The Activation Metric Hunt
Find the activation metric that predicts retention
We need our activation metric - the early behavior that predicts retention: [WHAT THE PRODUCT DOES, OUR CURRENT ONBOARDING STEPS, ANY COHORT OR USAGE DATA WE HAVE, GUESSES THE TEAM HAS]. Hunt it: the candidate behaviors worth testing (specific, countable, early), how to check correlation with retention using the data we have, the difference between a metric that predicts and one that causes, and the trap of picking something easy to move that does not matter. End with the one candidate to instrument first and the experiment that validates it.
Inputs needed
- ■Product description, onboarding steps, available data, team hypotheses
What good output looks like
Candidates are specific and countable; correlation check described for their actual data; prediction-vs-causation addressed; one pick with a validation experiment
The stream (0)
Reading the room…