The Roadmap to Product/Market Fit

TL;DR: Low retention kills companies that think growth can compensate. Fix retention first. Growth on a leaky bucket isn’t growth, it’s waste — and most founders learn this the expensive way, after burning a year of runway acquiring users who churned before they could ever pay back their CAC.

What it means

Cohen’s central argument is the kind of thing that should be tattooed on every growth marketer’s wrist: “Some people believe that low retention is fine because you can make up for it with growth. That’s not true.” Retention compounds. Acquisition doesn’t. A product with high retention and modest acquisition will reliably win against the inverse over any sufficiently long timeframe. You cannot outrun a leaky bucket; you can only delay the moment it becomes obvious.

This is the practical foundation underneath growth-as-compass — growth rate only matters if retained users are the ones growing. And simplicity-as-strategy helps here for a structural reason: simple products tend to have better retention because users actually understand the core value proposition, which means they remember why they came back.

The argument

Retention before growth. Every growth dollar spent on a product with poor retention is wasted, and the waste compounds because the cohorts you lose this month aren’t around to pay back the CAC next month. Fix the bucket first. This echoes Duolingo’s CURR-first strategy (duolingo-growth) and the hook-model's emphasis on building habits before scaling triggers. The pattern is so consistent across companies that “retention before growth” is one of the very few startup rules that actually generalize.

Product processes produce repeatable wins. “The key to driving growth is to create product processes that produce repeatable wins. Not one win, not two wins. But wins that can take you through decades.” This is process-power applied at the level of growth — organizational capabilities for finding and shipping retention improvements, compounded over years. The companies that figure this out look like they have unstoppable growth engines because they essentially do.

The connection to PMF

This piece is one of the cleanest companion reads to only-thing-that-matters and superhuman-pmf-engine. Andreessen tells you what PMF feels like. Vohra tells you how to measure it. Cohen tells you the diagnostic that distinguishes real PMF from the kind that looks like PMF on a monthly chart but evaporates in a cohort view: retention. If your retention curves don’t flatten out into a horizontal asymptote, you don’t have PMF — you have a leaky funnel that the law of large numbers is hiding from you.