Product-market fit examples: how 8 startups found (and proved) it

Everyone quotes Marc Andreessen's line about product-market fit being something "you can always feel." Less helpful is what to do when you can't feel it yet. The companies below couldn't, at first, either. What separated them wasn't a magic moment — it was a deliberate way of finding out whether the market actually wanted what they were building.

Eight examples, from survey scores to scrappy MVPs.

1. Superhuman: from 22% to 58% by treating PMF as a metric

When the email startup Superhuman first ran the Sean Ellis survey in 2017, the result was underwhelming: only 22% of users said they'd be "very disappointed" without it. Another 52% were merely "somewhat disappointed" — the polite middle that kills products quietly.

Instead of guessing, founder Rahul Vohra built an engine. He segmented respondents into personas and zeroed in on the cohort that loved the product most. He read the on-the-fence users' feedback (their top ask: a mobile app), and ignored the "not disappointed" group entirely. He split the roadmap between deepening what fans loved and converting the fence-sitters. Then he re-ran the survey, every cycle. The score became the product team's primary OKR. Three quarters later, it hit 58%.

It's the cleanest example there is of manufacturing fit on purpose. The full breakdown of the engine is here.

2. Slack: a 51% score from people who weren't even asked nicely

Slack felt like an overnight success, but the fit was real and measurable. In an independent research project, Hiten Shah surveyed 731 Slack users with the same "very disappointed" question. The result: 51% would be very disappointed to lose it — comfortably over the 40% line.

Just as useful was the why behind the score. Those users said Slack increased their productivity and improved team collaboration, and nearly everyone flagged the same missing must-have at the time: video conferencing. That's the survey doing double duty — confirming fit and handing you the roadmap in the same breath.

3. Baremetrics: proof that you don't need a perfect score to act

Not every example is a triumphant number, and that's the point. When the analytics company Baremetrics ran the survey in April 2015, it got 138 responses: 32% very disappointed, 58% somewhat, 11% not disappointed.

32% is below the 40% bar — so by the rule, not yet at fit. But the team didn't treat it as a grade. They treated it as a to-do list. The open-ended feedback drove a wave of product features and improvements. A "below threshold" score that you actually mine for direction is worth more than a vanity metric you frame on the wall.

4. Dropbox: the demo video that proved demand before the product

Drew Houston couldn't show investors traction for a file-syncing tool that barely existed. So in 2008 he made a short demo video — a screencast walking through how Dropbox would work — and posted it to a community of early adopters.

The waitlist jumped from around 5,000 to 75,000 people overnight. No finished product, no marketing budget. Just proof that the demand was real before a single line of scaling code got written. That's product-market fit validated the cheapest way possible: with a video and a signup form.

5. Airbnb: doing things that don't scale until the pull appeared

Airbnb's early numbers were flat, and the founders couldn't figure out why. So they flew to New York, knocked on hosts' doors, and noticed the listings had terrible photos. Their fix wasn't a feature — it was renting a camera and shooting professional photos themselves, by hand, one apartment at a time.

Bookings climbed. The unscalable, founder-does-everything hustle surfaced the real demand that the product alone hadn't. Paul Graham later turned the lesson into a mantra — "do things that don't scale" — but underneath it is a PMF story: they manually created the conditions for fit, saw it take, then automated.

6. Buffer: validating willingness to pay with two web pages

Before building anything, Joel Gascoigne tested whether anyone wanted Buffer at all. He put up a simple landing page describing the product with a "Plans and Pricing" button. Clicks led to a page that said the product wasn't quite ready, asking for an email.

People clicked through to the pricing. Some clicked the paid plans. That was the signal — demand, and willingness to pay, confirmed before he wrote the app. It's the leanest validation in this list, and a reminder that the first evidence of fit can come before the product does.

7. Instagram: the pivot that turned a cluttered app into fit

Instagram didn't start as Instagram. It started as Burbn, a cluttered check-in app with photos, plans, and points bolted together. Kevin Systrom and Mike Krieger looked at how people actually used it and found one feature carrying everything: photo sharing.

So they stripped the rest away and rebuilt around photos, filters, and instant sharing. The relaunch in October 2010 pulled in roughly 25,000 users on day one. Same founders, same core technology — but the focused version fit a market the bloated one never could. Sometimes finding fit is mostly an act of deletion.

8. Notion: near-death, then word-of-mouth pull

Notion almost didn't make it. The team scrapped an early version, nearly ran out of money, and rebuilt the product close to scratch. When the reworked version landed, something shifted: users didn't just adopt it, they evangelized it — templates, tutorials, and a passionate community spreading it for free.

That organic, compounding pull — growth the company wasn't paying for — carried Notion past a million users and beyond. It's the market-pull signal in its purest form: when you have fit, your users start doing your marketing.

What every example has in common

Different products, different eras, different tactics — a survey here, a demo video there, a brutal pivot, a camera in a stranger's apartment. But strip it back and the pattern is identical: none of them waited to "just feel it." They found a way to turn "do people want this?" into evidence they could read and act on.

For some it was demand before the product (Dropbox, Buffer). For others it was behavior they watched closely (Airbnb, Instagram, Notion). And for the clearest cases — Superhuman, Slack, Baremetrics — it was a single survey score they could measure, mine for direction, and track over time.

That last one is the most copyable, because you can run it on your own users this week and get a number, not a hunch.

How to write your own example

1. Run the survey. Ask engaged users how they'd feel without your product and calculate the "very disappointed" percentage. The step-by-step is here.
2. Mine the open-ends. Like Slack and Baremetrics, the written answers tell you your ICP and your roadmap, not just your score.
3. Act on the gap. Below 40%? Do what Superhuman did — deepen what fans love, convert the fence-sitters. Here's the improvement loop.
4. Track the trend. Fit is a moving number. Re-run it and watch which way it goes.

The companies in this list weren't luckier than you. They were just more deliberate about measuring whether the market wanted what they'd built. You can start being that deliberate today.