Your signups are fine. Your second sessions are not.

"Honestly, the funnel is working. We just need more top of funnel." The founder said it with the ad dashboard open, and the ad dashboard agreed with him. Cost per signup was down for the third straight month. Signups were at an all-time high. The confetti animation fired on every new account, the product congratulating itself for the one step that was never in doubt.

Revenue, meanwhile, was flat, and paid spend was not. The distance between those two lines lived in a place the dashboard did not have a chart for: the stretch between signing up and getting value, where most of those record signups arrived, looked around, and never came back. Second-session rate was 31 percent. Nobody knew that number, because nobody had ever asked for it. The funnel had a beginning everyone measured and an end everyone measured, and the middle was running on hope.

Acquisition buys attempts, not customers. The product has a few sessions to convert the attempt, and most teams cannot say what conversion even means there.

Every product has some early behavior that separates the users who stick from the users who evaporate. The folklore version is famous, the social network that found "seven friends in ten days," but the folklore obscures the method, which is empirical and available to any team with event data.

Take your candidate milestones: imported data, invited a teammate, created a second project, hit the core action three times. For each, split signups into those who did it within the first week and those who did not, and lay the retention curves side by side. One milestone will usually show a separation that is large, early, and stable across cohorts. That is your activation event, defined by evidence rather than by whichever step the onboarding tour happens to end on.

Then comes the step that separates analysis from decoration: validate that the milestone is a lever and not just a marker. Correlation is doing a lot of work in that curve, keen users do keen things, so you test it. Push a cohort toward the milestone, through onboarding changes or nudges, and watch whether retention follows. If it moves, you have found a control surface for the business. If it does not, you have found a horoscope, and the search continues.

Once the activation event is real, wire the operation around it. Activation rate per weekly cohort, on the wall, next to signups. Time-to-value as a distribution with a target, if the median user takes four days to reach the moment that makes them stay, every one of those days is a tax on every dollar of acquisition. Onboarding rebuilt as the shortest honest path to the milestone, which usually means deleting steps someone was proud of.

And the acquisition math changes currency. Cost per activated user, by channel, replaces cost per signup, and the league table rarely survives the translation. We ran this analysis inside a B2B product where the cheapest channel per signup turned out to be nearly the most expensive per activated user; the budget reallocation that followed came out of arithmetic, not opinion, and it was the least controversial spending decision that team had made all year.

Cost per signup ranks your channels by how well they generate confetti. Cost per activated user ranks them by how well they generate customers.

The fair warning: activation metrics rot. Products change, audiences shift, and the milestone that predicted retention in year one may be theater by year three. The teams that stay honest re-run the curves on a schedule, the same way they rerun load tests.

Before approving more top of funnel, ask for one number: of last month's signups, what share reached your best guess at the aha moment within seven days. If the room cannot produce the number, or cannot agree on the guess, the growth problem is not upstream. It is in the unmeasured middle.

TensorLabs builds this instrumentation and the analysis on top of it, activation discovery, validation experiments, the cohort plumbing that keeps it honest. If your signups and your revenue have stopped agreeing with each other, write to us and describe the gap. The middle of a funnel is usually the cheapest place in the company to find money.