Journal · 12 Mar 2026
Why activation rate is a poor proxy for feature adoption
Activation answers a different question from uptake. Activation asks whether a new account reached a first meaningful state in the product as a whole. Feature adoption asks whether a specific surface, shown to people who already have an account, was taken up. Collapsing the two is how a banner “success” survives a quarter without a second session on that banner’s destination.
In the Asoke room we keep a cheap test: if your activation chart would look identical whether or not the new feature existed, you are not measuring the feature. You are measuring onboarding weather.
A worked split
Take a wallet app that added bill-scan. Activation might be “first successful top-up within seven days of install.” Bill-scan adoption needs its own denominator: accounts that were shown the scan entry in a working Thai locale, with camera permission already granted or clearly prompted. Plenty of activated users will never be exposed. Plenty of exposed users will never complete a scan that posts to the ledger.
When teams report “activation held steady after bill-scan launched,” they have said nothing about scan. The feature could be unused. It could also be carrying a small, loyal group that activation will never reveal because those people activated months ago.
What to put on the page instead
Write three rates with explicit windows: exposure among eligible accounts, trial among the exposed, habit among those who tried. If you only have room for one sentence in a council memo, use habit/exposure against a pre-agreed floor. That sentence is harder to write and much harder to game.
Students sometimes ask whether activation can be a leading indicator for a brand-new app with one job. Even then, name the job. “Activated” that quietly means “opened settings” will waste the next review. Feature adoption measurement starts when you can point at the surface.