What is product market fit, and how do I know if I have it?

Where the definition comes from

The definition is from an essay Marc Andreessen posted on 25 June 2007, part 4 of his guide to startups. He credits the idea to Andy Rachleff, formerly of Benchmark Capital, and calls it Rachleff's Corollary of Startup Success: the only thing that matters is getting to product market fit.

When it is happening, he writes, customers buy the product as fast as you can make it and money from customers piles up in the company account.

What it looks like for a small product

Andreessen's picture is a company hiring sales staff. What follows is this page's own reading of his signs for a product one person built. All three are things strangers do.

• Strangers pay without being chased. Someone you have never spoken to finds the page and pays. Andreessen's sign is customers buying. Sean Ellis calls charging the final validation.
• Some come back. They use it again next week, or pay again next month. Andreessen's sign is usage growing. Retention is how you count it.
• Some tell others. A new customer says a friend sent them. Andreessen's sign is word of mouth spreading.

No source here sets a pass mark for any of the three, and this page will not invent one. Count them, and watch whether they rise without more pushing.

How to measure it with one survey question

Sean Ellis, a startup marketer, published a survey on 18 May 2009 and called one question in it the most important.

1. Ask people who have used the product, not everyone who signed up. Ellis's sample: people who have experienced the core of the product, used it at least twice, and used it in the last two weeks.
2. Offer four answers: very disappointed, somewhat disappointed, not disappointed, and one for people who no longer use it.
3. Divide those who answered very disappointed by everyone who answered. In the example, 10 divided by 40 is 25%.
4. Set it beside 40%. The survey post says over 40%. Ellis's post of 2 September 2009 says at least 40% of active users.

In the comments under that second post Ellis says he looks for at least 30 responses. He also names the limit of his own test: he asks while a product is still free, and if nobody is then willing to pay, the result was probably a false positive.

One founder's account of measuring it

Rahul Vohra's account, dated 12 November 2018, is first hand: he founded Superhuman, an email product. In the summer of 2017 he sent Ellis's question to recent users, from the 100 to 200 he had. 22% answered very disappointed. Counting only the kinds of user found among that 22% made it 33%. After three quarters of product work, half on what those users loved and half on what held others back, it was 58%.

Vohra reports that Ellis set the 40% line after benchmarking nearly a hundred startups, and says results start to point the right way at around 40 responses, where Ellis says 30. It is one company, with a team of 14 and money raised. Read it as a method, not a promise.

What it is not: the mistakes a new founder makes

• Signups. A signup costs the visitor nothing. Ellis's own sample leaves out anyone who has not used the product at least twice.
• Compliments. Kind words are free as well. The polite answer in the survey is somewhat disappointed, and Ellis reads it as a product that is only nice to have.
• Friends using it. Friends try it because of you. Count strangers only.

Will my app make money, and when should I stop? sets out the six things that decide whether a stranger pays. Why startups fail has what surveys and counts of post mortems say about poor fit, and who ran each.

Before fit and after it

Before fit, the work is the product and who it is for: ideal customer and positioning. Ellis's advice at that stage is to keep spending low. After fit, the work is reach, and cost starts to matter. Andreessen's essay sets one thing aside by assumption: that a customer brings in more than it cost to win them. Customer acquisition cost is the page for that sum, and every paid channel is judged on it, Google Ads included.

How Sami uses it

Sami works from the numbers you give it and the records AskSami keeps. The honest read AskSami writes, free once you sign up, scores the six things that decide whether a stranger pays, each yes, partly or no, and marks one unknown when nothing on record shows it.

Signups and payments are tied to their source through the snippet in your product and your own payment webhook. Sami does not send the survey. You do, and the answers you bring back as a file or a screenshot become dated records. Where there is no data, Sami says so and never invents a figure.