Instrument time-to-activation

Done when

Report the median minutes from signup to the activation event across the last 20 signups, and how many of the 20 could be measured. (Proof: a count.)

If this is a change to your product

What must become true

Every new account records two timestamps, its signup and its first activation event, so time to activation can be read per account without joining logs.

Where it usually lives

The user record and the place the activation event is emitted: the signup handler and the event emitter.

The trap

Choosing the activation event after seeing who stayed biases it. Fix the event first, the one named in your funnel step, then measure, and keep per-account values because a median hides the long tail.

In topics

• Activation and onboarding