I recently blogged about the importance of conducting a cohort analysis to track user retention. A good cohort analysis helps you understand how many customers you keep and how many you lose. So now what?
I'm surprised how little pro-active messaging/communication most Internet companies do. And if they do send me an email, it tends to be a generic weekly promotional email that they send to all users. One thing that I learned at half.com is the importance of lifecycle messaging -- in which you deliver different messages to different users based on where they are in their lifecycle. Some examples:
- The average fiction book is read within two weeks of purchase. So if you purchased a John Grisham book for $8.75 on Half.com, chances are that you will finish it within 14 days. We decided to implement an auto-email that was sent 17 days after purchase that said "Want your $8.75 back, click here to list your Grisham book for sale". We found that the open (and conversion) rate of that email was amazing -- and it greatly added to our ability to "turn" the same book multiple times.
- We noticed on our cohort analysis that if a first-time buyer did not make a second-purchase within 6 weeks of their initial purchase, there was a 80%+ chance that they would not return -- but that if a user made two purchases within that period of time, there was a 70+% chance they would return again. So we created an auto-email with a personalized, limited-time coupon that was sent after six weeks. Again, we experienced strong open/conversion rates.
- We noticed that if a buyer left negative feedback on a transaction it had a definite impact on their repeat purchase behavior -- so we created a customized email campaign towards those users.
- Like all e-commerce sites, Half.com had many "abandonded shopping carts" -- where people would put items in the cart and not check out. We created an auto-email campaign that would merchandise the items left in the cart -- and send it out two hours after the cart was abandoned. A very successful campaign.
And while these are examples of lifecycle messaging by email -- you can do the same thing on the site as well. Many sites still display the same homepage for first-time and repeat users. The leading e-commerce sites added personalized home-pages long ago -- so they can deliver different messages to new users (education, introduction, tutorials) and to repeat users (personalized recommendations, promotions). If you'd like to see a good example, go to eBay as an existing user -- then delete your eBay cookies from your browser and return.
Conducting a cohort analysis is just the first step towards increasing user retention and activity levels. Once you have your cohort analysis, you then should look at ways to systematically reduce churn and increase user satisfaction - by finding the key "touchpoints" which can dramatically change the course of a user's experience. This is not rocket science, but despite the fact that Jupiter Research data indicates that lifecycle marketing campaigns generate as much as nine times greater results, few marketers are taking advantage of this strategy to deliver the right message at the right time.
My co-founder at Half.com, Sunny Balijapalli, has recently started a new company (Zoomin.com - which was rated the #1 photo sharing and printing service in India by PC World) that First Round Capital has funded. Not surprisingly, one of his first priorities was to put together a lifecycle messaging program. With his permission, I've attached a copy of his program overview here.
A small digression. The current issue of Wired Magazine discusses the "spacing effect" for human learning -- which recognizes that if a person is reminded of a fact at certain specified intervals, they are far more likely to remember the fact later. (See chart below from Wired). I think the metaphor can be extended to consumer marketing as well -- if you communicate with a user during certain specific intervals, they are far more likely to return to the site later. They key is to figure out what those intervals are -- and what "triggers" you should use to drive the communication.