For two straight quarters, a two-sided marketplace's "activation rate" — the share of new sellers who completed their profile within seven days — climbed steadily. Revenue from new sellers did not move at all in the same period. Something didn't add up.
The metric looked healthy
Activation had been defined as "profile completed," which seemed reasonable: a completed profile was a known prerequisite for a seller's first sale. The growth team had spent real effort nudging sellers toward profile completion, and the nudges were clearly working — the number kept going up.
What the team found when they dug in
A cohort analysis split sellers by how they completed their profile: through the guided onboarding flow, versus through a one-click "auto-fill from social profile" shortcut the team had added to boost completion rates. The shortcut group completed profiles at a far higher rate — and listed their first item at less than half the rate of the guided group.
The shortcut was inflating "activation" without producing sellers who were actually ready to sell. The metric had quietly become a measure of how easy it was to click a button, not a measure of whether someone was on track to succeed.
What they measured instead
The team redefined activation as "listed at least one item within seven days" — a behavior directly tied to the outcome they cared about, rather than a step assumed to lead to it. The new metric dropped immediately when redefined, which was the point: it now reflected reality instead of a vanity number.
The result
With the honest metric in place, the team redirected onboarding effort toward the first listing instead of profile completion. Within two months, the corrected activation metric rose 21%, and — for the first time in two quarters — revenue from new sellers moved with it.
The takeaway
- A metric that's easy to move isn't the same as a metric that matters; check what's driving the number, not just its direction.
- Proxy metrics ("completed profile" standing in for "ready to sell") need to be re-validated whenever the flow that feeds them changes.
- A metric going down after a redefinition isn't always bad news — sometimes it's the first honest number you've had.