Andrew Chen Archives

(this is a cross-post of the guest blog I wrote over at the Startonomics blog yesterday)

So you want to make some money, eh?

Every startup gets to the point where they decide to bring in some ad revenue, and it's worthwhile to consider exactly how much money will be generated through advertising. So in the parlance of the ad industry, your "ad inventory" is the set of impressions that you have to sell to people. Let's talk about modeling that goes beyond the simple reports into thinking about what drivers influence your CPMs and impression counts.

This blog post will go over some of the key milestones that any ad-supported startup will have to face. In particular, we'll discuss 5 stages:

• Stage 1: Just getting started
• Stage 2: Layering in new sections
• Stage 3: Thinking about growth
• Stage 4: Adding in multiple ad networks
• Stage 5: Building up brand sales (rep, vertical, direct, or otherwise)

Let's get started with some basic metrics around CPMs and pageviews…

Stage 1: Just getting started
When you first get started, it makes sense to sign up to a self-serve, automated CPC ad network like Google AdSense. At the end of the month, you'll get a report for how much money you've made and what your CPM is, which will look something like this:

For many folks, just looking at this kind of number is enough, although it doesn't provide much detail beyond the core revenue numbers.

As I mentioned in my previous blog on 5 factors that determine your CPM rates, there are some rules of thumb for what kinds of CPMs you should expect on your site. Here are some very rough numbers:

• Social sites (forums/chat/etc) without direct ad sales teams: <$0.25 CPM
• Largely international sites: <$0.50 CPM
• Medium-sized sites that use banner ad networks: <$1 CPM
• Reference sites in a specific category: >$5 CPM or sometimes
  much higher, depending on category – we ran into home improvement
  reference sites that did $20 CPMs

Much of that is determined by the frequency of the pageviews, how specific it is, how much international advertising there is, etc.

Stage 2: Layering in new sections
For the sites out there that grow beyond AdSense, you need to have more granularity for thinking about and optimizing your ad inventory. You have to break down your site into the pieces that influence the performance the most. In general, that tends to be factors like:

• site section (or function), like Homepage, Profile, Media, Inbox, etc
• ad unit size, like 300×250 or 468×60 (full IAB ad units listed here)
• type
• targeting
• etc.

The point is, you should have a good sense of where your ad units and impressions are coming from, and how revenues overlay that. And the more impressions you get, the more granular you'll want to track your advertising.

Let's take a social networking site like MySpace as the example here. They might break down their inventory, at an early stage, into something like below:

Section	Impressions	Avg CPM	Revenue
homepage	100,000	$1.00	$100.00
forum	500,000	$0.50	$250.00
profile	200,000	$0.50	$100.00
message	100,000	$1.00	$100.00
Total	900,000		$550.00

(disclaimer: note that these numbers are for purely illustrative purposes only – your mileage may vary)

Once you have an inventory breakdown like above, you can start rolling up your aggregate revenues, impressions, and CPMs.

Stage 3: Thinking about growth
The next phase is scaling up your site and thinking about where you're likely to add to your ad inventory. In general, static areas like the homepage aren't necessarily going to grow very much. But other portions, like forums or profile or other social areas are likely to experience network effects that drive non-linear growth.

As a result, you'll end up adding traffic disproportionately to the social areas of the site, and your aggregate CPMs are likely to be brought lower by the higher-traffic, lower-value communication parts of your site. In the cases where the higher pageviews are also brought on by higher engagement, you might also expect the CPMs to decrease as you hit frequency caps imposed by the various ad networks upon your userbase.

Section	Impressions	Avg CPM	Revenue
homepage	200,000	$1.00	$200.00
forum	2,000,000	$0.40	$800.00
profile	1,000,000	$0.40	$400.00
message	500,000	$0.75	$375.00
Total	3,700,000		$1,775.00

(again, disclaimer: note that these numbers are for purely illustrative purposes only – your mileage may vary)

The point is that even though the traffic grew over 4x, the revenue only grew 3x. That's to be expected based on the reasons discussed above, primarily driven by lower-value inventory tending to grow like weeds.

Stage 4: Adding in multiple ad networks
It's also often the case that having many different ad networks can dramatically affect your CPM rates as well. Because of frequency capping issues and the availability of high-value campaigns for each ad network, a lot of times high-CPM opportunties are spread lightly across many sites and you don't get much of a benefit for having deep engagement. One way to solve that is by serving the ads of many different networks and optimizing between them to get the best prices at each point.

There's a great post by Mike Nolet of MikeOnAds on this subject, where he talks about how his previous company, RightMedia, handles the ad-network-to-ad-network optimization of revenue.

The key image from Mike's post is below:

What you're seeing here is that as user frequency increases, the monetization between the 3 networks (small/medium/large) is different. Ideally you'd start out on the small network, then go to the medium one, and then go to the large one in order to monetize for that CPM.

My personal experience (while at Revenue Science) working with the MySpace ad ops team led by the fantastically smart David Dipilato was that they juggled more than a dozen very large networks in order to maximize their revenues. This level of expertise in ad ops is a very hard skill to train and hire for, by the way.

Stage 5: Building up brand sales (rep, vertical, direct, or otherwise)
Once your site hits a certain level of ad inventory – and that number's certainly over 50M/month and maybe even several multiples of that – it's likely that you'd want to go after brand dollars. The reason is simple: remnant ad networks bring in CPMs in the dollars, whereas brand can take you into the tens of dollars. The problem is that the process of doing brand ad sales is very foreign and specialized, especially to folks who have thus far focused on technology, automation, and scalability. Brand sales is like enterprise sales – it's slow, custom, and not technology-intensive at all. To win you focus on excellence in sales and marketing operations.

Here are some of the common options for getting brand dollars:

Old school ad networks
Generally most folks get started through their existing ad networks, which start to rep their inventory as part of a brand category. This can happen if you're a tech blog and they sell you as part of their "Technology channel." You'll end up seeing these dollars mixed into your overall ad network CPMs, and hopefully they boost your aggregate numbers into something good!

Vertical ad networks
Another way that's getting much more common now is to get into some sort of vertical ad network like Glam.com or Jumpstart or whatever is appropriate for your site. (Glam = women and Jumpstart = cars, btw). These guys are far more sophisticated and specialized, and commonly move a ton of ad inventory as part of their business models.

Direct sales teams
Similarly, many companies go out and hire their own sales teams, or engage ad reps who already have those relationships. This is a hugely expensive process, but that's how the big money comes in.

Once you get your brand sales going, the site you're running will need a dedicated team to manage the active campaigns. The different ads coming through the system need to be carefully and constantly checked to make sure they are delivering in a timely manner, that they are running in the right sections, and that whatever performance metrics the advertiser is using goes well.

On the revenue side, you can no longer make little tables of inventory and use that to predict revenue. It's common that a very small % of ad impressions generates a sizeable % of the revenue. I've heard that some sites sell 5-10% of their impressions as brand ads which account for 30-40% of the revenue, for example. The right way to think of this is that your ad inventory ends up being a pyramid with the best stuff on the top and the worst stuff forming the base. 

So ultimately, you track the ad campaigns that the ad sales team is selling and your ad ops team is running exactly like an enterprise system. You have a sales pipeline with different campaigns going through based on RFPs your team is getting, and eventually once those flight, the revenue comes into your company.

In summary
There are many complexities in squeezing out the most dollars from your ad inventory – it's something that even big media companies struggle to master as they grow. Much of the process of doing this smoothly involves making sure you don't try to solve hard problems until they really become core to your business – for example, don't hire a big ad sales team before you have enough pageviews. Similarly, it's important to understand the implications of how your product ultimately impacts the monetization – social products, for various reasons, tend to monetize poorly and that needs to be built into your business model from Day One.

For more essays on advertising, social network monetization, and more, check out my directory of essays here.

PS. Get new updates/analysis on tech and startups

I write a high-quality, weekly newsletter covering what's happening in Silicon Valley, focused on startups, marketing, and mobile.

Leave this field empty if you're human: