Marc Andreessen defined product-market fit as "being in a good market with a product that can satisfy that market." It is a useful frame. It is also incomplete enough that the companies most confidently claiming PMF are often the ones furthest from it.
The signal founders confuse with PMF: early adopter enthusiasm
Every product that has ever failed had early adopters who loved it. Early adopters are not the market. They are the segment of the market with the highest pain, the lowest switching cost, and the highest tolerance for rough edges. Their enthusiasm signals that the problem is real. It does not signal that the product can reach the customers after them.
The distinction matters because early adopter behavior and mainstream market behavior are structurally different. Early adopters will find the product, tolerate friction, and evangelize without being asked. Mainstream customers need the product to find them, require lower friction, and will not evangelize unless the experience exceeds their expectations consistently. A product with strong early adopter retention can have zero PMF in the actual market it needs to penetrate to be a real business.
The PMF signal that is actually reliable: retention cohorts
The most reliable PMF signal is not the one founders most frequently cite — customer acquisition rate, NPS, or qualitative enthusiasm. It is the retention curve of acquisition cohorts over time. If consecutive cohorts of customers are retaining at similar rates and the retention curve flattens (rather than declining to zero), the product has demonstrated that it creates enough value for customers to keep using it. That is the foundation. Everything above that foundation is growth. Retention is PMF.
The founders who confuse acquisition with retention are the ones who scale into failure. They build distribution for a product that is not retaining, which means they are buying customers who leave. The acquisition number grows. The underlying business does not.
The PMF trap: segment fit mistaken for market fit
Segment fit is when the product works very well for a specific, identifiable subset of the potential market. It feels like PMF because the retention is strong, the referrals are real, and the customers are enthusiastic. The trap is building a growth strategy based on segment fit before understanding whether the economics of the segment support the business model the company has built.
The questions that distinguish segment fit from market fit: Is there enough of this specific customer type to build the business we have modeled? Can we acquire more of them at a cost that makes unit economics work? If the answer to either is no, the product has segment fit — which is valuable — but the founder needs to decide whether to rebuild the product for a larger segment or resize the business model for the segment they have.
What PMF actually feels like
Sean Ellis's definition remains the most operationally useful: if you asked your users "how would you feel if you could no longer use this product?" and more than 40% said "very disappointed," you have PMF. The threshold is not the point — the question is. It forces the founder to measure intensity of product dependency rather than surface-level satisfaction.
Andreessen's description of PMF from his 2007 essay is still the clearest qualitative framing: "The product is spreading from person to person only by word-of-mouth. You are not able to hire engineers fast enough. The press is writing about you. Customers are finding you. Investors are calling you." Not all of those will be true at once, but the direction of all of them should be right. When you have PMF, the market pulls. When you do not, everything is push.
The PMF mistake that kills good companies
The most expensive PMF mistake is scaling before you have it. The second most expensive is refusing to acknowledge when you had it in the wrong segment and need to find it somewhere else. Both mistakes come from the same root: treating PMF as a binary declaration ("we have it" / "we don't have it") rather than as a continuous signal that needs to be monitored, calibrated, and updated as the market changes. PMF is not a milestone you cross. It is a condition you maintain. And markets change often enough that the condition needs to be re-examined more frequently than most founders do it.