# Product-market fit: how to recognise it

Source: https://sarang.pro/product-market-fit-sean-ellis-test-retention/
Author: Sarang | Published: 25 October 2026

Product-market fit is the point at which a product has found a group of customers who want it badly enough to keep using it and paying for it. Before it, most effort goes into pushing. After it, the market starts pulling. Founders talk about it constantly, yet many struggle to say whether they have it.

This guide covers where the term comes from, the Sean Ellis survey as Ellis himself describes it, how to read retention curves, and what to do with the results. It assumes you already have some users. If you are still testing whether anyone wants the idea at all, start with a minimum viable product and customer interviews first.

## Where the idea comes from
Marc Andreessen popularised the phrase in a [2007 blog post](https://pmarchive.com/guide_to_startups_part4.html) titled "The only thing that matters". His definition is short: "Product/market fit means being in a good market with a product that can satisfy that market."

He also describes what it feels like. Without it, customers are not quite getting value, word of mouth is not spreading, usage is growing slowly and sales cycles drag. With it, "the customers are buying the product just as fast as you can make it." That description is vivid, but it is hard to measure. The two methods below give you numbers.

## The Sean Ellis test, in his words
Sean Ellis, who later co-wrote *Hacking Growth*, set out the method in a [2019 article](https://medium.com/growthhackers/using-product-market-fit-to-drive-sustainable-growth-58e9124ee8db). The core question is:

> How would you feel if you could no longer use [ProductName]?
The answer options are: very disappointed, somewhat disappointed, not disappointed, and "N/A, I no longer use [ProductName]".

Ellis writes that, in his experience, a product can grow sustainably once around 40% of users who try it say they would be very disappointed without it. Note the wording. He presents 40% as a threshold drawn from his experience, not as a statistical law.

His practical guidance is just as useful as the threshold:

- Survey people who have used the product recently, ideally within the last couple of weeks. People who never really used it, or stopped months ago, will skew the result.
- Treat 30 responses as the minimum for a directional read, and 100 or more as more reliable.
- Study the very disappointed group. Find out who they are, why the product matters to them and what the main benefit is.
- Use what you learn to sharpen targeting, positioning, onboarding and the roadmap.

## How Superhuman used it
The best-known application is from Rahul Vohra, founder of the email client Superhuman, described in [First Round Review](https://review.firstround.com/how-superhuman-built-an-engine-to-find-product-market-fit/). He surveyed users who had used the product at least twice in the previous two weeks. Alongside the Ellis question he asked who would most benefit from the product, what the main benefit was, and how it could be improved.

The first result was 22% very disappointed. By focusing on the types of user who were most likely to say very disappointed, the score rose to 33%. He then split the roadmap in half: one half doubling down on what the fans loved, the other fixing what held back the somewhat disappointed users who valued the same things. Within three quarters the score reached 58%. He tracked it regularly and made it the key result in a product OKR.

The survey worked for Superhuman because Vohra repeated it and used the answers to steer the roadmap each quarter.

## Reading retention curves
Surveys measure what people say. Retention measures what they do. Lenny Rachitsky, in a [2022 newsletter on product-market fit](https://www.lennysnewsletter.com/i/48314350/how-do-i-know-if-i-have-product-market-fit), puts flattening retention curves first among the signs of fit, because they show people's revealed preferences.

A retention curve starts with a group of users who signed up in the same period (a cohort) and shows what share of them are still active in each later week or month. Every curve falls at first, as people who were only curious drop off. What matters is what happens next:

- If the curve keeps falling towards zero, nobody sticks. You do not have fit yet.
- If it flattens above zero, some part of the market wants the product. Rachitsky's point is that a durable business needs that plateau to be much higher.
- If newer cohorts flatten higher than older ones, your changes are working.
He quotes Casey Winters: "You have product-market fit when your retention creates enough money (or content/virality) to drive sustainable acquisition."

## A worked illustration
An example (illustration only). A UK start-up offers a meal-planning app for parents. After six months it has about 1,500 monthly users. It surveys 120 people who used the app in the last two weeks: 30% say very disappointed. Not there yet.

Looking only at the very disappointed group, the team notices that most are parents of children with food allergies, who say the app saves them from checking every recipe by hand. Among the somewhat disappointed, the common complaint is that the shopping list does not work with their supermarket's app.

The team narrows its marketing to allergy-aware families, improves allergen filtering and adds a shopping list export. Three months later, a survey of a fresh set of recent users shows 41% very disappointed, and the retention curves for new monthly cohorts flatten at a higher level than earlier ones. Both signals now point the same way.

## Common mistakes
- **Surveying the wrong people.** Including dormant sign-ups drags the score down. Including only your biggest fans inflates it.
- **Too few responses.** Ten replies is an anecdote. Aim for Ellis's minimum of 30.
- **Treating 40% as a finish line.** It is a sign you can start investing in growth, not proof that growth will follow.
- **Reading blended retention.** An overall active-user number hides whether new cohorts stick. Look at cohorts separately.
- **Paying for growth too early.** Paid acquisition can mask poor retention for a while. Rachitsky notes that paid growth can mislead founders.
- **Re-surveying the same people.** Vohra surveyed new users each time to avoid repeat answers.

## When to use each signal
Use the Ellis survey early, when you have dozens or hundreds of active users but not enough history for clear cohorts. It also tells you why people value the product, which retention data cannot. Use retention curves once you have several months of cohorts, and give them more weight than the survey, because behaviour is harder to fake than opinion. When both agree, you can be reasonably confident. When they disagree, trust retention and go back to talking to customers.