Aha moments

A signup is not a user. It is a person who clicked a button and has not yet decided whether your product is worth the next five minutes. Activation is the early action (or set of actions) most strongly correlated with long-term retention.

Facebook's threshold — add 7 friends in 10 days — was an empirical finding from cohort analysis, not a brand story. Slack's early teams that sent a few thousand messages stuck. Dropbox users who put a file in a folder and opened it on another device got the product.

The big idea

"Completed the welcome tour" and "opened the app twice" are easy to instrument and usually weak. The definition must be specific enough to optimize:

Sometimes the aha is emotional ("I get it") and the measurable proxy is mundane ("imported CSV"). Use the proxy you can instrument, then tighten it.

Walking through cohort analysis

You do not invent activation in a brainstorm. You discover it by comparing retained users to churned ones.

1. Pick a retention definition (returned in week 4). See retention.
2. Pull first-week event histories for retained vs churned cohorts.
3. Rank early events by lift: P(retained | did X) / P(retained | did not X).
4. Prefer events users can choose to do, not ones only power users stumble into after already loving the product.
5. Validate with interviews: do people who did X describe a clearer "this works for me" moment?

Suppose 10,000 signups in January. Week-4 retention overall is 18%. Among users who invited ≥1 teammate in the first 3 days, week-4 retention is 41%. Among users who only completed the profile checklist, it is 19%. Inviting a teammate is the stronger candidate.

Watch selection bias. Power users invite teammates because they already like the product. If "invite" is mostly a lagging signal of love, optimizing the invite modal will not create love — it will nag people who have not felt value yet. Sequence: personal value first, then the social action.

Version the definition (definition_version) so when you change "aha" from "invite" to "first shared edit," historical charts do not silently lie.

Common mistakes

• Declaring tour completion the aha. Easy to raise, weakly tied to retention.
• Asking for the social action too early. "Invite your team" before personal value trains people to bounce.
• Never checking that the proxy still predicts retention. Retire it when the link dies.

Check yourself

Week-4 retention is 18% overall, 19% after the profile checklist, 41% after an invite. Which is the activation candidate?

The invite. The checklist barely beats the base rate.

Users who invite teammates retain better. Should you put the invite modal on step 1 of signup?

Not yet. Check for selection bias, and deliver personal value first.

Key takeaways

• Activation is the early behavior that predicts retention — not signup, not a tour.
• Find it by comparing retained and churned first-week histories.
• Social actions often lag love. Sequence personal value before the ask.

Further learning

Next, shrink the wait until that moment in time to value.