Decide explicitly: are you measuring activation at the user level or account level? Blurring the two creates a metric that is directionally interesting but operationally unusable.

Track the full progression: intent → creation → activation → habit. Drop-off between stages reveals where friction actually lives, not just whether the final threshold was cleared.

Define exact events, exact thresholds, exact windows, and documented exclusions. If two analysts get different counts, the metric is still too vague to run experiments against.

SSO edge cases, bot traffic, missing signup events — surface caveats openly. An honest metric with known gaps beats a fake-clean one that hides ambiguity and explodes in reviews.

Activation becomes operational only when it enters the weekly system: cohort review, experiment prioritization, ownership. Until then it remains a strategy artifact, not a management tool.