Journal
The cost of a miscounted activation
Activation is where app analytics goes to be flattered. It is a single word, it appears on slide three, and it can absorb staff accounts, QA, scheme codes, and the intern’s test login without changing colour. A London marketplace we taught in a four-week Ledger Seat had been buying growth work against a definition that included an internal code used to demonstrate the app on shop floors.
The cost was not a licence fee. It was a year of campaigns aimed at a number that moved when head office ran training. Creative was not “wrong”. The packet was promiscuous. Operations owned the demonstration flow; product owned the dashboard; neither owned the name. That is the usual geometry of a miscount.
We did not run a forensic audit for them. We made them print the event, assign an owner, and put staff packets on the forbidden-sea plate of the Telemetry Atlas. The board meeting that followed was, in their words, short. Growth paused two channels. The previous vendor workshop was not our contract and not our refund.
If your activation includes anyone who can be asked to tap the app as part of their job, you do not have an activation metric. You have a blend. Blends are allowed in research. They are expensive in a budget line. Write the exclusion in the charter before you hire another round of acquisition. The cheaper correction is a sentence, not another dashboard.