Andrew Chen Archives

Notifications strategy

Every startup has to think about their notifications strategy – it’s not enough to blast out e-mail whenever anyone on your site does anything, or to send out weekly newsletters. Notifications are an important part of how you manage your userbase’s lifecycle, and is a key part of any retention effort your business has.

I’ve often thought that online advertising folks are pure-plays for user acquisition. And that (traditional) games people are pure-plays for engagement. The question is, what’s the pure-play for retention? I’ve come to believe that businesses that focus on managing churn – be it telecom/wireless, catalog, financial services – are the guys who are the furthest ahead in understanding these concepts.

Learning from the catalog marketing world
I’ve recently been doing research in the retail and catalog marketing part of the world, where companies have thrown billions into understand when it’s a good idea to mail someone something and when it’s a bad idea. Unlike the online world, there’s generally a cost associated with sending out a “notification” and there’s usually a high LTV attached to each person, so the stakes are much higher.

Traditionally, companies in the catalog marketing world use a segmentation system called RFM:

• Recency
• Frequency
• Monetization

Basically every user in your database is graded from 1 to 5 across each variable from a quintiles basis, where 555 is the best.

The idea here is that the higher the numbers, the more likely the recipient is to respond to an offer. Thus, 555 customers are more likely to mail back something than 455, and Recency also plays a higher factor in response than the Frequency of that customer. (Which is why it’s RFM and not FRM)

But how do you know when someone would have responded anyway?
An interesting problem then occurs because what if all of your 555 customers would have responded anyway? Then you’re sending them catalogs but it’s pointless. I want to point at a great paper on “uplift marketing” that answers exactly this question. I attached it as a Scribd embed below, but if you’re reading this in an RSS reader, you’ll have to click through to this blog.

Anyway, the authors from a company called Stochasic Solutions divide the world into 4 groups:

1. Persuadables
2. Sure Things
3. Lost Causes
4. Sleeping Dogs

The idea is that Persuadables are the ones who are affected by your marketing (and notifications), whereas Sure Things would have embraced your products anyway.

Then the Lost Causes are never going to respond, no matter how much marketing you send their way.

And finally, an interesting group are the Sleeping Dogs, who are likely to use your product but as soon as you’re reminding them (especially too often), then they are likely to quit your service. Ouch. (Gym memberships anyone??)

The point is, the Persuadables are the ONLY group who can improve your business. Everyone either costs you dollars by expending marketing budget that didn’t need to get spent, or would quit your service upon remembering they ought to cancel.

Don’t assume that more notifications is better
I’ll let you guys read the document yourself, but I think the overall point is that first off, segmenting your audience based on their responsiveness is a smart idea. You’re able to figure out key variables that drive their participation, which leads to some nice strategies on the retention side. On the other hand, don’t forget that the Persuadables are the true audience that can be retained via notification. Everyone else is negative, and if you’re overly aggressive, it’s easy to turn those Persuadables into Sleeping Dogs.

The full document below:

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