What "product-market fit" actually means, and how to measure it

Most startups do not have product-market fit until Series B, and that is what the round is for rather than a verdict on you. So the job through seed and Series A is measuring it simply and then acting on the measurement. PMF is the point where you have real confidence the product solves a genuine need, in a way that can support a profitable business: a Sean Ellis survey for the need, LTV:CAC above 1 for the economics, both and not one. Then run the Superhuman analysis on the result, find who loves the product and exactly why, and double down on that until word of mouth turns up. And the failure I actually see isn't missing PMF, it's having no thesis to test it against, which is what the four-blank exercise is for.

Most startups don't have product-market fit until Series B.

Not seed. Not Series A. Nobody says it out loud because it sounds like an insult, when it is actually the job description.

That is what the money is for. You raise to buy the runway to scale the team and build the product while you go and figure PMF out, and figuring it out stays the main goal right through seed and Series A.

Which only works if you can measure it, and most of the conversation about this term is built to avoid measuring anything.

Ask five growth people to define PMF and you'll get five different answers, which is exactly why the term survives. It sounds authoritative while committing to nothing.

Listen, as a growth advisor who gives high-minded advice for a living, I get the appeal. 🙂

So make them commit. If your investor or advisor can't give you a concrete answer when you press them, take the advice with a grain of salt and watch them squirm.

A definition you can actually use

PMF is the point where you have real confidence the product solves a genuine need, in a way that can support a profitable business.

Three load-bearing parts:

• A specific point. A measurable moment you can say you've crossed, not a feeling that things seem to be going well.
• Solves a real need. The core problem-to-value hypothesis is validated. People genuinely want this.
• Supports a profitable business. At minimum, LTV:CAC above 1. Below that you're donating your marketing budget to your users.

Both. Not one.

That last part is where most definitions quietly cheat. A product people love with economics that never work is not fit, it's an expensive hobby. Growth that looks profitable on a product nobody needs is not fit either, it's a channel that happens to be working for now.

How to measure each half

For the need, Sean Ellis's survey is still the most reliable read: ask users how disappointed they'd be if they woke up tomorrow and couldn't use the product. The share saying "very disappointed" is your number.

For the economics, is LTV:CAC above 1? That's it. If you make more from a customer than you spend acquiring them, the math works and you can go argue about how much better than 1 you want it.

Now of course there are a dozen philosophies on this. These two are just the earliest things I can measure with confidence and the simplest to run, which is a different claim than saying they're the most sophisticated. If you already have retention cohorts deep enough to read a flattening curve, read it. Most teams asking me this question don't yet.

Running the Superhuman analysis on the result

The survey number on its own is a reading, not a decision. What turns it into work is the analysis Superhuman built on top of it, and it is four steps rather than a benchmark.

1. Get the number. Ask the question, and count the share who say they would be very disappointed. Forty percent is the bar people work to, and it comes from Sean Ellis rather than from anyone's intuition.

2. Throw away the rest of the responses. Not literally, but for this purpose. The people who said somewhat disappointed or not disappointed are not your market yet, and building for them is how a product gets blander. Segment down to the very-disappointed group and treat that group as the product's actual users.

3. Find out what they love, specifically, in their own words. This is the step that gets skipped, and it is the one that pays. Not "they like the product." The one thing they would fight you to keep. You are looking for a sentence you did not write, repeated by people who have never met each other.

4. Double down on exactly that. For as long as it takes. Not on the roadmap, not on the thing the somewhat-disappointed group asked for. On the thing your people already love.

Then you run it again, because this is continuous rather than a gate you pass. Understand the product. Understand what people love about it. Read the number. Find who loves it and why. Double down. Repeat.

You know you are there when word of mouth turns up. Not when the survey clears a threshold. When people love the product enough to tell somebody else about it without being asked, because nobody recommends a product that is merely fine. That is the same signal that decides how much paid you have to buy, which is a separate argument in referrals as a growth channel, and it is not a coincidence that the two land in the same place.

The part almost everyone misses: this runs to Series B

PMF is stage-dependent, and my read is more aggressive than the one you usually hear.

Most startups do not have product-market fit until Series B. Not seed. Not Series A. And that is not a failure, it is the entire point of the money: you raise to scale the team and build the product while you go and figure PMF out, and figuring it out stays the main goal right through both of those rounds.

I've sat in a lot of rooms where a founder is quietly ashamed of not having something they were never supposed to have yet, usually because an investor said it like a verdict. If you are at seed and you do not have it, you are on schedule.

So the useful question through seed and Series A isn't "do we have PMF." It's "are we set up to find out, and are we actually doing the finding."

The failure I actually see

It isn't missing PMF. It's having no thesis to test it against.

A team runs the survey, gets 22% very disappointed, and has no idea what to do with that, because they never wrote down who it was supposed to be disappointing or why. The number is only meaningful against a claim you made in advance. Without one, you're just collecting readings.

That's the real gap, and it's fixable in an afternoon.

The four blanks

With early-stage founders I run one exercise first. Four blanks that take an afternoon to argue about and change everything downstream.

1. Who is the ICP, specifically enough that you could list ten of them by tomorrow? If you can't name ten actual people or companies, the definition is still a category, not a customer.
2. What's the hair-on-fire problem we solve for them, better than anyone else manages to? Hair-on-fire is the bar. Mildly annoying does not get bought.
3. What alternatives are we actually up against, including doing nothing? Doing nothing is the one people leave off, and it's usually the market leader.
4. What two differentiators are we betting on, the ones that make us two to three times better at that one job? Two, not seven. If everything is a differentiator, nothing is being bet on.

Filled in, for a generic B2C habit app, it looks like this. Deliberately ordinary, because the value is in how specific it is rather than how clever.

Blank	Filled in
ICP	People who have restarted a running habit at least twice and quit both times, 28 to 45, already paying for one fitness app they barely open
Hair-on-fire problem	They do not trust themselves to keep going, and every product they have tried made that worse by showing them a broken streak
Alternatives	A free tracker, a coach at ten times the price, a friend who nags them, and doing nothing, which is what most of them pick
Two differentiators	Progress measured against the goal they set rather than against a streak, and a plan that survives a missed week without resetting

That is what "specific enough" means. Every one of those is checkable against real users, and every one of them is wrong in a way you would notice.

The point isn't to get it right first time. It's to have something specific enough that you can check yourself against it later.

Then you go and check

You build, you learn, and you come back to the four blanks with real usage in hand.

Does the ICP still hold, or did the people who actually stuck around look different from the ones you wrote down? Did the differentiation survive contact with real users, or did they like you for a third thing you didn't rank? Was the problem the one you thought it was?

That loop is what turns a PMF score from a vanity reading into a decision. The survey tells you where you stand. The thesis tells you what to do about it.

The takeaway

Stop using product-market fit as a vibe you gesture at, and stop treating not having it at seed as a verdict. Most startups don't have it until Series B. Finding it is the job that the seed and the Series A are paying for.

So measure it, simply. The need with a Sean Ellis survey, the economics with LTV:CAC above 1, both and not one. Then do the part that turns a reading into work: segment to the people who would be very disappointed, find out what they love in their own words, and double down on that until people love it enough to tell someone else.

And write the four blanks down first, because a measurement without a thesis behind it is just a number you argue about.

Common questions

What is a simple definition of product-market fit?

The point where you have real confidence the product solves a genuine need, in a way that can support a profitable business. It takes validating both the need and the economics, not one of them.

How do you measure product-market fit?

Two measures. A Sean Ellis survey, asking how disappointed users would be if they could no longer use the product, for the need. LTV:CAC above 1 for the economics. They're the earliest things you can measure with confidence, not the most sophisticated.

Should a seed-stage startup have PMF already?

Usually not, and that's fine. My read is that most startups don't have it until Series B, so at seed and Series A you are on schedule without it. Finding it is what those rounds are paying for. The question that matters at that stage is whether you have a specific enough thesis to know when you've found it, and whether you are actually doing the finding rather than assuming it will arrive.

What is the Superhuman PMF analysis?

It is what you do with the survey result rather than another way of taking it. Get the share who would be very disappointed if they lost the product, then segment down to only those people and treat them as your actual market. Find out what they love, specifically and in their own words. Then double down on exactly that instead of on what the less-enthusiastic group asked for. You know it worked when word of mouth turns up, because nobody recommends a product that is merely fine.

Can you have PMF without profitability?

Not by this definition. A product people love with unit economics that never work is not fit, it's an expensive hobby. You need the need and the profitable business.