finding an audience for your work. This kind of positive feedback will keep you going.

3) It’ll take years to become competent
It’s been discussed endlessly in books like Malcolm Gladwell’s Outliers, but it takes years of solid practice to be any good at anything. And then 10,000 hours (roughly 10 years) to become a world-class expert.

But even before you sink years into something, you’ll get frustrated much earlier on because you’ll think that you suck at it. There’s a great quote from Ira Glass (of This American Life) about the difficulty of getting good at anything, starting as a beginner:

“What nobody tells people who are beginners — and I really wish someone had told this to me . . . is that all of us who do creative work, we get into it because we have good taste. But there is this gap. For the first couple years you make stuff, and it’s just not that good. It’s trying to be good, it has potential, but it’s not.

But your taste, the thing that got you into the game, is still killer. And your taste is why your work disappoints you. A lot of people never get past this phase. They quit. Most people I know who do interesting, creative work went through years of this. We know our work doesn’t have this special thing that we want it to have. We all go through this. And if you are just starting out or you are still in this phase, you gotta know it’s normal and the most important thing you can do is do a lot of work. Put yourself on a deadline so that every week you will finish one story.

It is only by going through a volume of work that you will close that gap, and your work will be as good as your ambitions. And I took longer to figure out how to do this than anyone I’ve ever met. It’s gonna take awhile. It’s normal to take awhile. You’ve just gotta fight your way through.”

The period where your taste outpaces your ability to produce it is a hard one. You know your goals but don’t quite know how to fulfill them. That’s why it’s easier to be a film critic rather than a film director :)

The point of all of this isn’t to do it alone. In fact, you’ll find that it’s rare you can do something substantial by yourself. Instead, the above feedback loop most usually involves teams of people, at least once the basic groundwork has been done.

Technology startups are a perfect example of this- it exemplifies the process of getting a bunch of smart people together to learn and make something valuable for the world. It’s a remarkable thing to get the experience of creating something from scratch, and seeing it through to its success in the market. But even if startups aren’t right for you, and  you choose to write books for a living, in a success case you’ll work with editors, research assistants, other writers, etc.

On a final note, if this essay reminds you that your job sucks, just watch this clip from Office Space where they beat up a fax machine.

PS. I’m training Growth Hackers. Email me.

Written by Andrew Chen

June 3rd, 2013 at 11:00 am

Posted in Uncategorized

Does your product suck? Stop adding new features and “zoom in” instead

without comments


[Adapted from an answer I wrote on Quora, and thought I’d share it on my blog too.]

Adding a lot more features won’t save your product
Everyone’s worked on a product it’s failing despite a ton of work behind it. It’s not for lack of great ideas, or a lack of bright minds working long and hard on the product. In the startup world, often this comes because after a new product is launched, there’s a Trough of Sorrow where features are often added to try to spark traction. After a few months of this, and a few shifts in direction, it’s easy to get a Frankenstein product that tries to do too much.

At this point, adding new features won’t help– what’s broken is at the core of your product, not out on the edges. Adding more to edges won’t do anything, because most of your users aren’t even getting there.

Eric Ries has a wonderful term for what to do here, which is to consider a “zoom in pivot.” He talks about it in his book Lean Startup, as a kind of pivot you can do if your product isn’t gaining traction.

The idea of the zoom in pivot is:

A single feature in a product becomes the whole product, highlighting the value of “focus” and “minimum viable product,” delivered quickly and efficiently.

The question is, how do you pick the feature you’re going to zoom into? And how do you validate that it can work as a standalone product? And how do you execute the pivot itself and what metrics can you look at?

Picking the new product
The actual process of picking the new product is the same as picking any new product for a startup. Ultimately it still has to go after a huge market, it has to be differentiated against competitors, and have a distribution model. You have to be passionate about it. Etc, etc. All the standard strategy issues apply, and I’ll leave this as an exercise to the reader.

In terms of tactics though, the big thing from a metrics standpoint is to try and figure out what’s actually getting enough usage to actually execute the “zoom in” pivot. After all, if you zoom into a smaller featureset that isn’t being used currently, that’s obviously much risker than noticing that out of 10 features, 1 or 2 are getting all the usage, so then you dump everything else.

Based on developing a product strategy, and looking at current usage metrics, you can develop a hypothesis for what a smaller product might look like. You can also create some goals you want to hit as far as the metrics are concerned- obviously the usage of the zoomed in feature should be much higher, but by how much? And the usage of the secondary features should become zero or minimal- are you OK with that? The next step is to test it.

Iteration and testing
It should be easy to test a “zoom in” pivot- just default the navigation and the description of the product to focus on what you’re zooming into. You can even test a few ideas simultaneously if you want to.

Here are a few high-impact places to test:

  • Changing all the landing page where new/unregistered users arrive to reflect the new positioning
  • Taking users directly into the functionality after they sign in or sign up, so that you are defaulting to that usage
  • Using modal lightboxes or other highly prominent UI to channel users into the zoomed in featureset
  • At the end of the typical workflow of the user, to take them to the feature again

The above suggestions focus on making the zoomed in feature more prominent, but you can also make the other features more secondary. You can do the following:

  • Burying other features into submenus like “Extras” or “Goodies”
  • Removing other features from global navigation UI
  • Rewriting headlines to de-emphasize unneeded features, or removing text about them from landing pages, bulleted lists, etc.

The combination of all of the above – either by making the main feature more prominent, or the burying the secondary features – should help the goal. You can A/B test these, primarily focusing on new users, to see what the effect looks like.

From a metrics standpoint, I think as a baseline you’d want the zoomed in feature to increase significantly in usage, and for the secondary features to go to zero or nearly so. You also want to make sure some of the aggregate stats around frequency of use, time on site, content shared, etc. to be stable depending on what you care about.

Choosing a feature
After this iteration process, picking the zoomed in feature should be easy. You may have to go through an A/B testing process to smooth the transition from the old featureset to the minimalist one, but over some period of time you should be able to make the metrics move in the direction you want.

If it turns out the metrics are stubborn and some important metrics go down, then that’s much more problematic. It might turn out that the zoomed in feature you picked is somehow not right enough. Or maybe the userbase you’ve amassed isn’t right for the pivot. Or maybe you need to develop the featureset a bit more, in the direction you’ve pivoted, to get to the right product.

For all of these, the Plan B might be to either accept the new featureset and deal with the reduced numbers, hoping to fix them later. Or alternatively, the Plan B might be to pick a new featureset or continue iterating on the zoomed in featureset, until it works. That’s all gray area.

Written by Andrew Chen

May 16th, 2013 at 11:32 am

Posted in Uncategorized

Linkedin, Facebook, Google, Twitter, eBay, YouTube, Wikipedia, Amazon, Hotmail, Blogger, Apple: How they used to look

without comments

This is an oldie but a goodie I had to share again- here are some screenshots of early web products, some close to their inception, some a couple years later.

If you have some other screenshots of early products, be sure to tweet me at @andrewchen and I’ll try to put them up.


Written by Andrew Chen

May 13th, 2013 at 10:00 am

Posted in Uncategorized

The critical metrics for each stage of your SaaS business (Guest post by Lars Lofgren of KISSmetrics)

without comments

[My friend Lars is a product marketer at KISSmetrics and loves helping SaaS businesses understand how their business is growing. He writes regularly for the KISSmetrics blog and his personal marketing blog. He wrote the following post about SaaS products and the metrics you use to evaluate their success level. Lots of great information in there. You can follow Lars at @LarsLofgren -Andrew]


How healthy is your SaaS business?

We’re bombarded with KPIs and an endless series of metrics to tell us how we’re doing.

But instead of using data to measure our progress, it’s much more likely that we get lost and start focusing on metrics that are easy to track but don’t mean anything.

For a SaaS business, there are a few core metrics that need your undivided attention. And the priority of these metrics shift as you grow. If you’ve only had paying customers for 2 months, it doesn’t make much sense to track lifetime value. But later on, lifetime value is essential.

In this post, I’m going to break down the essential metrics for each stage of a SaaS business.

What this framework will give you:

  • By focusing on a few key metrics, you’ll also be focusing on the core problems you need to solve to get your business to the next level.

  • Data doesn’t do you any good unless you act on it. Each of these metrics clearly tells you how you’re doing. Right away, you’ll know where you need to spend your time.

  • Each stage has two metrics that balance each other. This keeps you from over-optimizing one metric and unintentionally harming the long-term health of your business.

Let’s jump in.

Before Product/Market Fit
You’ve just made the decision to start your business and you’ve got plans for world domination.

But before you can start building your empire, you need to make sure you have the right product for the right market.

For most new products, there’s usually a disconnect at the beginning and customers don’t quite want what you have. Either you need to go after a different target market or you need to change your product to fit their needs. When you get this match, we call it product/market fit.

You’re probably in this stage if:

  • You’ve just started.

  • You don’t know who your ideal customer is.

  • People are testing your product for the first time.

This is the first major hurdle you’ll need to overcome. But how do we measure our progress when we don’t have any data? You don’t even have any paying customers at this point and if you do, it’s not many. At this point, running a bunch of A/B tests won’t help you test your business model.

Instead, you’ll rely heavily on qualitative feedback and one critical survey question.

Primary Goals:

  • Validate core business assumptions by talking to people in your target market. If people ask you for your product before you even try to sell them, you’re going in the right direction.

  • Survey users and have at least 40% say that they‘d be very disappointed if they had to stop using your product.

Metric #1: Qualitative Feedback
Yes… this isn’t technically a metric. But it’s too early for data anyway so you’ll need to make the most of what you can get: feedback.

Right now, you really only have one goal: build the right product for the right market. And the fastest way to do this is to start talking to your customers.

If you have any users at this point, jump on Skype and get a deep understanding of their main problems. Ask them to show you how they currently solve the problem you’re going after. Then show them what you’ve been working on to see if they get excited about it. Usually, you’ll want to follow this format for the interview:

  1. Basic demographic questions to get a better sense for who you’re talking to.

  2. Deep questions about the current problem.

  3. Present your solution for feedback (don’t sell it, just get feedback).

You’ll want to do 10-20 of these customer interviews.

If you don’t have any users at this point, go and talk to people that you think would want to use your product. This is a great way to start testing different target markets efficiently. It’s a lot easier to schedule 10 more Skype meetings than it is to rebuild or rebrand your product.

When you want to start scaling feedback (especially as you move into the later stages of your business), use Qualaroo surveys, SurveyMonkey, feedback forms, and usability tests like But when you’re just starting, talk to people in your target market one-on-one. The insights will always be much better.

At KISSmetrics, we still do customer interviews each and every time we make a major change to our product. Adding a new feature? Go talk to customers. Revamping an old feature? Let’s talk to our customers that use it the most. Starting a new project like our Google Analytics app? Find a group of Google Analytics users to talk to. We do it every single time.

Metric #2: Measuring Product/Market Fit
There’s just one little problem with all this customer feedback though.

It’s super difficult to measure objectively. Are people REALLY interested in our product or are we only focusing on the positive feedback while downplaying the negative feedback?

Luckily, there is a survey question that will help you quantify whether or not you’ve reached product/market fit. Full credit goes to Sean Ellis for this question. Here it is:

How would you feel if you could no longer use [product]?

  1. Very disappointed

  2. Somewhat disappointed

  3. Not disappointed (it isn’t really that useful)

  4. N/A – I no longer use [product]

Send this to people that have used your product at least twice, experienced your core product offering, and used it in the last two weeks. The goal is to get at least 40% of your users to say “very disappointed.”

If you don’t meet the 40% benchmark, you may need to reposition your product or pivot entirely. If you do hit it, time to move on to the next stage.

More Resources
To dive into more detail on what you’ll need to make it through this stage, read through these posts:

Beginning to Scale
So you’ve found product/market fit.

You’ve got revenue coming in and a growing customer base. Now it’s time to build a business.

Up until this point, you didn’t really need to track much. Outside of basic user signups and revenue, there wasn’t anything to track. Now that you got the right product for the right market, there are two metrics that will keep you headed in the right direction.

You’re probably in this stage if:

  • You’ve found at least one way to acquire customers consistently.

  • Many of your customers stay subscribed and want to keep paying you.

  • Your monthly revenue is starting to grow.

Primary Goals:

  • Consistently grow MRR while controlling churn.

  • Get monthly churn to 1-2%. If it’s above 5%, ignore everything else until you lower it.

Metric #1: Monthly Recurring Revenue (MRR)
For a SaaS business, monthly recurring revenue is a much more valuable metric to track than traditional revenue. It’s the total revenue you received during the month that came from recurring subscriptions.

The health of a SaaS business heavily depends on recurring revenue. It can take months to regain the cost of acquiring a customer and the real profits come from increasing that subscription revenue. One-time windfalls just aren’t that valuable to us. By tracking monthly recurring revenue, we can see exactly how our business is doing month-to-month.

Unfortunately, tracking MRR can get tricky. There’s several use-cases that your tracking systems will need to be able to handle:

  • Having annual plans on top of your regular monthly plans complicates things a bit. The annual revenue actually needs to get divided between each month of the subscription, not just the month when the customer is billed.

  • Upgrades and downgrades get tedious to track. If a customer moves from a $10/month plan to a $50/month plan, you’ll need to add an extra $40/month to your MRR.

  • You’ll need to remove revenue when it churns with a cancellation.

Speaking of churn…

Metric #2: Churn
Growing MRR is one side of the coin at this stage. The other side is churn. If you can’t keep customers subscribed, it won’t be long before your MRR won’t budge and your business will stall.

The thing is, churn can be a devious metric. At the beginning, a monthly churn rate of 10% doesn’t seem so bad. If you have 100 customers, 10 of them left. Not that big a deal right? It’s pretty easy to get 10 more customers. But what happens when you have 10,000 customers? Now 1,000 of them left in a single month. Even the best marketing machines have a hard time keeping up with something like that.

Your churn rate starts out innocent and easy to handle. But it can quickly get out of control if you’re not keeping a close eye on it. In order to build a strong foundation that will help your company grow over the long-term, you absolutely, without a doubt, NEED to get control of your churn rate.

So what’s a good churn?

It always varies by industry. But in general, it’s critical that you get your monthly churn under 5% and your goal should be 1-2%. Later on, you can experiment with upsells and cross-sells to get negative churn.

Sooner or later, you’re going to hit a wall.

The main channel you’ve been using to acquire customers will start to slow down and you’ll hit diminishing returns. If you want to keep growing each month, you’ll need to find new sources of growth.

You might start testing affiliate programs, new ad networks, PR, business development, referral programs, new types of content marketing, conferences, event marketing, or whatever type of marketing happens to be hot at the moment. You’ve got LOTS of options to choose from. Some of them will be a great fit for your market, others will fail completely.

You’re probably in this stage if:

  • Growth is beginning to slow for the first time.

  • Continuing to improve your main channel is getting a lot harder.

  • You’ve successfully controlled your churn.

So as you start to experiment with new channels of growth, you need to focus heavily on two metrics. These metrics will keep your experiments in check and make sure you scale profitable channels.

Primary Goals:

  • Keep your cost per acquisition to one third of your lifetime value.

  • Get each customer to profitability within 12 months.

Metric #1: Lifetime Value (LTV)
How much revenue do you earn in total from a customer before they leave your business? For a SaaS business, it’s absolutely critical to track lifetime value. When you factor in acquisition, support, and product costs, it can take a SaaS businesses 6-12 months to turn a profit on a customer.

To make sure customers stay long enough to keep your business healthy, we use lifetime value (some people abbreviate it as CLV or LCV).

By now, you’ll have had customers long enough so that you can actually figure out your LTV. Use the formula here to get started. When you have more resources, you might also want to include second-order revenue in your LTV calculation.

Metric #2: Cost Per Acquisition
As we begin to experiment with new channels to keep growing, cost per acquisition keeps us in check. It’s the total cost it takes to acquire a customer from a particular source.

For the average CPA of your business, you can total up your entire marketing and sales expenses over a month then average that out over the total customers you acquired. But we need to take it a step further and segment CPA by acquisition channel. This tells us whether or not customers from new channels are worth the effort.

When you’re experimenting with new channels, it’ll usually be pretty obvious if the math won’t work out. Bad channels tend to be BAD channels. So keep experimenting until you find the ones that work.

Not only will CPA help you evaluate new channels for growth, it’ll help you figure out how far to push your main channels. How much can you actually spend to acquire a customer on AdWords or Facebook? How many writers can you hire to put together content? By keeping an eye on these metrics, you’ll know how far is too far.

A popular rule of thumb is to keep CPA to one third of your LTV. And a customer should become profitable within 12 months.

More Resources
Once you get past product/market fit, use these posts to help you work through all the details:

A Quick Overview
For each stage of your SaaS business, track these metrics:

  1. Before Product/Market Fit: Customer feedback and the product/market fit question

  2. Beginning to Scale: Monthly recurring revenue and churn

  3. Expansion: Lifetime value and cost per acquisition

Keep in mind that each stage is not completely exclusive. Let’s say that I’ve found product/market fit and I’m starting to scale. If I’m using AdWords to acquire my customers, I’ll definitely want to keep an eye on my cost per acquisition. But at this point, I’m still trying to get a handle on my churn for the first time. I don’t REALLY know how long these customers are going to stick around. So I’ll check my CPA to make sure it’s somewhat reasonable (if the total revenue from a 12-month subscriber doesn’t cover it, you have a problem). Otherwise, I’ll spend most of my time improving MRR and churn.

What about funnels? What about engagement metrics, ARPU, active users, number of visits to signup, and everything else? By all means, track the other metrics you need. But the above metrics are the bare minimum. Move mountains to track them before worrying about the rest. There’s little reason to track a random engagement metric if you don’t know what your MRR or churn is.

What have I missed? I’d love to know how you track your own SaaS business.

Written by Andrew Chen

May 6th, 2013 at 11:30 am

Posted in Uncategorized

The death of RSS in a single graph

without comments

Google Trend graph for “rss” – bad news.

I recently wrote a blog post about moving all my RSS readers to email subscriptions, and I immediately got 30+ negative comments on it. Obviously it struck a cord. I still believe what I said, and here’s some more data and reasoning to back it up:

RSS has been dying for years
First off, the image above is the Google Trends search on “rss” over the last few years. That tells you how many people are searching for RSS on Google. To me, that’s the best indication that as a consumer-facing technology, there’s been waning interest for years. Does any blog want to bet on that as a long-term trend? Combine that with the imminent shutdown of Google Reader, and you can guess that a lot of folks using RSS readers will move to non-consumption rather than switching to an alternative. Yes, there will always be a vocal minority that loves feed readers, ultimately RSS will be more like QR codes or Segways than a mainstream technology.

Ultimately, my bet is that RSS will stick around but more as a way for content services to talk to each other – you’ll see random blogs appear in places like Flipboard or Zite automatically – but the idea that people will see the little orange RSS button and click on it is a lost cause. (Oh, and searches for “google reader” don’t fare well either)

RSS doesn’t have a reply function
Interactivity between a writer and their audience is is one of the most rewarding aspects of maintaining a blog. RSS was meant to be a different way to present content, and doesn’t have identity or interactivity baked in. One of the best aspects of email subscriptions (and Twitter) is that you can actually see who’s taken interest in your work. You can even reach out to them and start a friendly conversation. Some of the most important relationships in my career have been made over email and Twitter.

As I switch over to emphasize email, my hope is that I can increase the level of interactivity with my audience. The way its set up now, if you hit reply to any email post, write a quick note, it’ll go directly into my inbox unfiltered. And better yet, we might even have an intelligent conversation!

Moving off RSS will lead to better content
Feedback loops let you iterate on what kinds of content resonate with your audience. Writers need feedback loops to improve their writing – everytime a new essay is emailed to my readers, I get a ton of feedback. I know exactly who and how many folks have unsubscribed. I can reply to ask them why, by writing an email. I also know how many new people have subscribed, and often look at their email domains to figure out if they’re a corporate, a startup, a VC, etc. This kind of detail helps me write better content and get to know my audience. All good stuff. And obviously RSS is just about content, and doesn’t have this kind of feedback built in.

Consumers are moving to “integrated” readers
Related to the negative trend in RSS interest, consumers have adopting other platforms instead. RSS readers were invented in a different era. Blogger, TypePad, and WordPress were created in an era where we thought of blog networks as a bunch of standalone websites, decentralized, like the internet. But it turns out that’s not as easy to use as it could be. Turns out consumers love it when they can follow, view feeds, and create content, all on the same site. This is the core of the feed-oriented homepages of  Twitter, Instagram, or Tumblr – the integrated reader has won out.

Email subscribers are 2x more active than RSS readers
The other thing I’ve noticed is that email subscribers are just stickier and more active. From my own personal data from my blog, I know that although I theoretically have 5x more RSS subscribers than email, from a traffic standpoint, the mass of RSS subscribers don’t make up for their numbers. On a per-email subscriber basis, I get about 2x the activity rate from people clicking links from RSS as compared to email.

So when it comes to the very practical question: When a blog is designed to prompt users to subscribe for future content, what should you push for? RSS or email? The answer is easy, go with email. In otherwards, in order for the numbers to work out, I’d need an RSS prompt to convert at 2x as email to get the same activity level. Given that the market size and interest in RSS is decreasing over time, and a small vocal minority uses an RSS reader, I think it’s pretty obvious where you want to go there.

Until RSS is redesigned (ha!), I repeat: RSS, I quit you. And if you have a blog, you should be thinking about this too.

Written by Andrew Chen

April 29th, 2013 at 9:45 am

Posted in Uncategorized

Featured essays from 2011-2013: Facebook, Growth Hacking, Mobile, and more.

without comments


I’ve recently tried to recommit myself to blogging :) and as part of that, I pulled together my recent set of essays and redid the Featured Essays section of this blog. If you missed anything, check them out below- they are a collection of what I’ve written over the last 18 months or so. In the coming months, I hope to continue writing more about mobile, especially the nascent field of mobile marketing. Thanks for reading.

Oh, and if you are reading this from an RSS feed, please subscribe to email instead. I explain why here: RSS I quit you.


Product/Market Fit



Industry and Investing

Written by Andrew Chen

April 23rd, 2013 at 10:00 am

Posted in Uncategorized