The Product Market Fit Survey: Questions, Template and Benchmarks

In the summer of 2017, Rahul Vohra sat staring at Marc Andreessen’s definition of product-market fit with tears in his eyes. Two years building Superhuman, no launch, and every description of “fit” he could find only told him whether he had it after the fact. None told him how close he was, or what to do next.

That is the gap a product market fit survey closes. It takes a feeling founders usually only recognise in hindsight and turns it into a single number you can measure this week, then improve on purpose. This guide covers what the survey actually measures, the one question at its heart and the four-question version that does the real work, exactly who to send it to, how many replies you need before the number means anything, how to read the result without kidding yourself, and what to do when you have no users to survey yet.

What a product market fit survey actually measures

Most definitions of product-market fit are lagging indicators. Andreessen’s famous one, that you feel fit when “customers are buying the product just as fast as you can make it”, is vivid and true, and completely useless before you get there. By the time money is piling up in the account, you already have fit. You wanted to know a year earlier.

Sean Ellis, who ran early growth at Dropbox, LogMeIn and Eventbrite, found the leading version. Ask your users one question: “How would you feel if you could no longer use this product?” Give them three options: very disappointed, somewhat disappointed, not disappointed. Then measure the share who say very disappointed. That percentage moves before revenue does, which is the whole point. It tells you whether you are building something people would miss while you can still do something about it.

The logic is simple. People who would be very disappointed to lose your product are the ones who have folded it into their lives. They have no good substitute. They are the users who stay, who pay, and who tell other people. The more of them you have as a proportion of everyone who has tried the product, the closer you are to a business that grows on its own.

The 40% benchmark, and what it really means

After running his survey across nearly a hundred startups, Ellis landed on a number: 40%. Companies where fewer than 40% of users said they would be very disappointed almost always struggled to grow. Companies above 40% almost always found traction. That threshold has been the working benchmark ever since.

A useful reference point comes from Hiten Shah, who put Ellis’s question to 731 Slack users in a 2015 open research project. 51% said they would be very disappointed without Slack, at a point when the product already had around half a million paying users. Slack is one of the fastest-growing software companies ever built, and it cleared the bar by eleven points. That tells you how hard 40% actually is.

Here is the part most guides skip: 40% is an empirical rule of thumb, not a law of physics. Ellis derived it by observing what separated the startups that grew from the ones that stalled. It is a strong signal, but the number you get is only as good as the people you ask, and a single reading tells you less than the trend over time. So read your score as a band, not a pass or fail line.

The bar also shifts with your category, which is why a flat 40% for everyone is too blunt. High-frequency consumer products, the apps and daily-use tools people touch several times a day, tend to need to clear 50% before the signal is convincing, because casual attachment is easy to come by and real dependence is the thing you are trying to detect. A B2B product used deep in someone’s workflow can be compelling at 40% from a tight ideal-customer segment. Calibrate the threshold to how central your product is to a user’s day, not just to the famous number.

The band matters because the same 30% can mean two completely different things. Averaged across a messy pile of tyre-kickers, 30% is a shrug. But if that 30% hides a tight segment scoring 55%, you do not have a product problem, you have a focus problem, and that is a much better problem to have.

The four questions (your template)

The score comes from one question. The reason to run the whole survey is the other three, which tell you what to do about the score. This is the exact four-question set Vohra used at Superhuman, generalised so you can lift it straight into a Typeform, Google Form or your survey tool of choice.

1. How would you feel if you could no longer use [product]?
   - Very disappointed
   - Somewhat disappointed
   - Not disappointed (it isn't really that useful)

2. What type of people do you think would most benefit from [product]?
   [open text]

3. What is the main benefit you receive from [product]?
   [open text]

4. How can we improve [product] for you?
   [open text]

Question one is your score. Questions two, three and four are open text, and they are where the gold is. Question two, “what type of people would most benefit”, is deceptively powerful: your happiest users almost always describe themselves, in the exact words you should be using in your marketing. Question three tells you the one benefit worth protecting above all others. Question four surfaces the specific thing standing between a fence-sitter and a fan. Those open-text answers are a voice-of-customer goldmine in their own right, and worth reading with the same care you would give any voice-of-customer research.

Keep the survey to these four. Every extra question you bolt on costs you completion rate and buys you noise. The four earn their place because each one feeds a specific decision later.

Who to survey (the step almost everyone gets wrong)

This is where most product market fit surveys go quietly wrong, and it is the single thing that ruins the number. You cannot send the survey to everyone who ever signed up.

If you poll people who registered and never came back, or who are three days into a trial they have barely opened, you are asking people to rate a product they have not really used. Their “not disappointed” answers are meaningless, and they will drag your score down and muddy your segmentation. The fix is to survey only people who have genuinely experienced the core of your product.

Vohra’s rule was clean: survey users who used the product at least twice in the last two weeks. Recent, repeated use is a good proxy for “this person has actually felt what the product does”. Adapt the exact threshold to your product’s natural rhythm, but keep the principle. Experienced users only.

There is a catch in that filter worth holding in mind, and most guides never mention it. Restricting to recently active users is correct, but it means your score is measured only over the people who stuck around. Everyone who tried the product and quietly left is not in the denominator, so a strong score can describe a small, loyal core sitting on top of a leaky bucket. Read the number next to your retention, not instead of it. 45% very disappointed among a base that shrinks every month is a warning, not a victory.

On sample size, the good news is that you need far fewer responses than you would guess. Superhuman had between 100 and 200 eligible users to poll when they started, and Vohra found the results become directionally correct at around 40 responses. That is well within reach of an early-stage product. More responses sharpen the picture and make segmentation safer, but do not wait for thousands before you learn something. One more rule from Superhuman: never survey the same person twice, because repeat responses skew the benchmark that the whole 40% figure rests on.

From a score to an engine: what to do with the answers

A number on its own changes nothing. The reason Superhuman went from 22% very disappointed to 58% in three quarters was not the survey, it was the four-step engine they built on top of it. Here is how to run it.

Segment to find who already loves you. Group every response by the answer to question one. Then look only at the very disappointed group and ask what they have in common. Are they a particular role, industry, company size, use case? At Superhuman, the very disappointed users clustered around busy professionals who lived in their inbox. Segmenting down to those people alone lifted the score by ten points, from 22% to 33%, with no product change at all. Nothing shipped. They simply stopped averaging their best customers together with people who were never going to love the product.

Paint your high-expectation customer. Take question two, “what type of people would most benefit”, from your very disappointed group, and use their words to write a single vivid profile of the person who needs you most. This is the high-expectation customer: not a broad persona, but the most demanding person in your target market who will get the most out of your core benefit and tell others. Superhuman wrote theirs as a named character, right down to how many emails she reads a day. It gave the whole team one person to build for.

Learn why they love you, and what holds the rest back. From the very disappointed group, read question three to find the main benefit. At Superhuman it was speed and keyboard shortcuts, over and over. That is the thing you protect and deepen. Then, and this is the counterintuitive move, ignore the “not disappointed” crowd entirely. They will ask for features that pull you off course and then churn anyway. Focus instead on the somewhat disappointed users for whom your main benefit already resonates. Question four tells you what is holding them back. For Superhuman it was the lack of a mobile app. Those are the people one or two fixes away from love.

Split the roadmap in half. This is the rule that ties it together. Spend half your roadmap doubling down on what the very disappointed users already love, and half fixing what holds the winnable fence-sitters back. Only double down on strengths and your score stalls. Only fix weaknesses and a competitor out-delights you. Do both, re-survey, and repeat. Superhuman made the very disappointed percentage the single most important number the product team tracked, and rebuilt the roadmap around it every quarter.

That cadence, survey, segment, learn, build, re-survey, is the difference between a metric and an engine. The score tells you where you are. The engine is what moves it.

Fit is not a finish line

One thing worth bracing for: your score can fall as you grow, and that does not mean you have broken anything. Early adopters are forgiving. They love your core benefit and shrug off the rough edges. As you push past them, you reach users who are more demanding and expect parity with whatever they use today, so a healthy 45% among your first cohort can slip when you widen the funnel.

Vohra flagged this directly. The way through depends on your model. If your product has strong network effects, like a marketplace, the core benefit keeps improving as more people join, so fit tends to compound. If you are a straightforward software or ecommerce business, there is no shortcut: you keep running the engine, rebuilding the roadmap each quarter so the product improves at least as fast as your audience broadens. The score is not a box you tick once. It is a vital sign you keep reading.

What if you don’t have any users yet?

Here is the honest limit of the whole method. A product market fit survey needs a live product in real users’ hands. It is a post-launch instrument. If you are pre-launch, or you are a merchant about to put a new product, pack or flavour on the shelf, there is nobody to survey yet, and running the survey on a handful of friends tells you nothing.

That does not mean you fly blind until launch. It means you are asking a different question. Before you have users, you are not measuring fit, you are testing demand: would the right people actually want this, and choose it over what they use now? The strongest answer is still a pre-sale or a deposit, because money is the one signal people do not give to be polite. The concept testing platforms worth using measure purchase intent against a benchmark rather than asking people whether they like an idea in the abstract, which they always do.

This pre-launch read is the gap TestFeed is built for. You can put a concept, a product, a pack, a claim, a name or an in-context price in front of your target audience and get back a purchase-intent score, the reasons behind it in shoppers’ own words, and a clear next move, in days rather than weeks. Treat it as what it is: a directional, pre-spend signal to decide which ideas deserve a real launch, not a market forecast or a guaranteed sales number. It gets you to the ideas worth building, so that when you do have users, the product market fit survey has a fighting chance of clearing 40%. And once you clear it, that is your cue to stop testing the water and push growth hard.

Five ways people ruin the number

Most bad product market fit surveys fail in one of a few predictable ways. Watch for these.

Surveying everyone. The biggest one. Include dormant sign-ups and trial-never-starteds and your score is noise. Filter to recently active users only.

Chasing the wrong feedback. The loudest requests often come from people who would not be disappointed to lose you. Acting on them builds a muddled product for users who will churn regardless.

Treating 40% as gospel. It is a rule of thumb from one person’s dataset, not a certification. A 38% with a 55% segment inside it is a far better position than a flat 41%.

Running it once. A single score is a snapshot. The value is in the trend as you improve, so build the survey into a repeating loop rather than a one-off audit.

Polling the same people again and again. Repeat responses distort the benchmark. Survey fresh, eligible users each round.

Run it this week

You do not need a research budget or a big user base to start. Pull the list of people who have used your product at least twice in the last fortnight. Send them the four questions. Read the very disappointed percentage, then go find out who those people are and what they love. That single loop will teach you more about your product’s future than another quarter of guessing. If your number is below 40%, you now know exactly where to look. If it is above, you know it is time to grow.

Frequently asked questions

What is a product market fit survey?

A product market fit survey measures how much your users would miss your product if it disappeared. The core question, popularised by Sean Ellis, is ‘How would you feel if you could no longer use this product?’ with three answers: very disappointed, somewhat disappointed, and not disappointed. The percentage who say ‘very disappointed’ is your score. It is a leading indicator, which means it tells you whether you are on track before revenue and growth make it obvious.

What is a good product market fit survey score?

The widely used benchmark is 40%. Sean Ellis found that companies where at least 40% of users said they would be very disappointed without the product tended to grow sustainably, while those below 40% usually struggled. Treat it as a strong rule of thumb rather than a law. Below 25% suggests you have not found the right market yet, 25% to 39% means a segment probably loves you and you should find it, and 40% or more is your cue to push growth.

What questions should a product market fit survey ask?

Four questions. First, ‘How would you feel if you could no longer use this product?’ with very, somewhat and not disappointed as the options, which gives you the score. Second, ‘What type of people do you think would most benefit from this product?’, which reveals who your product is really for. Third, ‘What is the main benefit you receive?’, which tells you the value to double down on. Fourth, ‘How can we improve this product for you?’, which surfaces what is holding fence-sitters back.

Who should you send a product market fit survey to?

Send it only to people who have genuinely experienced the core of your product, not everyone who ever signed up. A good filter is anyone who has used it at least twice in the last two weeks. Surveying brand-new sign-ups, inactive users or people still in a free trial they have barely touched will pollute the number and point your roadmap in the wrong direction. Never survey the same person twice, as repeat responses distort the benchmark.

How many responses do you need for a product market fit survey?

Fewer than most people think. Rahul Vohra of Superhuman found the results become directionally correct at around 40 responses, which even an early-stage startup with a small user base can reach. More responses tighten the picture and make segmentation more reliable, but you do not need thousands. What matters far more than volume is that every respondent has actually used the product.