Product-market fit: how to know you have it, and why most claims are wrong.

What product-market fit actually is

Product-market fit is the threshold at which a product satisfies strong, repeatable market demand. Marc Andreessen described it in 2007 as the moment when "you can feel the pull of the market." Sean Ellis turned it into a measurable test: at least 40 percent of active users say they would be very disappointed if they could no longer use the product. PMF is binary in spirit but the measurement is continuous — you track your score over time and watch it climb.

Andreessen's framing is qualitative. Customers are buying as fast as the team can produce. Usage is growing as fast as the team can add servers. Money is piling up in the company checking account. The team is hiring as fast as it can. The product is being pulled out of the team's hands. You do not have to ask whether you have PMF. You know.

Sean Ellis's framing is quantitative. Survey your active users. Ask the very-disappointed question. Count the percentage who say "very disappointed." If 40 percent or more answer that way, the product has crossed the PMF threshold for the segment you surveyed. Below 40 percent, you do not have PMF yet — even if revenue is growing, even if investors are excited, even if the team is busy.

Claim: Sean Ellis derived the 40 percent benchmark from survey data across roughly 100 startups in the early 2010s. Companies that hit the threshold went on to grow strongly. Companies below it almost always stalled. Source: "The Startup Pyramid" by Sean Ellis, Startup Marketing blog (2009-2012). Context: The 40 percent threshold is not magic. It is a regression line through real startup data. Some companies with 35 percent went on to find PMF and grow. Some at 45 percent stalled. But the threshold has held up as a reliable indicator across more than a decade of subsequent use.

Where Marc Andreessen named it (2007)

Marc Andreessen published the term in a June 25, 2007 blog post titled "The Only Thing That Matters" on his pmarchive blog. He credits Andy Rachleff at Benchmark Capital with the underlying idea, which Rachleff in turn credits to Don Valentine at Sequoia. The phrase has become the most widely cited concept in early-stage startup vocabulary.

Andreessen's essay made three claims that have aged well. First, that PMF is the only thing that matters before you have it — team, capital, market positioning, technology choices are all secondary to whether the product satisfies market demand. Second, that you can always feel PMF when it is happening — the symptoms are unmistakable. Third, that until PMF is reached, the team should be doing whatever it takes to find it, including pivoting the product or the target market repeatedly.

The essay is still the standard reference. Almost every subsequent piece on PMF builds on Andreessen's framing. Paul Graham at Y Combinator extended the idea with "Do Things That Don't Scale" in 2013, arguing that early-stage founders should manually onboard their first customers to learn what PMF actually means in their market. Brian Balfour at Reforge layered on the four-fits framework, which says PMF is necessary but not sufficient for scaling.

"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. 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)

The Sean Ellis 40 percent test

The Sean Ellis 40 percent test is a survey-based measurement of PMF. Ask active users how they would feel if they could no longer use your product. Three answer options: Very disappointed, Somewhat disappointed, Not disappointed. The percentage who answer "very disappointed" is your PMF score. Forty percent is the threshold Sean Ellis derived from research across 100 startups in the early 2010s.

The test works because it captures dependency, not satisfaction. A customer who is "somewhat disappointed" likes the product but has alternatives. A customer who is "very disappointed" cannot easily replace it. The 40 percent threshold separates products that customers depend on from products customers merely enjoy. Only the dependency segment drives durable growth.

Three implementation details matter. First, only survey active users — not signups, not trial users, not lapsed users. Second, time the survey two to four weeks into active use, once early-trial enthusiasm has settled. Third, segment the responses by persona, use case, and channel. The aggregate score often hides a clear PMF segment that the noise from other segments is dragging down.

Claim: Superhuman, the email client founded by Rahul Vohra, increased its Sean Ellis very-disappointed score from 22 percent to 58 percent in twelve months by systematically segmenting users and rebuilding the product for the high-expectation persona. Source: "How Superhuman Built an Engine to Find Product-Market Fit" by Rahul Vohra, First Round Review (2018). Context: Vohra's essay is the operating manual for using the Sean Ellis test as more than a single snapshot. The Superhuman team treated the score as a controllable metric and built quarterly product cycles around lifting the score. The methodology has been adopted by dozens of subsequent product-led companies.

Rahul Vohra's Superhuman PMF engine

The Superhuman PMF engine is Rahul Vohra's systematic method for finding product-market fit, published in First Round Review in 2018. It builds on the Sean Ellis survey by adding three steps: segment users by persona to find the high-expectation customer, double down on what the very-disappointed segment loves, and ignore the users who are not in your target segment. Superhuman used the engine to triple its PMF score in twelve months.

The engine has four parts. First, run the Sean Ellis survey and segment by user type. Second, look at the segment with the highest very-disappointed percentage — this is your high-expectation customer. Third, ask the very-disappointed users follow-up questions: What benefit do you get? How can the product be improved for you? Fourth, build quarterly product cycles around what the high-expectation segment loves and what they are missing. Stop trying to please users outside the segment.

The mental shift in Vohra's method is the hardest part. Most teams try to please every user. The engine says: identify the segment that already depends on you, double down on them, and let the other segments churn. Superhuman lost some users early in the process but the ones who stayed were the ones who built the company. The engine is a discipline more than a methodology.

Signals you have product-market fit

PMF shows up in patterns. Six signals separate companies that have crossed the threshold from companies that have not. None of them is the Sean Ellis score by itself. The score is necessary but not sufficient. The full picture comes from triangulating multiple signals over time.

Sean Ellis score of 40 percent or higher. The first and most direct signal. If less than 40 percent of active users would be very disappointed without your product, you do not have PMF yet, regardless of revenue or growth rate.

Retention curve that flattens. Cohort retention does not fall to zero. A meaningful share of users keeps coming back month after month. The curve might bend down for the first few months and then level off at 30, 40, or 60 percent. The flatness is the signal; the absolute level is secondary.

Organic word of mouth. Users tell other users without being prompted, paid, or incentivized. Net promoter score is one measurement. Referral coefficient (k-factor) is another. The qualitative version is just paying attention to how new users describe how they found you.

Demand exceeds your ability to serve. Andreessen's original framing. You are out of capacity. Sales is closing deals you cannot deliver. Engineers cannot ship features fast enough. Customer service is drowning. The team is hiring as fast as it can. The product is being pulled out of your hands.

Customer acquisition gets cheaper as you scale. Word of mouth and organic search start showing up in the new-customer mix. CAC trends down or stays flat as the company grows, instead of climbing the way it does for products without PMF. This signal takes 12+ months to confirm but is highly reliable when it appears.

Customers do not churn even when you raise prices. Pricing experiments reveal real dependency. A product with PMF can typically raise prices 20-50 percent and lose less than 10 percent of customers. A product without PMF loses customers proportional to the price increase, or worse.

Signals you do not have it (yet)

The signals you do not have PMF are mirror images of the signals you do. Each one shows up in audit after audit. The common thread is a team that is busy and a business that is not compounding. Recognizing the signals early saves quarters of misallocated capital.

Sean Ellis score below 30 percent. A clear no. Between 30 and 40 percent is borderline — you may be approaching PMF but have not crossed the threshold. Below 30 percent, the work is on the product, not on the marketing.

Retention curve that falls to zero. Every cohort eventually leaves. The product is a one-time use case or a trial that does not convert to durable usage. No amount of acquisition will produce durable growth when retention is broken.

CAC climbs as you grow. Each new customer costs more than the last. The acquisition channels that worked at small scale are saturating, and the channels that scale do not work for this product. This is the most expensive PMF failure to fix because the team usually does not notice until margin collapses.

Customers churn on price increases. A 10 percent price increase costs 20 percent of customers. The product is not differentiated enough to defend a higher price. Repeated pricing tests are the most reliable test of real dependency.

The team is busy but the business is not growing. Hiring is happening. Features are shipping. Investor decks look great. But cohort revenue is flat or declining. The team is mistaking activity for progress. This is the most common PMF illusion in series-A and series-B companies.

Real examples: Slack, Superhuman, Notion, Figma

Four well-documented PMF journeys: Slack went from internal Tiny Speck tool to public product with PMF in roughly a year. Superhuman used the Sean Ellis engine to triple its score in twelve months. Notion took years of iteration before crossing the threshold. Figma found PMF after multiple pivots away from the original concept.

How four companies found product-market fit, with sources for each

Company	Journey to PMF	Key signal	Source
Slack	Built internally at Tiny Speck as a tool for the gaming company; pivoted to ship the chat tool itself in 2013	2,000 messages sent per team predicted 93% retention	First Round Review (2017), Stewart Butterfield interviews
Superhuman	Took 4 years of building before charging; used Sean Ellis engine to systematically improve PMF score from 22% to 58% in 2018	Sean Ellis very-disappointed score crossed 40% in segment of "founders and managers under 30"	Rahul Vohra essay in First Round Review (2018)
Notion	Multiple rewrites between 2013 and 2018; the team almost shut down twice before the 2018 launch that found PMF	Organic word of mouth and Twitter virality after 2018 launch	Ivan Zhao public talks, Lenny's Newsletter interviews
Figma	Founded 2012, public beta 2015, paid product 2017; multiple pivots from generic design tool to browser-based collaborative design	Designer teams using Figma in production design workflows by 2017-2018	Dylan Field public talks, Figma S-1 (eventual filing)

How to run the Sean Ellis survey (5 steps)

Here is the 5-step method I use to set up the Sean Ellis PMF survey, refined with the segmentation logic from Rahul Vohra's Superhuman methodology. The survey itself takes 14 days end to end — one week to set up and send, one week to collect responses and analyze segments.

1. Identify your active users.Filter to users who have used your product at least twice in the past two weeks. Only ask people who actually know the product. Surveying inactive users gives you noise, not signal. For most products, "twice in two weeks" is a reasonable definition of active. For high-frequency products like Slack, raise the bar; for low-frequency like Airbnb, lower it.
2. Send the very-disappointed question.Ask: "How would you feel if you could no longer use [product]?" Three options: Very disappointed, Somewhat disappointed, Not disappointed. The percentage who answer Very disappointed is your PMF score. Send via email, in-app, or both. Aim for at least 100 responses for statistical reliability.
3. Segment by user type.The aggregate score can hide the truth. Slice the responses by user persona, use case, and acquisition channel. You will often find one persona at 60 percent very-disappointed and others at 15 percent. Build for the 60 percent persona. This is the most important step in the Superhuman methodology.
4. Ask follow-up questions.For users who say "very disappointed," ask three follow-ups. "What type of person do you think would benefit most from [product]?" "What is the primary benefit you get?" "How can we improve [product] for you?" The answers form your high-expectation customer profile and your product roadmap.
5. Rebuild for the high-expectation customer.Use the follow-up answers to double down on what is working for the very-disappointed segment. Stop trying to please users who are not in that segment. Repeat the survey quarterly to track the score over time. Superhuman ran this cycle quarterly for two years and moved its score from 22 to 58 percent.

Claim: Across roughly 50 audits we run per year at Real Growth Matters, the most common PMF measurement mistake is surveying lapsed users or inactive trial users. The score that comes back is usually 12-18 percent very-disappointed, and the team concludes the product has no PMF. Re-running the survey on active users only typically pushes the score 15-25 points higher. Source: Real Growth Matters Inc., internal audit data, 2024-2026. Context: The Sean Ellis test only measures signal against the active-user denominator. Including users who have already churned dilutes the very-disappointed count with people who have already voted with their feet. The fix is operational: filter active users before sending the survey, every single time.

Three myths about product-market fit

PMF is one of the most cited concepts in startup vocabulary and one of the most misunderstood. Three myths show up in almost every founder conversation. Each one has a kernel of truth and a misleading conclusion. Untangling them is the difference between scaling a real business and pretending you have one.

Myth 1: revenue growth means you have PMF

Revenue can grow for reasons that have nothing to do with PMF. A team can buy revenue with discounts, paid ads, or expensive sales motions. The revenue chart climbs. The unit economics get worse every month. Six months later, margin collapses, churn climbs, and the team discovers they never had PMF in the first place. Revenue is a lagging indicator. The Sean Ellis test is a leading indicator. Trust the leading indicator.

Myth 2: PMF is permanent once you have it

Brian Balfour at Reforge has written extensively on this. Markets shift. Competitors enter. User expectations change. The product that had PMF in 2018 can lose it by 2024 without the team noticing. The fix is to re-run the Sean Ellis survey at least annually. A team that thinks PMF is a one-time achievement is a team that will lose it.

Myth 3: PMF means you can scale

Balfour's four-fits framework argues that PMF is necessary but not sufficient for scale. You also need market-model fit (the market supports the business model), model-channel fit (the model supports the channels you can use), and channel-product fit (the channels shape the product). A product with PMF but without the other three fits will hit a scaling ceiling. PMF is the floor, not the finish line.

How PMF fits with adjacent concepts

PMF is the gate before scale. The frameworks around it answer different questions. AARRR pirate metrics is the funnel decomposition for diagnosing where in the customer journey PMF is missing. The north-star metric is the single number a team rallies around after PMF is achieved. Jobs-to-be-done framing is the qualitative lens for finding what value the product is actually delivering. Growth loops describe the structure of how PMF compounds over time.

For early-stage companies, PMF pairs with jobs-to-be-done framing as the qualitative lens and with cohort analysis as the quantitative method for tracking retention. Once PMF is found, the work shifts to picking a north-star metric and decomposing the funnel with AARRR pirate metrics.

For scale-stage companies, PMF pairs with the four-fits framework (market-model, model-channel, channel-product) as the diagnostic for why a product that had PMF cannot scale further. CAC payback and LTV ratio is the unit-economics check that confirms PMF is real and not just a temporary surge.

Quick answers about product-market fit

What is product-market fit in plain English?

The point at which a product satisfies real, repeated market demand. Customers want it, buy it, use it, and would be upset if they could not have it. Marc Andreessen named the concept in 2007.

Who came up with PMF?

Marc Andreessen popularized the term in a June 2007 essay on his pmarchive blog. He credited Andy Rachleff at Benchmark Capital with the underlying idea. The phrase itself is Andreessen's contribution.

What is the easiest way to measure PMF?

The Sean Ellis 40 percent test. Survey your active users with one question: how would you feel if you could no longer use this product? If 40 percent or more say "very disappointed," you have PMF.

Why 40 percent specifically?

Sean Ellis derived the threshold from data across roughly 100 startups in the early 2010s. Companies above 40 percent went on to grow strongly. Companies below it almost always stalled. The threshold has held up across more than a decade of subsequent use.

What is the Superhuman PMF engine?

Rahul Vohra's methodology, published in First Round Review in 2018. It builds on the Sean Ellis survey by segmenting users by persona and rebuilding the product for the high-expectation segment. Superhuman tripled its PMF score in twelve months using it.

Can a company lose PMF?

Yes. Markets shift. Competitors enter. User expectations change. A product that had PMF five years ago may not have it today. The fix is to re-run the Sean Ellis survey at least annually and watch the score over time.

Frequently asked

What is product-market fit?

Product-market fit is the threshold at which a product satisfies strong, repeatable market demand. Marc Andreessen defined it in a 2007 blog post as the moment when you can feel the pull of the market. The Sean Ellis 40 percent test gives it a measurable form: at least 40 percent of users say they would be very disappointed if they could no longer use the product.

Who came up with product-market fit?

Marc Andreessen popularized the term in a June 2007 essay titled The Only Thing That Matters on his pmarchive blog. He credits Andy Rachleff at Benchmark Capital with the underlying idea, which Rachleff in turn credits to Don Valentine at Sequoia. The term itself is Andreessen's contribution.

What is the Sean Ellis 40 percent test?

A survey where you ask active users how they would feel if they could no longer use your product. Options: very disappointed, somewhat disappointed, not disappointed. If 40 percent or more answer very disappointed, you have PMF. Sean Ellis derived the 40 percent threshold from research across roughly 100 startups.

How do you measure product-market fit?

Four main methods. The Sean Ellis 40 percent very-disappointed survey. The retention-curve method (does retention flatten or fall to zero). The organic-growth method (do users tell other users without prompting). The Superhuman PMF engine, which combines the Sean Ellis survey with persona segmentation.

What is the Superhuman PMF engine?

Rahul Vohra's systematic method for finding PMF, published in First Round Review in 2018. It builds on the Sean Ellis survey by segmenting responses by user persona and using follow-up questions to identify the high-expectation customer. Superhuman went from 22 percent very-disappointed to 58 percent in twelve months using this method.

Can you lose product-market fit?

Yes. Brian Balfour at Reforge has written extensively on this. Markets shift. Competitors enter. User expectations change. A product that had PMF in 2018 can lose it by 2024 without the team noticing. The fix is to re-run the Sean Ellis survey at least annually.

How long does it take to find PMF?

Highly variable. Some companies find it in months (Notion, Figma). Most take years of iteration. Paul Graham at Y Combinator argues that most successful companies pivot at least once before finding PMF.

How is product-market fit different from product-channel fit?

PMF is whether the product satisfies the market. Product-channel fit is whether the channels you can afford match the product you have built. Brian Balfour's four-fits framework argues that you need both PMF and product-channel fit to actually scale. PMF without channel fit is a product no one can find.