Product-market fit
The concept every founder chases and few define
Description
In June 2007, Marc Andreessen published a blog post titled The Only Thing That Matters. The single most important factor in a startup's success, he argued, is neither the team, nor the market size, nor the cleverness of the product — it is whether the product has achieved what he called product-market fit. Product-market fit was the condition of being in a good market with a product that can satisfy that market. A startup without it, no matter how well-funded and well-staffed, was doomed. A startup with it, even poorly executed, would be pulled forward by what users wanted. The rest of the operational work — hiring, fundraising, marketing — was secondary.
The post reshaped how founders and investors talked about early-stage companies. Before it, there was no widely-used term for the condition Andreessen named. The available terms — traction, market validation, product resonance — were either narrower or less precise. After the post, product-market fit became the standard shorthand. Eighteen years later, the concept is ubiquitous. Its definition, however, remains contested, its measurement is inconsistent, and whether a given startup has achieved it is usually answered through an uncomfortable mix of intuition and ex post facto rationalization.
What product-market fit actually is, how to tell whether you have it, and what to do when you do not are questions that have generated substantial literature without a settled answer. The concept is real — something happens when a startup crosses the threshold from struggling for every user to being pulled forward by demand — but the threshold is not a crisp boundary. It is a fuzzy transition zone whose features differ across markets and business models. Understanding it requires working through what Andreessen said, how it has been operationalized, the heuristics founders use, and the framework's limits.
● The question we're asking: what is product-market fit, how do you know when you have it, and why does the concept resist precise definition?
● What we'll see: the original formulation, the refinements that followed, the measurement heuristics that have emerged, and the limits of the framework.
Table of contents
01The original formulation
Andreessen's 2007 definition was deliberately qualitative. A startup has product-market fit when it is in a good market with a product that can satisfy that market. The phrasing foregrounded the market over the product. A mediocre product in an unusually good market could achieve fit. A spectacular product in a bad market generally could not. The asymmetry was important because it pushed against the founder instinct to focus on product quality as the primary success driver. The market, Andreessen argued, was the more important input. The best product in a dying industry would not save the startup. A barely-adequate product in an exploding market could ride the market up.
Andreessen's signs of fit were phenomenological. When you have it, customers are buying the product as fast as you can make it, usage is growing faster than you can add servers, money piles up in the bank, you are hiring sales and support as fast as you can. When you do not have it, customers are not getting value, word of mouth is not spreading, press reviews are blah, the sales cycle takes too long, deals never close. The list described what fit looked like once unambiguously present. It did not give a formal test for the ambiguous middle cases.
02The refinements
The decade after Andreessen's post produced refinements that tried to operationalize the concept. Sean Ellis, an early growth marketer at Dropbox, LogMeIn, and other successful startups, proposed the most widely-used survey test. Ask current users how they would feel if they could no longer use the product. Options: very disappointed, somewhat disappointed, not disappointed. Ellis argued, based on observations across dozens of startups, that at least forty percent needed to answer very disappointed to have achieved fit. Below the threshold, retention and word-of-mouth were typically insufficient. Above it, they were typically sufficient.
The Ellis test has become standard because it is simple to run and produces a defensible threshold. The forty-percent number is not rigorous in the sense of being derived from a controlled study; it was a pattern Ellis observed across his consulting. But subsequent practitioners have found it reasonably predictive. Teams whose users cluster at very-disappointed typically go on to scale. Teams clustering at not-disappointed typically do not. The test is not infallible — wording can bias responses — but it is the closest the industry has come to a reliable quantitative indicator.
03The measurement heuristics
In practice, founders rely on a collection of heuristics rather than a single metric. Organic growth, where users arrive without paid acquisition, is one of the clearest signals — products with fit produce word-of-mouth that pushes up the funnel without marketing. Retention, measured as the percentage of users still active at 30, 60, or 90 days, is another. Net promoter score above fifty is another rough proxy.
For B2B startups, the signals differ. The sales cycle shortens as the buyer quickly understands why the product solves their problem. Deal sizes become more consistent. Customer success conversations focus on expanding usage rather than justifying continued purchase. Reference calls produce enthusiastic responses. A B2B startup at fit has a sales motion that feels like pulling an oar through loose water, not battering through ice on every deal.
04The limits of the framework
The first limit is that the framework is fundamentally post-hoc. Founders know they have achieved it when they have already achieved it. The framework does not tell you how to get there, only how to recognize arrival. This is important because most of the practical work happens before fit, and the framework has little specific guidance for this period beyond the Lean Startup loop. Founders iterating toward fit often describe the experience as disorienting because no single action feels like moving the needle, and the framework cannot tell them whether they are close or far.
The second limit is that fit is not always a binary threshold. Many startups achieve partial fit — product works for a specific segment but not others, or in a specific geography but not elsewhere. The binary language — you have it or you do not — misleads in these cases. Fit often exists in specific cells of the market and is absent in others. The strategic question becomes which cells to focus on and which to abandon, which the binary framing does not help with.
05Why it still matters
Product-market fit matters as a subject because it is the concept that the startup industry uses to distinguish companies that are going to succeed from companies that are not. The distinction is not perfect, but it is the best shorthand the industry has. Founders who understand the concept are better at directing their attention to the thing that actually determines startup success; founders who do not understand it often waste time and capital on the operational improvements that cannot substitute for it. Investors who understand it can read the early signals more accurately; investors who do not often pattern-match on the wrong indicators and invest accordingly.