The situation
Everything upstream is in order. Acquisition works, the message is aimed at a real buyer, and new customers set the product up and get exactly what they were promised. The first month is excellent. Usage then decays on a curve so consistent you could set a calendar by it.
This is the least comfortable of the six, because there is nothing to fix in the marketing and the marketing is where the company has capacity.
What the company usually believes
That retention was a communications problem. Lifecycle email, in product nudges, a customer success hire. Each of these produces a small improvement that decays, because they are addressing attention rather than reason.
Which of the three it is
A downstream leak, and the hardest of the three kinds.
How the other two get ruled out
Not a positioning problem wearing a costume. The buyers had arrived expecting exactly what they got, and said so. The promise and the product agreed.
Not an activation failure. They reached the moment of value and recognised it. The thing worked.
The evidence
The distinction that matters is between value that did not land and value that landed and did not repeat. Look at whether the job the product does recurs at all for that buyer, on their calendar rather than on yours.
A product used in bursts, with nothing to do between bursts, produces this exact curve. No amount of lifecycle email creates a reason that the buyer’s own work does not.
What we do
We write the conclusion the company does not want, in the first two pages, with the cohort curves next to it, because a diagnosis that can only ever conclude that marketing is at fault is not worth paying for.
The work that follows is product shaped rather than message shaped. Find the adjacent job the same person does on a far more regular clock, using the same data and the same access, and earn the habit there.
What changes
The argument inside the company moves. The roadmap conversation and the growth conversation stop being separate conversations, which is usually what actually moves this case, and is worth more than any single feature.
What the company stops doing
Buying attention to compensate for a missing reason. Hiring against retention before knowing whether retention is achievable with the product as it stands.
The lesson
Not every growth problem is a marketing problem. A diagnostic that is not willing to conclude that the product is the constraint will always find a marketing answer, because it is being paid to.