Resurrection and Win-Back
By the end of this lesson, you should be able to: compute the denominator that makes a resurrection number mean something, compare a returning user to a new one honestly, tell a growing pool apart from an improving product, and say what a win-back campaign can and can't take credit for.
The term with no denominator
Every other term in the growth equation gets a rate. Acquisition has cost per signup. Activation has an activation rate. Churn has a churn rate.
Resurrection gets a count. "We brought back 5,000 users last month" is how it's always reported, and on its own that number is unreadable. Five thousand out of what?
Lesson 3 named the state that answers this. Stale users signed up, weren't active last period, and aren't active now. Nobody puts them on a dashboard because they do nothing, and they're the entire pool resurrection is drawn from.
The denominator is everyone who could have come back: the dormant, plus the ones who actually did. Without it you can't tell a good month from a big pool.
Basket's dormant pool, and what comes out of it
| Week | Brand new | Came back | Dormant pool | Resurrection rate |
|---|---|---|---|---|
| Mar 16 | 2,965 | 1,351 | 15,015 | 9.0% |
| Mar 30 | 2,213 | 2,578 | 17,164 | 15.0% |
| Apr 13 | 2,133 | 3,289 | 20,098 | 16.4% |
| Apr 27 | 2,549 | 3,940 | 24,291 | 16.2% |
| May 11 | 2,691 | 4,727 | 28,407 | 16.6% |
| May 18 | 2,774 | 5,154 | 30,659 | 16.8% |
Two things stand out.
The dormant pool is 30,659 people, roughly twice the number who were active that week. For any product past its first year this is normal and almost always unmeasured: the largest group associated with your product is the group doing nothing.
And 5,154 came back against 2,774 who were brand new. Resurrection delivers nearly twice the volume of acquisition. It has no budget line, no team, and no dashboard.