A click in the first hour is not adoption

Product groups in Calhoun and elsewhere still treat the first-day graph as a verdict. Someone ships a filter, a share sheet, or a new empty state. The graph jumps. Slack fills with congratulations. Two weeks later the same graph is a polite slope, and nobody can say whether the slope is seasonal, accidental, or a sign that the feature never belonged in the main path.

Adoption, as we use the word in a feature adoption reading, is a return. A person who found the control, used it for a real task, and came back without being prompted by an in-app banner. That definition is stricter than a unique click. It is also kinder: it stops teams from arguing about vanity spikes they cannot reproduce.

When we sit with a local usage ledger, we mark three stamps: first encounter, first successful task, and second unaided return. If the ledger cannot support those stamps, we say so in the brief instead of inventing a proxy. Proxies are how “opened the menu” becomes “the team loves the redesign.”

Keep the announcement-week numbers. Just do not let them stand in for the third week. The third week is usually short enough to read by hand, which is the point of keeping records on machines you already own.