When we mapped events to user behavior across all segments, the real problem emerged. It wasn't one problem. It was three.
Finding 1 — Activation was a single metric that meant three different things
The team defined "activation" as one binary event: "completed setup." But the analytics persona ran a query, the reporting persona created a dashboard, and the admin persona configured a data pipeline. All three counted as "activated" in the same bucket. The 14% headline number was an average of three wildly different persona completion rates: the analytics persona sat at 22%, the reporting persona at 11%, and the admin persona at 8%. The product was failing three different groups in three different ways, and the single metric hid every signal.
Finding 2 — The aha moment was not one action, but three connected actions
For every persona, real activation — the moment users understood the product's value — required three connected actions, not one. The analytics persona needed to (1) connect a data source, (2) write a query, and (3) see results rendered as a chart. The single "activated" event only fired when all three were completed, so there was no visibility into where within that sequence users were dropping off. The real bottleneck was invisible because the intermediate steps were never tracked as distinct events.
Finding 3 — Onboarding had 14 friction steps before the first value moment
The complete onboarding flow contained 14 steps — account creation, email verification, workspace setup, data source connection, schema mapping, user invite flow, role assignment, API key generation, sample data import, query builder intro, first query attempt, error handling, dashboard creation, and then finally seeing a result. Users completed an average of 8.2 steps before abandoning. They never reached the value moment because they exhausted their patience before the product delivered anything useful.
