What is product-market fit, and how do you know you have it?
The point at which a product satisfies a real, strongly felt need in a market large enough to build a business on — and the honest answer to how you know is that it is unmistakable when you have it and ambiguous when you do not.
Marc Andreessen's framing, which remains the clearest: before fit, you are pushing — customers are not quite buying, sales cycles are long, press does not care, and the product is not being used much. After fit, you are being pulled — you cannot make it fast enough, servers fall over, money piles up, and you are hiring as quickly as you can.
The signals worth watching:
Retention that flattens. This is the strongest single indicator. Most products lose users continuously; a product with fit shows a cohort retention curve that levels off rather than decaying to zero. A flat tail means some group genuinely needs it.
Organic growth and word of mouth, without paid acquisition doing the work.
Usage frequency matching the intended pattern.
Customers complaining when it breaks, and pursuing you about it.
Sean Ellis's survey question — asking users how they would feel if they could no longer use the product, with the widely cited benchmark that around 40% answering "very disappointed" suggests fit. A useful heuristic, not a law.
What it is not:
Not a launch. Interest at launch is curiosity, and it decays.
Not funding. Investors have been wrong repeatedly.
Not growth alone. Paid growth with poor retention is buying customers, not fit — and it can look excellent for a year.
Not permanent. Markets shift, competitors arrive, and fit can be lost.
Why it is worth being strict about it. The most common and expensive startup error is scaling before fit — hiring salespeople, spending on acquisition and building infrastructure for a product people do not quite want. That converts a survivable problem into an unsurvivable one, because burn rises while the underlying issue remains.
Before fit, the only job is finding it: talk to users constantly, watch behaviour rather than opinions, and narrow the target until someone genuinely cares.
General information, not business advice.