Activation

For a long time, the health of a product was read off the signup count. Count how many new people arrived this month, compare it to last month, and if the number is bigger, the product is growing. For software that was sold once and installed on a machine, that arithmetic held up fine. The money had already changed hands at the moment of the sale.

Subscriptions broke the arithmetic. Signing up is free, cancelling takes one click, and the revenue only shows up if the person comes back month after month. Someone who creates an account, looks around for forty seconds, and closes the tab still lands in the signup count. They never land in the revenue.

Cheaper signups widened the gap further. When getting in required a card number and a contract, the people who signed up were the people who intended to use the thing. Once an email address was enough, signing up stopped being a decision and became something people do while browsing. In a lot of products today, sixty of every hundred new accounts never get past the first screen. The number being counted stayed the same; what that number meant changed completely.

So planning meetings kept stalling in the same spot. Marketing would report three thousand new signups for the month, the CEO would ask how many people were actually using the product right now, and the answer would be two hundred. Nobody could account for the other two thousand eight hundred. The ads team had brought people in and the product team had shipped features, but the few days between the signup screen and a second visit belonged to no one.

Activation is the name that got attached to those few days. It was not a new discovery so much as a decision to stop treating the gap as nobody's problem, and once someone had to report the number every week, the gap started getting smaller. The metrics branch of the product field is laid out in the order a user passes through it: acquisition covers people arriving for the first time, activation covers what happens right after, and retention covers whether they keep coming back. Once the middle slot had a name, it also got an owner, a way to measure it, and a list of places to intervene. Teams that treat expansion as an experiment discipline — the growth side of the house — use the same word for the same stretch, which is why it shows up in both sets of meetings.

Five terms sit under activation. Four of them name something that happens to the user, and one names the work a team does to make those things happen.

The four concepts fall in the order a new user meets them.

1. Setup moment — the configuration a user has to finish before any value is visible. In a shared document tool that means creating a first document and inviting a colleague. The longer and fiddlier this stretch is, the fewer people reach whatever sits on the other side of it.
2. Aha moment — the point where the user realizes this product is useful to them specifically. It lands through doing rather than through reading marketing copy, which is why it usually shows up as one identifiable action in the event log, such as editing a document alongside a colleague in the same window for the first time.
3. Habit moment — the point where a one-time realization hardens into repeated behavior. Past it, people stop opening the product because they remembered to and start opening it because the work took them there.
4. Time to value (TTV) — how long it took to get from signup to the aha moment. The same activation rate means very different things depending on whether that took thirty minutes or nine days. The first three terms ask where the user got to; TTV asks how long the trip was.

The activity is onboarding: everything a team lays down in advance so new users reach those points. The copy on an empty first screen, the email that goes out the next morning, the guided tour that clicks through the first three steps — all of it is onboarding material. The distinction matters because the other four describe things that happen to a user, while onboarding describes work the team controls. When a review concludes that activation is weak, onboarding is usually what actually gets rebuilt.

All five get rolled into one number inside product metrics: the activation rate. There is no industry formula for it. Each team draws its own line, usually as a bundle of two or three behaviors inside a time window, such as the share of new users who create three documents and invite at least one colleague within seven days. Draw the line too low and people who did nothing but sign up count as activated. Draw it too high and users who are doing fine get filed as failures.

Drawing the line usually goes like this.

1. Split users who stuck around from users who left early, then compare what each group did in week one. If eighty percent of the people still around a month later invited a colleague in the first week, against twelve percent of the people who left, invitations become a candidate behavior.
2. Attach a count and a window to the candidate. One invite or three, seven days or fourteen — the same product can report an activation rate twice as high depending on which you pick.
3. Leave the definition alone for at least a quarter. Move the line often and nobody can say whether this month was better or worse than last month.

"This quarter we're working on activation instead of acquisition. We can pour three thousand people in the top and it keeps leaking out the bottom."

A statement that the quarter's resources are moving from bringing in more people to keeping the ones who already arrived.

"What are we using as the activation bar? One document created feels too loose to me."

The team has not yet agreed on how far a user has to get before we count them as having seen the value.

"Signups are up but the activation rate is flat. I think we should look at the ad creative."

A suspicion that the product didn't get worse — the people arriving did. If the ads pulled in an audience with no use for the product, no amount of product work moves the number.

At Pagedeck this comes up as the second line of the Monday metrics review. Ravi reads out last week's new signups, and Mia immediately reads the share of that same cohort who created a document and invited someone within a few days. The two numbers sit side by side because the first one on its own never tells anybody what to do next. A week where signups doubled and the activated share halved is a worse week than one where nothing moved, and only the pair of numbers shows that.

Last month signups rose eighteen percent while the activation rate fell from thirty-four to twenty-nine. Theo pulled the drop-off points on the screens right after signup and found that half of the new accounts were closing the tab on the field that asks for a team name. Noa moved that field later so it could be filled in any time and replaced the first screen with a single empty document, and the rate was back to thirty-three within two weeks. None of that required new instrumentation, only a habit of reading the two numbers together. Without a word for this stretch, the conversation would have ended at "signups are up, so why isn't revenue." With the slot named, it was obvious who owned the missing five points, and the search narrowed to the screen after signup rather than to the ads.

The most common error is counting a completed signup as activation. It feels natural — they cleared the signup screen, so they're our users now — and the signup timestamp is the easiest thing in the system to count. But activation asks whether someone got value, not whether someone got in. A person who created an account and never opened a document is, from the product's side, someone nothing has happened to yet. Collapse the two and the dashboard looks healthy every month while next month's payments stay flat. Worse, a table that reads three thousand signups and three thousand activations hides where the two thousand eight hundred went. The whole reason for measuring activation separately is to see into the gap between signing up and coming back, so filling that column with the signup number empties it of purpose. A team that does this usually notices a year later, when someone asks where the drop happens and there is no record that can answer.

The second error is assuming a better onboarding flow will lift activation on its own. The flow is the visible, editable part, so it's where hands go first, and polishing the copy genuinely does help sometimes. But weak activation has three different sources. The wrong people are arriving; the setup stretch is so long that users give out before reaching any value; or the value itself is thin. The first two move when the ad targeting and the setup moment change. The third moves only when the product changes. Rework the guided tour without diagnosing which one you have and the screens change every quarter while the number drifts between twenty-nine and thirty percent.

Splitting the data usually settles it. Break the activation rate out by acquisition channel: if one channel sits in the low teens while the rest look normal, the problem is who is being brought in. If every channel is evenly low and users pile up on the same screen, the problem is the setup stretch. Running that split takes an afternoon and saves a quarter of guessing, which is why most teams do it before touching a single screen. The third case is the hardest to face — when setup is short, the traffic is right, and even the users who finish everything don't come back, the thing to fix is the value the product promised, not the instructions for finding it.