Product-Market Fit

Marc Andreessen popularized the term as "being in a good market with a product that can satisfy that market." His description is famously felt rather than calculated: usage grows as fast as you can add servers, money piles up, you are hiring sales and support as fast as you can. Before PMF you are pushing the product onto the market; after PMF the market pulls it out of your hands.

The two halves both matter. "Product" is whether the thing works and is wanted; "market" is whether enough people want it badly enough to sustain a business. A great product in a tiny or indifferent market does not have fit, and a mediocre product in a desperate market can have it. PMF is the moment those two sides lock together for a specific, nameable segment — not "everyone," but a concrete who.

The most cited instrument is the Sean Ellis test: survey users who have experienced the product's core value and ask, "How would you feel if you could no longer use this product?" If at least 40% answer "very disappointed," you likely have fit. The 40% line is an empirical benchmark from Ellis's work, not a law — treat it as a strong signal, and read the segment of "very disappointed" users to learn who your real market is.

Retention is the harder, truer signal. Plot a retention curve (the share of a cohort still active over time) and look for it to flatten rather than decay to zero — a flat tail means a stable group keeps coming back, which is the quantitative shape of demand. Pair it with organic pull: word-of-mouth referrals, unprompted sign-ups, and a shrinking cost to acquire each new user. When growth comes from users bringing users, the market is doing your selling for you.

PMF is not a launch, a funding round, or a spike in sign-ups. Top-of-funnel traffic and trial starts measure interest, not fit — a great landing page can manufacture a sign-up surge that retention then erases within two weeks. If new users do not come back, you have demand for the pitch, not the product.

It is also not permanent, not binary, and not one number. Fit is segment-specific (you can have it for startups and lack it for enterprise), it erodes as markets and competitors move, and it can regress when you expand into adjacent segments that do not share the original job. Revenue alone can mislead too: heavy discounting or a few large, at-risk logos can prop up MRR while underlying retention quietly rots. The honest read of PMF triangulates the disappointment survey, the retention curve, and where the organic pull is actually coming from.

Each PMF signal lives in a different place by default: survey responses in a feedback tool, cohort retention in product analytics, referrals and sign-ups in web analytics, and revenue in billing. To read fit honestly you have to join them — which segment is "very disappointed," whether that same segment retains, and whether its revenue is healthy rather than discount-propped.

A product operating system built on a shared spine joins those records by design. Product analytics and web analytics share the same identifiers as the revenue data on the spine, feedback and insights attach to the customer record, and Customer 360 shows one segment's usage, requests, and revenue in a single view — so you can cut retention and the disappointment signal by the segment that matters instead of stitching exports together. That makes the PMF read sharper; it does not manufacture the fit itself.