Product-Market Fit Signals and How to Measure Them

Validation tells you whether an idea is worth building. Product-market fit tells you whether what you built is working. The term was popularised by Marc Andreessen in a 2007 essay, in which he described it as being in a good market with a product that can satisfy that market. It is easier to feel than to define, which is why founders look for measurable signals.

Signal 1: the "very disappointed" test

The growth marketer Sean Ellis proposed a simple survey question for active users: "How would you feel if you could no longer use this product?" with the answers very disappointed, somewhat disappointed, not disappointed and no longer use it. In his experience, companies where at least 40% of users answered "very disappointed" tended to find it much easier to grow. The 40% figure is a benchmark, not a law, but it is widely used.

Rahul Vohra, the founder of the email client Superhuman, described in a 2018 essay how his team turned this question into a regular process: segment the respondents, find the users who would be most disappointed, learn what they love and what holds back the "somewhat disappointed" group, and track the score over time.

Signal 2: retention curves that flatten

Plot the share of each sign-up cohort still active week by week. In most products the curve drops sharply at first. The key question is whether it flattens out at a meaningful level or keeps sliding towards zero. A curve that levels off means a core group of users has found lasting value. A curve that never flattens means you are filling a leaky bucket, and no amount of marketing will fix it.

Signal 3: growth you did not pay for

Word of mouth, organic search, unprompted referrals and inbound sales enquiries are strong signs that the product sells itself. Track the share of new customers who arrive without paid acquisition. If it rises as you grow, that is a good sign.

Signal 4: customers pulling, not you pushing

Users complain loudly when something breaks.
Sales cycles shorten; customers ask to buy before you finish the demo.
Customers use the product in ways you did not plan, and ask for more.
Prices can rise without a jump in churn.

The casino that made control groups a habit

One of the most cited business stories about measuring what works comes from a casino company. In the late 1990s Gary Loveman, a Harvard Business School professor, joined Harrah's Entertainment and went on to lead it. The company built its Total Rewards loyalty programme into a large customer database and became known for testing marketing offers with proper control groups: one group of guests received an offer, a matched group did not, and the difference in visits and spending showed whether the offer actually changed behaviour or just rewarded people who would have come anyway. Loveman described the approach in a 2003 Harvard Business Review article, and later spoke of not running a control group as one of the few things that could get an employee fired.

The lesson for product-market fit is the same: a rising number is not proof unless you know what would have happened without your change. Cohort analysis and holdout groups are how you get there. We look at the broader testing culture of casinos, including the darker side of optimising for time on device, in casino product design as a case study in testing, and at how to size your bets once the signals appear in startups as bets.

When you do not have fit yet

Most products do not have it at first. Go back to your most enthusiastic users, find out what they have in common, and narrow your focus to people like them. Revisit the riskiest assumption from your validation plan; often the product is fine and the segment is wrong.