011 min
What breaks without product market fit
A team can ship on schedule, hear no complaints and still be building something nobody needs. If it cannot tell the difference, it behaves the same either way. It hires, buys ads and adds features to a product that people try once and drop.
Michael Seibel of Y Combinator wrote in 2016 that founders who wrongly think they have fit start hiring, spend more each month and polish the product before they know what to build. The money goes out before the team has learned what the product should be.
Rahul Vohra, the founder of the email app Superhuman, faced a version of this question. He did not want a feeling that people liked the product. He wanted a number that said whether people needed it. The sections below follow how he got one, after a look at what fit feels like and where the idea came from.
021 min
What product market fit feels like
Seibel's warning raises a plain question: what does real fit look like from inside a company? Andreessen described it in an essay posted on June 25, 2007. He said a team can feel the absence of fit. Customers get little value from the product, word of mouth does not spread, usage grows slowly, sales take too long and many deals never close.
He said a team can feel fit just as clearly. Customers buy as fast as the team can supply the product, or usage grows as fast as the team can add servers. Money from customers piles up, and the team hires sales and support staff as fast as it can.
Seibel gave a shorter test. A team has fit when usage overwhelms it. He added that he looks for a founding team that is rushing to serve a growing number of happy, loyal and ideally paying customers.
When fit is almost too much
Heavy demand brings its own problem. Seibel notes that a team with fit is often so busy keeping the service running that it cannot make major changes to the product. That is a good problem to have. It still means the product the team has on that day is the one it must keep running, so the team should settle the basics before demand arrives.
032 min
Where product market fit came from
If fit is so obvious once it arrives, who gave it a name? The short answer is a venture capital firm and a blog post. Andy Rachleff says he learned the idea at Sequoia Capital, where Don Valentine invented it. Rachleff recalls that Valentine looked for companies with so much market demand that they could make many mistakes and still succeed.
Andreessen then spread the term. In his 2007 essay, Marc Andreessen said that fit exists when a team has a product that satisfies the needs of a market worth serving. He called the idea Rachleff's corollary and wrote that the only thing that matters for a startup is reaching it.
Rachleff also gave the idea a precise test. Rachleff's own test is whether a team has proven its value hypothesis: what it builds, who it is for and how the business makes money. He draws the sequence from Steve Blank, who argued that a team should prove the value hypothesis first and only then test a growth hypothesis, which is a plan for scaling up. That order explains why a team with fit can still be losing money: it has shown that people want the product, and it has not yet shown how to grow.
Why the market comes first
Andreessen put the market ahead of the team and the product. Andreessen argued that the market, not the team, decides the result when the two are mismatched. A very good team in a market with little demand struggles. An average team in a market with strong demand still sells. So in his view the first job is to judge the market, and the product follows from what that market needs.
042 min
How to measure product market fit
Andreessen's description depends on feel, and a team close to its own product cannot always trust its feel. So the next question is how to measure fit with something harder than a mood.
Rachleff starts with word of mouth. Rachleff warns that teams fool themselves when paid growth looks like demand, and says word of mouth is the real test. A team can buy signups with ads, so the question is whether growth continues when the buying stops. Growth that compounds through recommendations is hard to fake.
Sean Ellis built a survey for the same purpose. It asks users one question: how would you feel if you could no longer use the product? The answers are very disappointed, somewhat disappointed or not disappointed.
Score = users who answered "very disappointed" ÷ all users who answered the question
Ellis found that companies struggling to grow almost always scored under 40 percent on this question, and companies with strong traction almost always passed it.
A survey is one signal among several. Each has a weak spot, so teams read them together.
| Signal | What to look at | Where it can mislead |
|---|---|---|
| Ellis survey | Share answering very disappointed | A few answers, or only the most loyal users |
| Retention | Share of users who keep using the product over time | Habit or lock-in can look like need |
| Word of mouth | Share of new users who arrive on recommendation | Paid growth can look the same |
| Sales and support | Short sales cycles, requests for more | Early customers may not be typical |
Numbers say how big the gap is. They do not say why, and user interviews fill that in.
052 min
How Superhuman measured product market fit
A single score says where a team stands, not what to do next. Vohra's account of Superhuman shows the second half. Superhuman's first survey found that 22 percent of users would be very disappointed to lose the product, which told its founder that it had not yet reached fit. He was not discouraged by the low number. It gave the whole team a target to work on.
He added three questions to the survey: what kind of person would benefit most, what the main benefit is, and how the product could improve. Each one was used in a step.
Find who the product is for
Vohra first looked only at users who said very disappointed, and grouped them by the kind of person they were: founders, managers, executives and business development. He then counted only the survey answers from users who looked like those fans. The people who loved the product were the market, and the others were a different market. This is segmentation applied to a survey.
Learn what the fans love
The fans' answers to the main-benefit question had a pattern. They valued the speed of the product, its focus and its keyboard shortcuts.
Split the roadmap
Next came the somewhat-disappointed group. Vohra kept those who also valued speed, since they already wanted what the fans wanted. Their answers to the improvement question pointed to one main barrier: Superhuman had no mobile app. He then split the plan in two. One half put more effort into what fans loved. The other half removed what held back the near-fans.
Some users will not convert
Not every somewhat-disappointed user is worth chasing. Vohra warned that some of the somewhat-disappointed users would never become devoted users, whatever the team changed. That is why he filtered the group by the benefit it valued before spending effort on it.
The result was large. After Superhuman narrowed the count to users who looked like its fans the score was 33 percent, and about three quarters of product work later it had reached 58 percent.
061 min
A second case: Slack and Glitch
Superhuman improved a product that some users already loved. Another team faced weaker demand and reached a different decision. By October 2012, Stewart Butterfield decided that Glitch, the team's online game, would never turn a profit. The company had built internal chat tools while developing that game, and Butterfield turned those tools into a new product, Slack.
The variable that differs from Superhuman is what the team changed. Vohra kept the product and sharpened it for the users who loved it. Butterfield's team gave up the product and kept its tools and its skills. Both decisions started from evidence that demand was too weak, and neither started from a launch.
Slack launched publicly in August 2013, and 8,000 customers signed up within 24 hours. Sign-ups show interest, not need, so the stronger signal came later. By February 2015 about 10,000 new daily active users were joining each week. That pattern matches what Andreessen called fit: usage that grows about as fast as the team can serve it.
071 min
False signs of product market fit
Both cases ended well, but teams often mistake other things for fit. Three mistakes are common, and each one feels like success while it is happening.
Headcount and funding
Founders often point to the size of the team or a large funding round as proof of fit. Seibel's view is that these show that investors and employees believe in the plan. They do not show that customers need the product.
Growth that was bought
Rachleff says he is always amazed at how many teams fall for growth they paid for. Ads can fill a signup page and still leave a product that nobody mentions to a friend. The test is whether usage grows after the spending stops.
Wanting a solution, not your product
A buyer who says a problem matters has not said that your product is the answer. Interest in the solution to a problem can be strong while interest in one specific product is weak. This is why a survey about the product, such as the Ellis question, tells a team more than a survey about the problem. Teams that mix the two often feel sure of fit too early.
082 min
Where product market fit measures fall short
The measures above are useful, but none of them is a law.
The 40 percent line is a rule of thumb
Vohra says Ellis found the 40 percent figure by benchmarking nearly a hundred startups. That is a pattern across a small group, not a proof. The question also asks people to predict how they would feel, which is different from watching what they do. A team should read the score with retention and word of mouth beside it.
The score can fall when the product succeeds
Vohra notes that the score can fall as a product reaches past early adopters, because later users expect everything their current tool already does. A falling score in that case does not mean the team did something wrong. It means the audience changed, and the team needs to learn what the new users require.
Fit belongs to a group of users
Superhuman's score rose when it counted only users who looked like its fans. The same product had strong fit with one group and weak fit with another. A single number for all users hides that difference, so a team should look at groups of users separately.
Fit is not a finish line
Markets and customers change, so fit measured once may not hold. Vohra's team kept tracking the score after it reached 58 percent for this reason. The practical move is to measure on a schedule and to treat a drop as a question to investigate.
091 min
Product market fit vs. nearby concepts
Teams often use these ideas as if they meant the same thing. Each answers a different question.
| Concept | The question it answers | What separates it from fit |
|---|---|---|
| Not in the library yetProduct market fit | ||
| ProductMinimum viable offer | ||
| MarketingProduct positioning | ||
| Not in the library yetJobs to be done |
The deciding fact is order. Positioning and the job a customer wants done help a team find the right market, a minimum viable offer tests the idea cheaply, and fit is the evidence that all of it worked.
101 min
How to check product market fit this week
You can have a first score in a few days using only a survey tool and a spreadsheet. The artifact you end with is a table of survey answers grouped by user type.
- Pick the users. Choose people who have used the product more than once, so they can imagine losing it.
- Send four questions: how would you feel if you could no longer use the product, who do you think benefits most, what is the main benefit you get, and how could we improve it for you.
- Work out the score, the share who answered very disappointed. Write it at the top of the sheet.
- Group the very disappointed users by role and by main benefit. Count the score again using only users who look like them.
- Read the somewhat-disappointed answers. Keep the ones that value the same benefit as your fans, and set the rest aside.
- Split your next plan in two: half on what fans love, half on what holds the near-fans back.
- Run the survey again after the next batch of changes. Each round is one turn of a feedback loop.
Treat the first score as a baseline, not a verdict. A low one is the normal place to start, as it was for Superhuman.
?3 questions
Questions people ask
Does product market fit mean the business is profitable?
What should a team do before it has product market fit?
Does product market fit apply to a new product inside a large company?
§5 sources
Sources
Marc Andreessen, "The Pmarca Guide to Startups, Part 4: The only thing that matters", 2007.
Michael Seibel, "The Real Product Market Fit", Y Combinator, 2016.
Rahul Vohra, "How Superhuman Built an Engine to Find Product Market Fit", First Round Review, 2018.
Andy Rachleff in conversation with Mike Maples, "How to Know If You've Got Product Market Fit", Starting Greatness, 2020 (a transcript).
Show all 5 sourcesShow fewer sources
A summary of Slack's history, Wikipedia, "Slack (software)". (software)
