Measure onboarding to first value

An onboarding checklist tells you that users passed screens. It does not tell you that they reached a reason to return. The useful activation question is: what observable event means this customer has received the first complete unit of value?

The economics chapter defines activation as a cohort reaching a first-value event within a window and warns that teams can improve the number by weakening the definition. Read that context at metrics and economics chapter.

Define the event in customer language

Avoid events such as 'completed onboarding,' 'visited dashboard,' or 'created account.' They measure your interface. Prefer a completed customer outcome: reconciled the first statement, shared an approved plan, resolved an exception, published a usable asset, or received a response from a teammate.

Write an event contract with actor, action, object, quality condition, time window, and exclusions. Amplitude's documentation says events record user actions and properties provide context; it also warns that tracking everything can bury the signal (Amplitude event selection). This is vendor documentation about its own analytics product. The event-design principle does not require that vendor.

Choose the window from the workflow

A five-minute consumer task and a two-week enterprise setup should not share an activation window. Use the shortest period in which a prepared target customer can fairly reach value, allowing for workday and approval delays. Report the distribution of time to value, not only whether a threshold was met.

Behavioral-cohort documentation shows how a cohort can be defined by an event count and a window from first use (Amplitude cohort documentation). Again, this is product documentation, not evidence that a particular event predicts retention. You must test that relationship in your data.

Worked hypothetical cohort test

Hypothetical: A team-planning product defines first value as: 'A new workspace owner publishes one plan containing at least three assigned tasks, and another invited teammate views or edits it within seven days.'

In April, 120 qualified workspaces start; 42 reach the event. In May, a guided sample plan is introduced for a randomized half of 140 new workspaces. The team compares event completion, median time to event, week-four return, support contacts, and deletion of sample content.

If the guide raises the event rate but not week-four return, or users publish untouched sample plans, it improved ceremony rather than value. All numbers are fictional; no benchmark claim is implied.

Read the funnel and the cohort together

A funnel locates the step where users leave. Google Analytics documentation describes funnel reports as a way to define steps and examine abandonment between them (Google Analytics funnel reports). A cohort asks whether people who reached value later behaved differently. Neither proves causation by itself.

Compare cohorts by segment, acquisition source, device, plan, and start week when sample size permits. Keep the event definition fixed through the test. If it changes, version it and restate earlier results where possible.

Activation artifact

• Event name: a stable verb-object name
• Customer value: why the event matters to them
• Exact rule: actor, object, properties, and exclusions
• Window: from which start event to what deadline
• Quality guardrail: what prevents empty or automated completion
• Cohort slices: segment and acquisition source
• Downstream check: a later return, repeat outcome, or paid continuation
• Owner: who approves definition changes

Limits

Early cohorts are noisy, identity resolution can merge or split users, and enterprise value may occur offline. Vendor analytics can sample, model, or omit data depending on configuration. Activation is a leading indicator you define and validate; it is not product-market fit, retention, or revenue.