We Redefined Our Activation Metric and Retention Numbers Changed Overnight

The old activation definition was 'completed signup' — created an account, verified email, done. By that definition, activation rate was a healthy 78%. Retention, measured against that activation number, looked consistently disappointing, and the team spent two quarters trying to fix a retention 'problem' that turned out to be a measurement problem instead.

Signup completion predicts nothing

We pulled every user from the prior six months and ran a correlation between a long list of early behaviors and 90-day retention. Completing signup correlated with almost nothing — unsurprising, since it's a nearly universal step. The behavior that actually predicted retention with real strength was connecting at least one data source within the first session. Users who did that retained at roughly four times the rate of users who signed up but never connected anything.

An activation metric isn't supposed to measure how many people got through the door. It's supposed to measure who did the one thing that predicts whether they stay.

What changed once we redefined it

• Activation rate dropped from a reported 78% to 34% overnight, which was alarming until the team understood it was now measuring something meaningful instead of something trivial.
• The onboarding flow got redesigned around getting a user to connect a data source in session one, rather than around completing the generic signup form faster.
• 'Retention is broken' stopped being the framing — it became 'activation is too low,' which pointed the team at a fixable, upstream problem instead of a vague downstream symptom.
• Activation rate under the new definition became the metric leadership actually tracked weekly, because it moved in response to onboarding changes in a way the old one never did.

Retention itself never changed the day we redefined activation — the underlying user behavior was identical. What changed was that we stopped measuring the wrong upstream number and started being able to see, and act on, the thing that actually predicted whether someone stayed.