The situation
Acquisition is genuinely working and the dashboard says so. Then the cohort chart bends. People sign up and never reach the point where the product is obviously useful, or they reach it, use the thing for a few weeks, and quietly stop.
Revenue looks flat rather than falling, which is the detail that makes this hard to see from inside. New business refills the bucket at close to the rate it empties. The team is busy, everybody is working hard, and the number will not move.
What the company usually believes
That the answer was more at the top. Traffic is the lever most teams already know how to pull, and pulling it produces visible activity inside a week. The second theory, once the first stopped working, was that onboarding needed rebuilding.
Which of the three it is
A downstream leak.
How the other two get ruled out
Not a positioning problem, on the evidence. The accounts that stayed and the accounts that left had arrived with the same expectation, and that expectation matched what the product does. Where it does not match, this looks identical from the outside, which is exactly why the segmentation comes before the repair.
Not a lack of urgency. People were buying. The difficulty started after the purchase, not before it.
The evidence
Separate the accounts that stayed from the accounts that left and look for what divided them on day one rather than on day thirty. The answer is nearly always one of three, and the three need different repairs.
They arrived expecting something the product does not do, which is a positioning problem wearing a retention costume. They never reached the moment of value, which is a product and onboarding repair. Or the value was real and did not repeat, which is the hardest of the three and the only one that is genuinely about the product.
What we do
We name which of the three it is, with the cohort evidence attached, and then build against that one rather than against all three at once. Choosing wrong here costs a quarter, so the diagnosis is worth more than the speed.
What changes
Acquisition spend stops being poured into the same gap. The work moves from the top of the funnel, where it is visible and comfortable, to the part of the system that is actually failing.
What the company stops doing
Increasing spend against a leak. Rebuilding onboarding on the assumption that the problem is the path rather than who is walking it.
The lesson
Adding traffic to a system that leaks makes the leak more expensive, not smaller. Find the leak first, and be specific about which of the three it is, because the repairs have nothing in common.