Finding product-market fit — and how to measure it

Product-market fit means a specific group of people uses the product repeatedly, would be genuinely disrupted if it disappeared, and tells others about it without being asked. Every part of that matters: a specific group, repeated use, real disruption, unprompted referral.

The reason it is worth defining precisely is that the alternative — a product people find interesting — produces most of the same early signals. Sign-ups, positive feedback and press all occur without fit. Only sustained behaviour distinguishes them.

Take everyone who first used the product in a given week and measure what fraction return in each subsequent week. Plot it. Every product declines at first. The question is whether the curve flattens.

A curve that flattens at any positive level means a durable group has found real value, and growth compounds because each cohort adds to a stable base. A curve that continues toward zero means every customer acquired is eventually lost, and growth requires permanently increasing spend. This distinction matters more than the absolute retention number.

The often-cited survey question — how disappointed would you be if this product no longer existed — is a reasonable secondary measure. A commonly used benchmark is that above forty percent answering 'very disappointed' suggests fit, though the threshold is a rule of thumb rather than a law.

Behavioural signals are more reliable than stated ones. Usage that grows within an account without sales involvement. Customers building workarounds to use the product for things it was not designed for. Renewals that happen without a conversation. Support tickets that are requests for more rather than complaints about basics.

Total registered users, funding raised, press coverage, waitlist size, social following and conference interest are all compatible with having no fit whatsoever. Each is generated by novelty or marketing rather than by value delivered.

The most misleading of these is a large sign-up number from a launch. Launch traffic is curiosity. What matters is the fraction of it still present eight weeks later, and that number is usually a small fraction of what the founders hoped.

Narrow rather than broaden. The most reliable route to fit is to find the small segment that retains best and rebuild the product specifically for them, accepting that this makes it worse for everyone else. Products that serve one group intensely can expand later; products designed for everyone rarely become essential to anyone.

This means examining cohort data by segment rather than in aggregate. It is common to find that overall retention of fifteen percent conceals one industry retaining at forty and everyone else at five. That forty-percent group is the business.

Scaling acquisition before the curve flattens is the expensive one. Growth spend on a leaking product converts capital into a temporarily larger number, and the underlying problem is unchanged when the money runs out.

The other is reading aggregate growth as fit. Total revenue can rise for a year while every individual cohort performs worse than the last, and by the time it shows in the aggregate the company has hired against it.

Garage to IPO expresses fit through churn and product quality. Investment in R&D lowers churn and raises revenue per user, which changes the entire trajectory of a run — the same marketing budget produces a durable base rather than a treadmill.

The most common way new players lose a run is the mistake above: spending heavily on acquisition while churn is high, watching users climb and cash fall, and running out before the base stabilises.