Product-market fit is not a feeling and it is not a launch. It is pull: customers arriving faster than you can explain, retention flattening instead of decaying to zero, and the roadmap being dragged out of you by paying users. Before that, everything is push — and the honest name for the push phase is "before PMF," which is where almost every company spends years.
Read the retention curve, ignore the rest
Take everyone who started using the product in a given month and plot how many are still active each week after. Decay is normal at the start. What matters is whether the curve flattens. A curve that flattens at 30% means a real segment found real value — go find out who they are. A curve that slides to zero means nobody needs it yet, no matter what the signup chart says.
Growth hides this. You can double signups every month while the curve quietly hits zero, and the day the marketing stops, the company stops. Retention first, growth second — always.
Segment before you build
When the curve is weak, the reflex is to add features. The better move is to slice the curve: by customer type, by acquisition channel, by first action taken. In every plateau I have worked through — including two of my own — there was a sub-segment retaining dramatically better than the average. The path to PMF was narrowing to that segment, not widening the product.
Ask the question that hurts
Ask your active users: "How would you feel if you could no longer use this?" If well under half say "very disappointed," you do not have fit yet — and the follow-up answers tell you which users are closest to needing you, and what is missing for them. Survey the payers, not the fans; fans are free and forgiving, payers are neither.
Change one lever per cycle
Pick the segment. Fix the one thing their answers point to. Give it a full cycle — two to four weeks — and re-read the curve. Founders lose the plateau months by changing five things at once and learning nothing from the result. Pre-PMF iteration is a lab; contamination ruins the experiment.
Milestones that mean something
- Ten paying customers you did not personally befriend
- A retention curve that flattens for one named segment
- Customers describing the product to each other in words you did not write
- The first inbound "can I pay you more for X" request
When those four are true, stop iterating and start pouring fuel. Until they are, every dollar spent on growth is renting users the product cannot keep.