| Stage | Measure and condition | Planning use |
|---|---|---|
| Sean Ellis survey | 40 percent or more of qualified respondents answer “very disappointed” | A strong quantitative signal that acquisition spend is worth increasing. A lower share points to product work first. |
| Sample size | At least 30 qualified responses; 100 or more is more defensible | Below 30, treat the score as directional only. |
| Retention curve | The curve flattens at some point instead of sliding to zero | Suggests fit with some audience. The next job is to find out which one. |
| Acquisition mix | Ideally more than 50% of new accounts come from direct or organic traffic | Shows customers are pulling the product in, not only responding to paid effort. |
Figures come from Kromatic’s guide to surveying for product-market fit and Lenny’s Newsletter on how to know if you have product-market fit.
The practical answer to how to find product market fit is to measure it. Survey qualified users with the Sean Ellis test, then check that retention flattens before you spend heavily on growth.
Fit is one stage in the wider path from idea to launch, covered in our guide on how to start a startup. The phases below run from defining fit, through the survey and retention checks, to adjusting when the numbers fall short.
Before measuring: define fit and talk to customers
Marc Andreessen’s description, as quoted by Quaff Media, is being in a good market with a product that can satisfy that market. Lenny’s Newsletter records his claim that the only thing that matters is getting to product-market fit. The same piece gives his best-known symptom: customers are buying the product just as fast as you can make it.
That definition has two halves, a market worth serving and a product that satisfies it. Early conversations should test both before any survey runs. If the idea itself is still shaky, work through how to come up with a startup idea first.
- Choose a specific group. Name the customers you believe the product satisfies, so the market half of the definition is testable.
- Talk to people who use the product. Ask what they use it for and what they would do without it.
- Write down their words. The wording customers use for the benefit becomes useful when you read open-ended survey answers later.
- Note who is enthusiastic. Patterns in who cares most tell you which audience to survey and which to ignore.
Measuring: the Sean Ellis test
The Sean Ellis test is a one-question survey. According to Kromatic, the core question is “How would you feel if you could no longer use this product?” Respondents choose how disappointed they would be, and the figure that matters is the share answering “very disappointed”.
Setting the benchmark
Ellis said that every successful business he worked with had a 40 percent or greater result, as reported by Quaff Media. Kromatic describes 40 percent or more of qualified respondents picking “very disappointed” as a strong quantitative signal. It also notes that below that threshold, spending heavily on acquisition tends to be inefficient. Use the benchmark as a gate for growth spending, not as a trophy.
Running it properly
- Filter the audience. Send it only to people who used the core product at least twice in the last two weeks.
- Collect enough answers. Kromatic advises at least 30 qualified responses before treating the score as more than directional, and calls 100 or more defensible.
- Ask why. Pair the score with a follow-up asking what the product’s main benefit is, so you can see what the “very disappointed” group values.
- Read it with other data. Kromatic recommends reading the result alongside retention and engagement data rather than on its own.
Confirming: retention and business signals
A survey captures what users say. Behaviour confirms it, and the retention curve is the clearest behavioural check. Lenny’s Newsletter says that if the curve flattens off at some point, you have probably found fit for some market. A curve that keeps sliding toward zero means the product has not yet found an audience that stays.
Churn is the other side of the same picture, since every user who leaves shows up as a drop in the curve. Quaff Media defines one cohort measure as the percentage of users who continue to pay after onboarding for eight weeks. Tracking cohorts this way shows whether newer groups stay longer than older ones.
Other signals worth tracking
- Net promoter score. Quaff Media describes it as a measure of how likely people are to tell others about your product. It adds a word-of-mouth angle that the disappointment question does not capture.
- Customer acquisition cost against lifetime value. Lenny’s Newsletter says you have found fit when you can repeatably acquire customers for less than their lifetime value. Quaff Media describes the LTV/CAC ratio as comparing what you spend to acquire a customer with what you earn from them.
- Burn multiple. Lenny’s Newsletter gives the formula as Net Burn divided by Net New ARR, which links cash consumption to growth in recurring revenue.
- Organic share. Lenny’s Newsletter says more than 50% of new accounts should ideally come from direct or organic traffic.
- Sales yield. For enterprise products, the same source points to a sales yield greater than 1.0 as a marker of fit.
Pick the signals that match your business model. A consumer app will lean on the survey and retention curve, while an enterprise seller will watch sales yield and burn multiple.
A documented example: Lookout
The clearest documented case in public accounts is Lookout, the mobile security company. As reported by Quaff Media, only 7% of poll respondents said they would be very disappointed without the product. The company swiftly changed its business strategy in response to the poll and focused on its antivirus product. Ellis said the figure was 60% by the time he left the company.
The lesson is in the sequence. The low score was a prompt to change direction, not a verdict on the company. A strategy change followed, and the same question later returned a result well above the benchmark.
Adjusting when the score is low
Kromatic calls the survey a leading indicator, not a finish line. A low score is information about where to work next, and it is most useful early. If you are still deciding how to structure the team and support around the product, the comparison of startup incubator vs accelerator programmes may help.
- Score below the benchmark. Hold back heavy acquisition spend, since Kromatic notes it tends to be inefficient at that level. Look instead at what the “very disappointed” respondents value and build toward it.
- Strong score but a retention curve that never flattens. The survey and behaviour disagree, so check whether the respondents are truly qualified users and whether cohorts leave after onboarding.
- Good retention with a small audience. A flattening curve signals fit for some market. Work out who that group is and whether it is large enough.
- Mixed results across segments. Split the survey by customer type, then concentrate on the segment that answers most strongly, as Lookout did when it narrowed to one product.
Change one major thing at a time, then measure again, so you can tell which change moved the number.
Frequently asked questions
Is 40 percent a hard cutoff?
No. Kromatic describes the 40 percent figure as a heuristic, not a hard cutoff. Treat a result near the line as a reason to look at retention and engagement before deciding.
How often should the survey be repeated?
Ellis’s approach, as reported by Quaff Media, is to distribute the survey once a month until product-market fit is achieved.
Does a steep drop-off in the first month mean there is no fit?
Not by itself. Lenny’s Newsletter says a fairly deep drop-off in the first month is normal in cohort analysis. What matters is whether the curve flattens afterwards.
Which stage of company is the survey for?
Quaff Media describes the question as best suited for early-stage startups, where there is a small group of real users to ask and the product can still change direction.
