Onboarding6 min read

The Activation Metric

One measurable event that separates accounts likely to stay from accounts likely to leave.

The problem it solves

Without a defined activation: The moment a new user first gets the value the product promised. event, onboarding work cannot be judged. Teams optimise signups and completions while retention stays flat.

Signup to activation

An illustrative path from signup to an activation event defined as a second teammate posting. The largest single drop sits between account created and first item created, which is where onboarding work has the most room, not at the invite step teams usually focus on.

An activation metric is a single event, or a small combination, that marks the point where an account has understood the product well enough to be worth something. It is not signup. It is not tour completion. It is the behaviour that separates the accounts still present in ninety days from the ones that are not.

You find it by looking backwards. Take accounts from a cohort old enough to have retained or churned, and look for early behaviours that differ sharply between the two groups. The candidate that separates most cleanly, and that a new account can plausibly reach in its first week, is your metric.

1Pick a cohort at least three months old, so the outcome is known.
2Split it by whether the account is still active now.
3List early behaviours in the first seven days and compare the two groups on each.
4Shortlist behaviours that differ sharply and are reachable by a new account.
5Test the shortlist for causation by making one easier and watching whether retention follows.

Step five is the one teams skip, and skipping it is how organisations end up chasing correlations. Engaged users do many things. Some of those things caused the engagement and some were symptoms of it. Pushing users toward a symptom produces the behaviour and not the retention.

Weak activation metric

Completed onboarding.
Logged in twice.
Clicked into three sections.
Viewed the dashboard.

Strong activation metric

Invited a teammate who then accepted and posted.
Created three items in the first week.
Connected a data source that returned real records.
Published something that another person opened.

The right column shares a shape. Each one involves real effort, real data or another person, which is why they are hard to fake and hard to gamify. The left column can be satisfied by a curious visitor who will never return.

Worked example

Hypothetical: reading a cohort

A team looks at accounts from six months ago. Among those still active, most had added a second team member within the first week. Among those gone, most had not. That is a correlation and not yet a finding. They make inviting easier for half of new accounts and watch whether ninety day retention moves. If it does not, the invite was a symptom of a team that was already committed.

Review the metric each year. Products change, audiences change, and an activation definition that predicted retention two years ago can quietly stop predicting anything while still being reported every week.

When it fits

Products with enough volume and history to compare retained and churned cohort: A group of users bucketed by when they joined, tracked over time rather than blended together. credibly.
Teams that need a shared target for onboarding work across design, growth and engineering.
Products where a small number of early behaviours plainly separate outcomes.
Organisations willing to run a test rather than shipping on correlation alone.

When it backfires

When correlation is treated as cause, so the team pushes a symptom and gets the behaviour without the retention.
When the metric is easy to satisfy trivially, which invites interface pressure that produces the event and nothing else.
When one metric is applied to genuinely different segments, hiding that each needs a different path.
When it is set once and never revisited, so the team reports a number that stopped predicting anything.

Products using it

Slack
Workspace value has long been framed around teams exchanging a meaningful volume of messages rather than around account creation.
Dropbox
Early activation was framed around putting a file in a synced folder, which is the behaviour the product depends on.
Canva
Activation is framed around completing and exporting a design rather than around opening the editor.

The psychology under it

The takeaway

Find the early behaviour that separates retained from churned accounts, then test that making it easier actually moves retention.

Finished the teardown? Bank it and the day counts toward your run.