Andy Rachleff didn't set out to create the most referenced concept in startup strategy. As co-founder of Benchmark Capital and later CEO of Wealthfront, Rachleff was teaching a course at Stanford's Graduate School of Business in the mid-2000s when he articulated an observation from two decades of venture investing: the companies that succeed are the ones that find a market with urgent demand and build a product that satisfies it. Not a good product in a good market. A product that the market pulls out of the company — where demand is so strong that the product barely needs to be sold.
He called it product/market fit.
Marc Andreessen took the concept from classroom discussion to startup canon. In his 2007 blog post "The Only Thing That Matters," Andreessen argued that product/market fit is the single variable that separates companies that matter from companies that don't. Not team quality. Not product elegance. Not funding. The market. "In a great market — a market with lots of real potential customers — the market pulls product out of the startup," Andreessen wrote. A great team in a bad market will fail. A weak team in a great market can stumble into success because the market's demand compensates for the company's deficiencies. The only scenario where team and product dominate is when the market is already validated — and the question becomes who captures it.
The diagnosis is visceral, not analytical. Andreessen described both states in terms any founder recognises. When product/market fit is not happening, every metric moves reluctantly. Customers aren't getting value. Word of mouth isn't spreading. Usage isn't growing. The sales cycle drags. Deals collapse. The company feels like it's pushing a boulder uphill. When product/market fit is happening, customers are buying as fast as the company can produce. Servers can't scale fast enough. Revenue accumulates without proportional sales effort. Journalists call because they've heard about the product before you pitch them. The boulder is rolling downhill and the founders are sprinting to stay ahead of it.
Sean Ellis, who led growth at Dropbox, LogMeIn, and Eventbrite, developed the most practical measurement tool for PMF in 2010. His survey asks a single question: "How would you feel if you could no longer use this product?" If 40% or more of respondents answer "very disappointed," the product has achieved fit. Below 40%, it hasn't. The threshold isn't arbitrary — Ellis derived it from studying companies that achieved sustainable growth versus those that stalled. The 40% benchmark became the standard diagnostic across Silicon Valley, giving founders a concrete number to target instead of relying on intuition.
The concept divides a startup's life into two fundamentally different operating modes. Pre-PMF is search mode. Every decision, every hire, every feature serves a single purpose: finding the configuration of product and market that creates genuine pull. The roadmap is flexible. The burn rate is low because capital spent before PMF is capital spent on guessing.
Post-PMF is execution mode. The company has found the thing that works, and every dollar, every hire, every process exists to do more of it, faster. The questions change. Pre-PMF: "What should we build?" Post-PMF: "How do we build it faster?"
Scaling prematurely — before PMF is established — is the most common cause of startup death. The Startup Genome Project's analysis of over 3,200 startups found that premature scaling killed 70% of the companies that failed. They hired sales teams before anyone wanted to buy. They built infrastructure before demand justified the cost. They optimised a machine that hadn't yet found its purpose.
The most telling example of accidental PMF discovery is Slack. Stewart Butterfield's company Tiny Speck spent years building Glitch, a multiplayer online game that never found an audience. During development, the team built an internal messaging tool to coordinate across offices. The game struggled. The chat tool thrived. Every new hire at Tiny Speck became a power user of the internal system within days. When Butterfield pivoted and released the tool as Slack in 2013, the product hit 8,000 signups on its first day of limited preview. Within two years, over a million daily active users. The game never had product/market fit. The communication tool — built as a side project, without a business plan — had it from the moment it existed. PMF doesn't care about your intentions. It cares about whether the product solves a problem so real that people reorganise their workflows to accommodate it.
The concept has an uncomfortable corollary that most startup advice avoids stating directly: if PMF is primarily determined by the market, then most of what founders spend their time on — product design, team building, fundraising — is secondary until the market question is answered. A brilliant product in a market that doesn't care is a beautiful failure. A mediocre product in a market that's desperate for any solution is a startup that can be fixed. The market cannot be fixed. It can only be found or abandoned.
Product/market fit is the moment when a product stops being something the company pushes and becomes something the market pulls. Everything before that moment is a search. Everything after it is a race.
Section 2
How to See It
Product/market fit — and its absence — leaves unmistakable signatures in customer behaviour, growth dynamics, and competitive response. The signals are not subtle once you know what to look for. The challenge is that many of PMF's imposters — manufactured growth, viral mechanics, promotional spikes — produce metrics that look similar on a dashboard but feel completely different inside the company. Learning to read the real signals separates investing in a search that's converging from one that's going nowhere.
Startup
You're seeing Product/Market Fit when organic growth outpaces paid acquisition. Customers refer the product without being asked. Retention curves flatten — users who try the product keep using it week after week rather than dropping off. The strongest signal: customers express anxiety at the possibility of losing access. When Sean Ellis asked Dropbox users how they'd feel if the product disappeared, over 40% said "very disappointed." That emotional intensity — not usage metrics, not revenue — is the most reliable indicator of genuine fit.
Business
You're seeing the absence of Product/Market Fit when growth requires constant, expensive effort. Every new customer is acquired through advertising, discounting, or aggressive outbound sales. Churn is high. Usage is shallow — people sign up but don't return. The team keeps adding features hoping the next one will be the breakthrough. Internally, there's a persistent sense that the company is pushing rather than being pulled. The product works. The market doesn't care enough.
Investing
You're seeing Product/Market Fit when a company's unit economics improve as it grows. Customer acquisition cost drops because word of mouth replaces paid channels. Lifetime value rises because retention strengthens. Revenue per employee climbs. These are the financial signatures of a product the market wants — each new customer costs less to acquire and generates more value because the product itself does the selling.
Markets
You're seeing Product/Market Fit when competitors begin copying the product's specific design choices rather than competing on their own terms. Imitation is the market's acknowledgment that the product has defined the correct solution shape. When every CRM startup began mimicking Salesforce's dashboard layout in the early 2000s, or when every project management tool adopted Trello's card-based interface, the copying wasn't laziness. It was recognition that PMF had been achieved and the market's expectations had been set by the leader.
Section 3
How to Use It
Decision filter
"Has this product achieved genuine pull — are customers actively seeking it, retaining on it, and recommending it without incentives? If the honest answer is no, every dollar spent on scaling is a dollar spent amplifying something the market hasn't yet validated."
As a founder
Treat PMF as a binary state, not a gradient. You either have it or you don't. The temptation is to declare partial fit — "we're getting close" — and begin scaling. Resist. Partial fit means customers use the product but don't depend on it. They churn when a better option appears or when the promotional pricing expires. Run the Sean Ellis survey. If fewer than 40% of active users would be "very disappointed" without your product, you haven't found fit. Stay in search mode. Reduce burn. Talk to the users who are disappointed — they're the signal. The users who are indifferent are the noise.
Superhuman, the email client founded by Rahul Vohra, operationalised this precisely. In 2017, Vohra surveyed users and found that only 22% would be "very disappointed" — well below the 40% threshold. Instead of launching broadly, he segmented the responses. Users who loved it were power email users who valued speed. Users who were lukewarm were casual senders who didn't need keyboard-first design. Vohra narrowed the target audience, rebuilt features around the power users' needs, and resurveyed. The score climbed past 40%, then past 50%. Only then did he scale. The discipline of staying in search mode until the score crossed the threshold saved Superhuman from the most common startup death: scaling a product that isn't yet loved.
As an investor
PMF is the single most valuable diagnostic for evaluating growth-stage companies. Revenue growth alone is unreliable — it can be manufactured through discounting, channel partnerships, and aggressive sales teams. The question isn't whether the company is growing. The question is whether the growth is organic and durable. Ask for retention cohort data. If the earliest customer cohorts retain at 60%+ after twelve months, the product has genuine fit. If early cohorts decay steadily, the company is refilling a leaking bucket. Ask how much of new customer acquisition comes from word of mouth versus paid channels. A company where 50%+ arrives organically is being pulled by the market. A company acquiring 90% through paid channels is pushing.
As a decision-maker
Before committing resources to scaling any product or initiative, apply the PMF diagnostic. Does the product retain users without intervention? Do customers describe it as essential rather than useful? Would a significant portion of the user base be genuinely upset if it disappeared? If the answers are equivocal, the product is in search mode — and the correct allocation is a small team iterating quickly, not a large team executing a scaling plan. The most expensive mistake in corporate strategy is applying scale resources to a product that hasn't found fit.
Common misapplication: Confusing growth with fit. A product can grow rapidly without achieving PMF — through viral mechanics that drive signups without retention, through heavy discounting that attracts price-sensitive users who churn when prices normalise, or through distribution partnerships that deliver volume without engagement. Groupon grew faster than almost any company in history during 2010–2011. But the growth was driven by deep discounts that attracted bargain hunters, not loyal customers. When the discounts ended, the customers vanished. Revenue peaked at $3.2 billion and the company's market cap fell from $16 billion at IPO to under $1 billion. Growth without retention is not product/market fit. It's a leaking bucket with a fire hose pointed at the top.
Second misapplication: Treating PMF as a product achievement rather than a market relationship. Founders who say "we've built product/market fit" are framing it as something they constructed. The more accurate statement is "we've found product/market fit" — it's a discovery about the relationship between what they've built and what the market demands. The relationship can change. The market can evolve. A competitor can shift expectations. The product that achieved PMF six months ago may need to earn it again if the landscape shifts. PMF is maintained, not preserved.
Section 4
The Mechanism
Section 5
Founders & Leaders in Action
Product/market fit is not a theory these founders studied and applied. It's a condition they experienced — sometimes by design, sometimes by accident, and almost always with a clarity that retrospective analysis makes look inevitable but was anything but obvious at the time.
The recurring pattern: the moment of fit was recognisable not from a dashboard but from the quality of the chaos. Support queues overwhelmed. Servers crashing under load. Customers angry about features that didn't exist yet — which meant they cared enough to be angry.
Pre-PMF, the company begs for attention. Post-PMF, the company struggles to keep up with it. The transition happens fast — and the founders who recognised it earliest were the ones who scaled most effectively.
Before Andreessen codified product/market fit as a concept, he lived it at a velocity that shaped his entire intellectual framework. In 1993, as a 22-year-old developer at the University of Illinois' National Center for Supercomputing Applications, he co-authored Mosaic — the first web browser with an intuitive graphical interface. Within months, Mosaic captured 53% of all web traffic. Users downloaded it at a rate that overwhelmed NCSA's servers. Nobody had marketed the browser. Nobody had a distribution deal. The product spread because it solved a problem — making the early web navigable by non-engineers — so completely that anyone who tried it immediately shared it.
When Andreessen co-founded Netscape in 1994 with Jim Clark, the company inherited Mosaic's fit and amplified it commercially. Navigator launched in December 1994 and reached ten million users within four months — an adoption curve steeper than any prior software product. The company's revenue went from zero to $85 million in its first full year. Netscape didn't have a marketing budget proportional to its growth. It didn't need one. The market pulled the product with an intensity that made traditional go-to-market strategy irrelevant. Sales representatives reported that customers called them, not the reverse. The product was being recommended faster than the company could track the recommendations.
Andreessen didn't theorise PMF from textbooks. He experienced it — and then experienced its opposite. After Microsoft bundled Internet Explorer with Windows in 1996, Netscape's browser share began its decline. The product hadn't gotten worse. Navigator was still technically capable. But Microsoft's distribution advantage — every new PC came with IE pre-installed — changed the market structure. Netscape's fit depended on being the easiest way to access the web. When IE became pre-installed and free, that job was already done before the user ever heard of Netscape. The lesson hit hard: PMF depends on market structure, not just product quality. A product can maintain every feature that created the original fit and still lose it when the competitive landscape shifts beneath it. That experience — ecstatic fit followed by structural erosion — became the foundation of everything Andreessen later wrote about why the market is the dominant variable.
Amazon's book-selling phase is frequently cited as a PMF story, but the more revealing case is Amazon Web Services. In 2003, Amazon's infrastructure team was building internal tools to manage the company's growing computational needs. Chris Pinkham and Benjamin Black wrote a paper proposing that Amazon's internal infrastructure could be offered to external developers as a service. Bezos approved the project. AWS launched S3 in March 2006, followed by EC2 in August.
The fit was immediate. Startups that had been spending months configuring servers and negotiating hosting contracts could now launch infrastructure in minutes at a fraction of the cost. Within the first year, AWS had hundreds of thousands of registered developers. Y Combinator startups began defaulting to AWS as their infrastructure layer. The growth required no enterprise sales team — developers adopted the service bottom-up because it solved a problem that every startup faced and no existing provider addressed at the right price point and speed.
Bezos recognised the PMF signal through a characteristic he'd later describe as the customer working backwards for you. AWS customers didn't just use existing services. They submitted feature requests at a pace that outstripped Amazon's ability to build. The demand was so intense that it justified reallocating engineering resources from Amazon's retail business — a counterintuitive capital allocation decision that only made sense if the PMF signal was unambiguous.
The AWS story also demonstrates that PMF can emerge from unexpected directions within an established company. Amazon's core retail business had PMF with consumers. AWS achieved an entirely separate PMF with developers and startups — a market Amazon's board of directors had never targeted and most analysts didn't understand for years. Bezos's willingness to follow the PMF signal even when it led away from Amazon's core identity is the reason AWS existed at all. By 2024, it generated over $90 billion in annual revenue and accounted for the majority of Amazon's operating profit. The retail business was the original plan. The infrastructure business, discovered through an internal tool that found unexpected external demand, became the profit engine.
The iPod offered Apple's first clean example of PMF after years of struggling for relevance in personal computing. In October 2001, the MP3 player market was crowded with devices from Creative, Diamond, and a dozen smaller manufacturers. Jobs ignored their feature comparisons entirely. The iPod's 5 GB drive held 1,000 songs — less storage per dollar than several competitors. But the click wheel, iTunes integration, and the simple promise of "1,000 songs in your pocket" created a product experience that no spec sheet could capture. First-year sales were modest — 125,000 units in the holiday quarter. But the retention signal was unmistakable: iPod owners didn't switch to cheaper alternatives. They upgraded. They evangelised. By 2004, Apple owned 65% of the hard-drive-based MP3 player market. The fit wasn't with the device. It was with the system — device plus software plus music store — that made every alternative feel broken by comparison.
The iPhone extended the insight into a different kind of PMF — fit achieved through category creation rather than category entry. When Jobs unveiled it in January 2007, the smartphone market already existed. Nokia sold 435 million phones that year. BlackBerry had intense PMF with enterprise email users. Jobs wasn't entering this market. He was replacing it. The iPhone didn't compete on existing terms — worse at email than BlackBerry, worse at calls than Nokia, no app ecosystem, no 3G, no physical keyboard. On a feature comparison, it lost to a $200 Nokia N95 on almost every measurable dimension.
The PMF signal came from a metric no competitor tracked: time spent with the device. iPhone users didn't just make calls and check email. They used the device for hours each day — browsing the web, looking at photos, watching video, navigating maps. The engagement pattern was closer to a personal computer than a phone. Apple sold one million iPhones in 74 days. By the time the App Store launched in 2008, the device had become a platform — and the PMF wasn't just with the hardware. It was with the ecosystem of applications the hardware enabled. Apple hadn't found fit within the existing smartphone market. It had created a new market — mobile computing — and achieved fit so strong that the old market ceased to exist within five years.
Netflix achieved product/market fit twice — in two different markets, with two different products, serving the same underlying need.
The first fit was DVD-by-mail. When Netflix introduced its subscription model in 1999 — unlimited rentals for a flat monthly fee, no late fees, no due dates — the response from film enthusiasts and serial viewers was immediate. These customers, who rented fifteen or more movies per month from Blockbuster, switched overnight. Subscriber count grew from 300,000 in 2000 to 4.2 million by 2005. Blockbuster, with 9,000 stores and $5.9 billion in revenue, dismissed the threat because its own customer base — impulse renters visiting stores for Friday night entertainment — wasn't defecting. Blockbuster was measuring a different market.
The second fit was streaming. In 2007, Netflix launched streaming with roughly 1,000 titles — a fraction of the 100,000-plus DVDs available by mail. The early product was objectively inferior on selection, quality, and device support. But the PMF signal emerged within months: streaming usage per subscriber climbed week over week even though no marketing promoted it. Customers who had been hiring Netflix for "access a deep catalog of films" were now hiring it for "give me something to watch right now." The convenience premium of instant access overwhelmed the selection deficit. By 2010, streaming had overtaken DVD-by-mail in hours watched.
The transition also illustrates how fragile PMF becomes during a format shift. In 2011, Hastings attempted to split Netflix into two companies — Qwikster for DVDs and Netflix for streaming. The move triggered a customer revolt. Netflix lost 800,000 subscribers in a single quarter and the stock dropped 77%. Hastings reversed the decision within weeks. The lesson: Netflix had PMF for a unified service that served both the "browse a deep catalog" and "watch something now" jobs simultaneously. Splitting the product destroyed the fit — not because either product was worse, but because the combination was what customers had actually hired. PMF can be with the bundle, not just the individual components. Hastings learned that distinction at a cost of $12 billion in market cap.
Section 6
Visual Explanation
Section 7
Connected Models
Product/market fit sits at the centre of startup strategy — the threshold that separates search from execution, hypothesis from evidence, burning capital from generating returns. No model operates in isolation. PMF draws its power from frameworks that improve the search process, creates productive tension with models that operate on different timescales, and leads naturally to questions about scaling that other frameworks address. The connections below map the most important interactions.
Reinforces
Founder-Market Fit
A founder with deep, earned knowledge of the market finds product/market fit faster because they already understand the customer's real problem — not the version that appears in surveys, but the version that exists in daily frustration. The founder-market fit advantage is speed: a fitted founder starts the PMF search closer to the answer.
Stripe's Patrick and John Collison were developers who had personally experienced the pain of integrating online payments. Their founder-market fit meant Stripe's first product wasn't a guess — it was a solution to a problem they'd lived with for years. The product achieved PMF almost immediately because the founders' experience had pre-validated the hypothesis. Without founder-market fit, the search for PMF is a random walk. With it, the search is directional — the founder's instinct narrows the space to a neighbourhood where fit is likely to exist.
Reinforces
Jobs to Be Done
Clayton Christensen's Jobs to Be Done framework provides the analytical structure for identifying what product/market fit actually means in a given context. PMF answers "does the market want this product?" JTBD answers the prior question: "what is the market actually hiring this product to do?" Without JTBD, PMF can be achieved for one dimension of the customer's need while leaving other dimensions unaddressed — creating vulnerability to a competitor who serves the full job.
Slack didn't just achieve PMF for "team messaging." It achieved PMF for the job of "keep me connected to everything happening at my company without checking six different tools." That job definition — broader than chat, narrower than enterprise software — explained why Slack replaced email, file sharing, and notification systems simultaneously. The JTBD lens ensures that PMF is measured against the real job, not a reductive feature comparison. A product can have PMF for a partial job and still be displaced by a competitor that addresses the whole one.
Section 8
One Key Quote
"You can always feel product/market fit when it is happening. The customers are buying the product just as fast as you can make it — or usage is growing just as fast as you can add more servers. Money from customers is piling up in your company checking account."
— Marc Andreessen, 'The Only Thing That Matters' (2007)
Section 9
Analyst's Take
Faster Than Normal — Editorial View
Product/market fit is the most important concept in startup strategy and one of the least precisely applied. Everyone uses the phrase. Few companies measure it rigorously. Fewer still have the discipline to stay in search mode when the board is pressuring them to show growth metrics that only scale mode produces.
The concept's greatest strength is its honesty. PMF is a market verdict, not a management opinion. No amount of internal conviction can substitute for the external signal of customers who retain, refer, and express genuine distress at the prospect of losing access. The Ellis survey works because it bypasses the social dynamics of product feedback — where customers tell founders what they want to hear — and asks a question that reveals actual dependency. "How would you feel if you could no longer use this product?" is not a satisfaction question. It's a loss question. The psychological distance between "somewhat disappointed" and "very disappointed" is the distance between a product people like and a product people need.
The biggest misapplication in practice: declaring PMF prematurely. Founders are under enormous pressure — from investors, from employees, from their own psychology — to declare that fit has been achieved so they can shift to the more rewarding work of scaling. The result is companies that scale a product the market tolerates rather than one it demands. The difference is invisible in the first year and catastrophic by the third. A tolerated product grows through effort. A demanded product grows through pull. The economics diverge exponentially: the tolerated product's acquisition cost stays flat or rises, while the demanded product's drops as word of mouth compounds.
What Andreessen got exactly right: the market is the dominant variable. This remains counterintuitive to founders who believe their team's quality or their product's design is the primary determinant of success. A mediocre product in a market with desperate demand will find PMF because the market will pull it into shape. Customer complaints become the roadmap. Support tickets become the product spec. The market does the work of product management — but only if the demand is real and urgent. A brilliant product in a market with no urgency will die beautifully.
The Ellis test's limitation worth acknowledging: it measures current fit, not durable fit. A product can score above 40% at a point in time and lose fit six months later as competitors improve, novelty wears off, or the market's needs shift. PMF is not a permanent condition. It's a relationship between a product and a market at a specific moment. The most dangerous moment in a startup's life is the quarter after PMF is achieved, when the urgency to listen feels less acute precisely when the consequences of not listening are greatest.
Section 10
Test Yourself
Product/market fit is easy to define and hard to diagnose. Rapid growth can mask its absence. Slow growth can obscure its presence. Viral mechanics can manufacture a convincing imitation. These scenarios test whether you can distinguish genuine PMF from its imposters — and recognise the real thing when it appears in forms that traditional analysis would dismiss.
Is product/market fit at work here?
Scenario 1
A B2B SaaS company grows revenue 300% year-over-year by offering the first year at 80% discount through a channel partnership with a major consulting firm. Customer acquisition is almost entirely through the partnership. When surveyed, 18% of users say they'd be 'very disappointed' if the product disappeared.
Scenario 2
A consumer app grows slowly — 8% month-over-month — for its first eighteen months. The founding team is frustrated by the pace. But when they run the Ellis survey, 52% of active users say they'd be 'very disappointed' without the product. Retention at twelve months is 65%. Almost all new users arrive through word of mouth.
Scenario 3
A social media platform reaches 50 million users in its first year through viral growth mechanics — each new user invites friends to unlock features. Daily active usage is high. But when a competitor launches with a similar feature set, 30% of the user base migrates within three months.
Scenario 4
A developer tools company has 2,000 paying customers and an Ellis score of 63%. The product has no marketing team — all customers discovered it through GitHub, Hacker News, or colleague recommendations. The waitlist for the enterprise tier has 400 companies. The founding team of five can barely keep up with support requests.
Section 11
Top Resources
The PMF literature is concentrated in a handful of foundational essays and books — fewer than for most mental models, because the concept is young and emerged from practitioners rather than academics. Andreessen provides the concept. Ellis provides the measurement. Rachleff provides the origin. The practitioner accounts — particularly Vohra's Superhuman case study — fill in the operational texture of what the search actually looks like from inside a company trying to cross the threshold.
The foundational text. Andreessen's argument that product/market fit is the dominant variable in startup outcomes — more important than team, product design, or funding. The essay provides the clearest taxonomy of a startup's three elements (team, product, market) and argues that market always dominates. The descriptions of what PMF feels like — both present and absent — remain the most cited passages in startup literature. Essential starting point.
Ellis introduced the 40% "very disappointed" threshold that became the standard PMF measurement. The post describes the survey methodology, explains why the benchmark works, and shows how to segment feedback to identify where fit is strongest. The practical contribution: converting Andreessen's abstract concept into a measurable test that any startup can run within a week.
The most detailed public account of a company systematically engineering its way to PMF. Vohra describes how Superhuman used the Ellis survey, segmented users by fit score, identified the features driving the strongest responses, and iterated until the score crossed the threshold. The article operationalises PMF measurement in a way that Andreessen's essay and Ellis's survey do not — turning the concept into a repeatable process with concrete steps.
Ries's build-measure-learn loop is the operational framework for the PMF search. The entire methodology — validated learning, the pivot, innovation accounting — describes the process of searching for product/market fit through rapid experimentation. The chapters on pivot decisions are directly relevant: when should you change direction because the current product isn't finding fit? The most systematic guide to the search process itself.
Griffin synthesises the PMF literature with a practitioner's lens, collecting the most useful frameworks from Andreessen, Rachleff, Ellis, and others into a single reference. Particularly valuable for its treatment of the relationship between PMF and adjacent strategic concepts — distribution, pricing, competitive dynamics, and the distinction between PMF for consumer versus enterprise products.
Leaders who apply this model
Playbooks and public thinking from people closely associated with this idea.
The MVP framework says ship the smallest thing that tests your hypothesis. Product/market fit says the product must be good enough that 40% of users can't live without it. The tension between "minimum" and "fit" is real and productive. An MVP is designed to be disposable — a learning instrument. PMF requires the product to be indispensable — a retention engine. The gap between disposable and indispensable is where most startups stall.
The resolution is sequencing. The MVP tests whether the hypothesis is directionally correct — whether anyone cares enough to engage. PMF requires the product to evolve past the MVP into something that genuinely satisfies the market's need. Teams that treat the MVP as the destination — shipping the minimum and expecting fit to appear — misunderstand both concepts. The MVP identifies the terrain. PMF requires building a real structure on it. The first cycle answers "is there demand?" The subsequent cycles answer "is this product good enough to retain?"
Tension
Network Effects
Products with network effects create a specific PMF paradox: the product can't achieve fit until it has enough users to deliver its core value, but it can't attract enough users until it demonstrates fit. Facebook wasn't useful with ten users. It needed density within a social graph — enough friends present that the feed was interesting — before the product delivered on its promise. This chicken-and-egg dynamic means network-effect products often show misleading PMF signals: rapid growth driven by novelty or social pressure rather than genuine product satisfaction.
The reverse tension matters too: network effects can sustain a product past the point where it would otherwise lose fit. Users stay on a social platform not because the product is excellent but because their contacts are there. The switching cost created by the network masks declining satisfaction. For investors, traditional PMF signals — retention, usage frequency — can overstate fit for network-effect products. The diagnostic must separate "stays because the product is great" from "stays because leaving is costly." One is genuine fit. The other is captivity.
Leads-to
Do Things That Don't [Scale](/mental-models/scale)
The search for product/market fit almost always requires unscalable founder effort. PMF is not found from a conference room. It's found by talking to customers individually, building features for specific users, providing concierge-level service, and doing manual work that no scaling company could sustain. Paul Graham's principle is the operational methodology for the PMF search.
Airbnb's founders found PMF by personally photographing apartments, rewriting listings, and meeting hosts in their homes. DoorDash's founders found it by personally delivering food. These unscalable actions weren't inefficiencies — they were the search process itself. Each manual interaction taught the founders something about what the market actually wanted that no analytics tool could have revealed. The unscalable work creates the understanding. The understanding creates the product. The product creates the fit.
Leads-to
Iteration [Velocity](/mental-models/velocity)
Once the PMF signal begins to emerge — retention improving, organic growth appearing, the Ellis score climbing — the speed of product iteration determines how quickly the company converts a weak signal into a strong one. PMF strengthens through cycles of user feedback, product improvement, and re-measurement. The company that iterates weekly converges on strong PMF faster than the company that iterates monthly.
Superhuman's journey from 22% to 58% on the Ellis survey was driven by rapid iteration on the specific features power users valued most. Each cycle — survey, identify gaps, build, resurvey — tightened the fit. Iteration velocity is what turns "approaching PMF" into "achieved PMF." Slow iteration means the market moves while the product stands still. Fast iteration means the product chases the market's feedback signal with enough speed to catch it.
One pattern the literature underweights: PMF can be destroyed by the company's own actions. Digg achieved undeniable fit with its link-sharing community between 2004 and 2010 — organic growth, passionate users, cultural influence. Then in 2010, Digg launched v4, a redesign that prioritised publisher content over community submissions. Users revolted. Traffic dropped 26% in a single week. The community migrated to Reddit. Digg's PMF wasn't lost to a competitor. It was abandoned by a redesign that broke the contract between the product and its users.
The most useful reframe for founders: PMF is not something you build. It's something you find. The distinction matters because "building" implies control — you design the product, you ship it, you achieve fit. "Finding" implies exploration — you test hypotheses, follow signals, and hold your product vision loosely enough to go where the market leads. The market has no obligation to validate your original thesis. It will validate the thesis it cares about, and your job is to find it before you run out of capital.
The current frontier: PMF measurement for AI products. Most AI tools are being positioned as feature improvements to existing software — "like X, but with AI." The PMF question cuts through the positioning: are users retaining because the AI capability genuinely transforms their workflow, or because novelty hasn't yet worn off? Early data suggests that AI products face a distinctive retention challenge — the initial "wow" of generated output declines rapidly as users discover the tool's error patterns and limitations. The 40% Ellis threshold is as applicable to AI products as to any other category, but the measurement window may need to be longer. The true test of AI PMF isn't whether users are impressed in week one. It's whether they're dependent in month six.
One honest observation about the concept itself: PMF is easier to recognise than to engineer. Andreessen, Ellis, and Rachleff all describe what fit looks like. None of them provide a reliable recipe for creating it. The best frameworks — customer development, lean methodology, JTBD analysis — increase the probability of finding fit by improving the quality of the search. But the search itself remains unpredictable. Slack found PMF by accident while building something else. Superhuman found it through methodical measurement and iteration. Amazon found it in AWS by offering an internal tool to outsiders. There is no single path. The only constant is that companies that stay in search mode until the signal is unambiguous outperform those that declare victory prematurely.
