Activation is a retention problem in disguise.

Why the teams with the best activation numbers are quietly running retention experiments on day one.

The lazy definition.

Most teams define activation as "the user completed signup and reached the first meaningful action." Depending on the product, that might be sending a first message, placing a first order, or funding a first account. These metrics are easy to measure, which is why they exist. They are also lazy.

The real definition of activation is the one that correlates with week-four retention. Anything else is a vanity number dressed up as a funnel metric. Teams optimising the lazy definition feel productive — numbers move — but the cohort at the other end doesn't retain, and nobody can quite explain why.

The honest question.

The honest question is: "of the people who did X, how many were still here four weeks later?" Run that query for every candidate activation event, rank them by retention lift, and you have your activation metric. It is almost never the one marketing picked.

How to run it.

Start with a list of candidate activation events: everything a new user might do in their first few days. Then, for each candidate event, run the same query. Of the users who did this event in their first week, what share were still active four weeks later? Compare that with users who didn't do it.

Next, test pairs and sequences, not just single events. "Did X" is one question. "Did X, then came back and did Y the next day" is another, and it is often the more predictive one. The return visit tends to carry more signal than the first action, because it shows the product earned a second look.

Then rank the candidates by retention lift and pick the earliest behaviour near the top of the list. That becomes your activation metric. If it changes what onboarding should optimise for, say from "complete the first action fast" to "give people a reason to come back tomorrow", the redesign follows from there: a timely notification, a recap the next morning, something worth returning for on day two.

Treat this as a method, not a result. The behaviour that predicts retention is different for every product, and the only way to find yours is to run the query on your own cohorts.

Three questions to ask.

• 01
  What's the earliest behaviour that predicts week-four retention? Not the first conversion. The first behavioural tell. If you can't answer this, you don't have an activation metric yet.
• 02
  What's the second behaviour? Retention rarely hinges on a single event. Look for pairs or sequences. The one-two behaviour is almost always more predictive than either event alone.
• 03
  Is your onboarding optimising for the first behaviour or the pair? Most onboarding flows are built around the first behaviour. The best ones engineer the second behaviour deliberately — by building reasons to come back, not just reasons to sign up.

Why teams resist.

Because the lazy activation metric is convenient. It moves in response to onboarding tweaks. It gives the team a weekly win. The honest activation metric often doesn't move for weeks — because the retention-predictive behaviour happens on day two or day three, not day zero — and weekly wins are the currency of a growth team.

The teams that push through this feel the honest metric for about a month, then never want to go back. The lazy metric is a mirage. The honest one is a compass.

Where this fits.

Start with the retention query. Pick the cohort behaviour that most predicts week-four retention. Make that your activation metric. Build the onboarding around it. Watch the retention cohort numbers before you watch the activation rate. That is the work.