In the summer of 2017, Rahul Vohra had a problem he couldn't name. He'd spent two years building Superhuman, an email client designed to be the fastest on earth, and everything looked right. Users loved the product. The team was talented. Investors were interested. But Vohra couldn't shake the feeling that something was incomplete. He had no framework that could tell a pre-launch startup whether it had crossed the invisible line between "promising" and "pull."
He found a single question, designed by growth strategist Sean Ellis: "How would you feel if you could no longer use this product?" Ellis had benchmarked nearly a hundred startups and found that if 40 percent or more answered "very disappointed," the company had product-market fit. Below 40, it didn't. Vohra surveyed his users. Twenty-two percent.
The number confirmed what his nervous system had been signaling for months.
Product-market fit is the moment when your market starts pulling the product out of your hands. But the signal arrives in your body before it arrives in your data. Marc Andreessen, who popularized the term in his 2007 essay "The Only Thing That Matters," described it in almost physical terms: "You can always feel when product-market fit isn't happening." And you can feel when it is. The phone rings. The inbox fills. Customers start telling other customers. The feeling is real, and it precedes the metric by weeks or months. The question isn't whether your intuition matters. It's whether you can trust it, sharpen it, and build a system around it before it misleads you.
Why Does Your Body Know Before Your Dashboard Does?
In 1994, a neuroscientist named Antoine Bechara walked into Antonio Damasio's lab at the University of Iowa with a card game that would change how science understood decision-making. The game was simple. Four decks of cards sat face down on a table. Participants flipped cards one at a time from any deck they chose. Each card delivered a gain or a loss. Two of the decks were rigged to be profitable over time. Two were traps, offering large immediate gains but larger losses that would quietly bankrupt you.
Bechara wired the participants with skin conductance sensors, measuring the electrical activity produced by their sweat glands, a reliable proxy for emotional arousal. Then he watched.
After about ten cards, something remarkable happened. The participants' palms began to sweat before they reached for the bad decks. Not after. Before. Their bodies had detected the pattern and were generating a warning signal, a faint spike of arousal that said "danger," even though the participants couldn't yet explain which decks were dangerous or why. When researchers asked them directly, they had no conscious hunch. They couldn't articulate a strategy. But their skin conductance told a different story. The body had learned the game roughly fifty cards before the conscious mind figured it out.
Bechara and Damasio called this the somatic marker hypothesis. The idea, published across a series of papers through the late 1990s, was that the body generates emotional markers that tag experiences as good or bad, and that these markers influence decisions before conscious reasoning engages. The ventromedial prefrontal cortex, a region just behind the bridge of your nose, integrates these body-based signals with memory and context. The amygdala encodes the emotional valence. And the anterior insula translates the whole package into what you experience as a gut feeling.
This isn't mysticism dressed in lab coats. The Iowa Gambling Task has been replicated hundreds of times across decades. The finding holds: the body detects statistical patterns in uncertain environments before the conscious mind can articulate them. Participants move through three distinct stages. First, the prehunch period, where skin conductance shifts but the person has no awareness of the pattern. Then the hunch period, where they start to sense something but can't explain it. Then, finally, the conceptual period, where they can describe the strategy in words. Most of the learning happens in stage one. The body figures it out first, and it communicates through feeling, not language.
For founders, this matters more than it sounds like it should. Product-market fit generates hundreds of micro-signals: the tone of a customer's email, the speed of a reply, the specific words a user chooses when describing your product to someone else, the way a demo conversation shifts from polite questions to urgent ones. No dashboard captures these. No analytics platform tags the moment a customer's voice changes from curious to committed. But your brain does. The same neural architecture that helped Bechara's participants avoid bad decks is running in the background every time you talk to a customer, read a support ticket, or watch someone use your product for the first time. The signal is real. The question is whether you've learned to listen to it.
The 40 Percent Line
Sean Ellis didn't set out to invent a test for product-market fit. He was trying to solve a more practical problem: how do you know, before you pour money into growth, whether a product is ready to grow?
Ellis had worked on early growth at Dropbox, where a three-minute explainer video had turned a waiting list of five thousand people into seventy-five thousand overnight. He'd been on the founding team at LogMeIn, where the product had taken much longer to find traction. The difference between the two had nothing to do with growth tactics. It came down to whether the product had something worth growing. He started asking the "very disappointed" question to every startup he advised, and the pattern emerged quickly. Companies below 40 percent could spend money on acquisition, but the users would leak out the bottom. Companies above 40 percent couldn't stop growing even when they tried.
The elegance of the question is that it measures something no other metric captures cleanly: emotional dependency. Not satisfaction, which is polite and easy to inflate. Not Net Promoter Score, which asks about future behavior that may never happen. The Ellis question asks about loss. It measures the size of the hole your product would leave. And loss, as Daniel Kahneman demonstrated across decades of research, is the emotion humans feel most intensely. We experience losses roughly twice as powerfully as equivalent gains. A product that 40 percent of users would be very disappointed to lose has created something closer to a need than a preference.
Vohra took the 22 percent score and built an engine around it. First, he segmented. Instead of treating all users as a single population, he identified which user personas appeared most often in the "very disappointed" group. Founders, executives, people in business development. These people lived in their inboxes, sent dozens of emails a day, and cared about speed the way a surgeon cares about a scalpel. When he filtered for only these users, the score jumped from 22 to 33 percent. The product didn't change. The audience narrowed.
Then he profiled. He created a fictional character called Nicole, a composite of his most passionate users. Not average, not a blend. A specific person with specific habits who would be devastated without Superhuman. She read a hundred to two hundred emails a day. She sent fifteen to forty. She wanted inbox zero and she wanted it fast.
Then he built a roadmap split in half. One half doubled down on what the very disappointed users already loved: speed below fifty milliseconds, keyboard shortcuts for everything, design details like automatic conversion of "-->" into an arrow. The other half attacked the barriers that kept "somewhat disappointed" users from tipping over: no mobile app, missing integrations, weak search. He ignored feedback from "not disappointed" users entirely. Their requests were noise.
Three quarters later, the score hit 58 percent. Superhuman hadn't pivoted. It hadn't added a killer feature. It had gotten more specific about who it was for and more relentless about serving those people. The metric tracked the feeling, and the feeling tracked the reality.
On a napkin, the Ellis test works like this: if fewer than four in ten users would be gutted to lose your product, you don't have product-market fit. You have product-market proximity.
The False Positive Problem
The most dangerous version of product-market fit is the one that isn't real.
In April 2020, Jeffrey Katzenberg launched Quibi with $1.75 billion in funding, partnerships with every major Hollywood studio, and a Super Bowl commercial. The premise: premium short-form video designed for mobile, episodes under ten minutes, consumed during commutes and lunch breaks. Katzenberg had co-founded DreamWorks. His co-founder, Meg Whitman, had run eBay and Hewlett-Packard. The investors included every major entertainment and technology company in the world.
Quibi had every signal that looks like product-market fit if you squint. Celebrity talent. Massive funding. Press coverage. Brand partnerships. Industry validation from people whose judgment had been right many times before. What it did not have was a single piece of evidence that anyone wanted to watch short premium video on their phone during a commute. The company had raised $1.75 billion before testing the assumption that mattered most.
The first month brought 3.5 million downloads. By the time the service shut down, six months after launch, only about 500,000 people were paying subscribers. The gap between those two numbers is the anatomy of a false positive. Downloads are curiosity. Subscriptions are dependency. Quibi had the former and none of the latter.
Katzenberg and Whitman later acknowledged that "the idea itself wasn't strong enough to justify a standalone streaming service." But the deeper problem was cognitive, not strategic. The company suffered from what psychologists call the false consensus effect, the tendency to assume that other people share your preferences, beliefs, and behaviors. When you are a billionaire entertainment executive who consumes content constantly and believes short-form premium video is the future, it is genuinely difficult to perceive that the market might not agree. Every person in the room nodded. Every investor wrote a check. The consensus felt like evidence. It was an echo.
Juicero committed the same error with different props. Doug Evans raised $120 million to build a $400 Wi-Fi-connected juicer that pressed proprietary produce packets into juice. The machine was a marvel of engineering, containing custom parts, a press mechanism powerful enough to lift two Teslas, and a scanner that read QR codes on the packets to verify freshness. In April 2017, Bloomberg reporters discovered that you could squeeze the packets by hand and get the same juice. The product wasn't solving a problem. It was engineering a solution for a need that didn't exist, then interpreting investor enthusiasm as market validation. Five months later, Juicero shut down.
The pattern is consistent. False product-market fit looks like traction from the outside. Funding, press, downloads, celebrity endorsements. But none of these are the signal. The signal is what happens when you try to take the product away. Would Quibi's users have been very disappointed? The churn numbers answered that before anyone thought to ask.
What Happens When the Signal Is Real and You Almost Kill It?
The opposite error is equally instructive. Some founders hold product-market fit in their hands and nearly throw it away because it doesn't look the way they expected.
In 2010, Kevin Systrom was running Burbn, a location-based social app that let users check in, share plans, and post photos. It had some traction but couldn't differentiate itself in a crowded field of check-in apps. Systrom and his co-founder Mike Krieger noticed something in the data that didn't match their strategy. People weren't using Burbn to check in. They were using it to post photos. The check-in features, the ones Systrom had built the company around, were being ignored. The photo-sharing feature, a secondary add-on, was generating all the engagement.
Systrom did something that most founders find almost physically painful. He stripped out every feature except photos. Not reduced. Stripped. He killed his own product to follow the signal. His girlfriend told him she wouldn't share photos because hers didn't look good enough, so he built the first filter, X-Pro II, that afternoon. The rebuilt app launched as Instagram. Within two months it had a million users. Within a year, ten million. Eighteen months after launch, Facebook bought it for a billion dollars.
The signal had been there the whole time. It just didn't look like what Systrom expected product-market fit to look like. He expected check-ins. The market wanted photos. His willingness to listen to the data instead of his plan is the reason Instagram exists.
Stewart Butterfield had an almost identical experience. His company, Tiny Speck, was building a multiplayer online game called Glitch. The game was struggling. But the internal communication tool the team had built to coordinate development was generating an unexpected reaction: people outside the company kept asking to use it. Butterfield shut down Glitch and launched the communication tool as Slack. The beta drew 8,000 users in 24 hours and 15,000 within two weeks. These were companies showing up organically, asking to pay money for a product that had been a side project. Within eight months of its full public launch in February 2014, Slack was valued at a billion dollars.
Slack's internal data eventually revealed a metric that crystallized the feeling into a number: when a team had sent 2,000 messages, 93 percent of those teams never left. That was the activation threshold, the point at which the product had embedded itself deeply enough that removing it would tear a hole. But Butterfield didn't need the metric to know. The phones were ringing. The emails were arriving faster than his team could answer them. The market was pulling the product out of his hands. He could feel it.
Product-market fit is not a finish line. It is a direction your customers drag you toward, and the founder's job is to stop resisting.
Try This: The Signal-to-Noise Audit
A protocol for separating genuine product-market fit signals from vanity metrics and wishful thinking.
Step 1: Run the Ellis survey, but segment before you read the score. Survey your active users with the "very disappointed" question, but don't look at the aggregate number first. Break responses by user persona, acquisition channel, and use case. The aggregate might be 25 percent, but the founders who found you through word of mouth might be at 55 percent. That segment is your market. Everyone else is noise you've been averaging in.
Step 2: Listen to the language, not the rating. When users describe your product to someone else, write down the exact words. Not your summary of the words. The actual words. If they're using language you didn't put in your marketing copy, that language is the signal. It tells you what your product actually does for them, which may be different from what you think it does.
Step 3: Track the unprompted signals. Product-market fit generates behaviors that nobody asks for. Users who email unsolicited feature requests. Customers who tag you on social media without being asked. People who bring your product into their company without going through procurement. Make a list of every unprompted action your users have taken in the last 30 days. If the list is short, your product is satisfying but not essential. If it's long and intense, the 40 percent number will confirm what the behavior already told you.
Step 4: Measure the hole, not the happiness. Don't ask customers if they like your product. Ask what they'd do if it disappeared tomorrow. The specificity of their workaround plan tells you more than any satisfaction score. If they'd be mildly inconvenienced and switch to a competitor by Thursday, you don't have fit. If they describe a painful, cobbled-together process involving three tools and a spreadsheet, you've become infrastructure.
Step 5: Audit your roadmap for consensus-chasing. Look at your current feature roadmap and mark each item with who requested it. If most items came from "somewhat disappointed" or "not disappointed" users, you're building for the middle. Vohra ignored the "not disappointed" users entirely. The features they wanted were distractions from the features that would push "somewhat disappointed" users into the "very disappointed" camp. Build for your most passionate users. Let the middle find their way to you through the product's specificity, not its breadth.
The Pull
Marc Andreessen wrote that you can always feel product-market fit when it's happening, and you can always feel when it's not. "The customers are buying the product just as fast as you can make it," he wrote, "or usage is growing just as fast as you can add more servers." The description is physical. It's gravitational. The market reaches for the product, and the founder's job shifts from pushing to steering.
Bechara's card players felt the bad decks in their palms before they understood them in their minds. Vohra felt the gap before the Ellis score named it. Butterfield felt the pull before Slack's retention numbers confirmed it. Systrom saw the photos accumulating before he understood what they meant. The body knows. The anterior insula fires. The somatic markers tag the experience as significant, and the ventromedial prefrontal cortex integrates it with everything you've already seen, every customer conversation, every support ticket, every moment a user's face changed during a demo.
The metric matters. The 40 percent threshold is real, the segmentation is essential, and the roadmap discipline is what turns signal into traction. But the metric is a confirmation of something that already happened in your nervous system. Product-market fit is a feeling before it is a number. The founders who find it are the ones who learn to trust the feeling, test it ruthlessly, and never mistake applause for pull.
If you want to build a minimum viable product that actually tests for pull instead of just collecting feedback, the validation loop matters more than the feature set. And if you're struggling with why users nod politely but don't convert, the 9X problem explains the gap between what people say they want and what they'll actually switch for. The signal is always in the behavior, never in the survey.
Building for the right audience at the right level of specificity is what separates a product with traction from a product with traffic. The Launch System walks through the complete validation loop that tells you whether you have product-market fit or product-market proximity, from your first unique value proposition through segmented testing to the moment the market starts pulling. The blog showed you how to read the signal. The system shows you how to build toward it.
FAQ
What is product-market fit and how do you know when you have it? Product-market fit is the point at which a product satisfies a strong market demand so thoroughly that the market begins pulling the product forward. Marc Andreessen, who popularized the term in 2007, described it as "being in a good market with a product that can satisfy that market." The most reliable quantitative test is Sean Ellis's survey: ask users "How would you feel if you could no longer use this product?" If 40 percent or more answer "very disappointed," you have product-market fit. Below that threshold, the product may be liked but is not yet essential, which means growth investments will leak users faster than they acquire them.
What is the Sean Ellis 40% test for product-market fit? The Sean Ellis test asks existing users a single question: "How would you feel if you could no longer use this product?" with three response options: very disappointed, somewhat disappointed, or not disappointed. Ellis benchmarked nearly a hundred startups and found that companies where 40 percent or more of users answered "very disappointed" consistently achieved strong organic growth, while companies below that threshold struggled regardless of how much they spent on acquisition. Superhuman used this test systematically, starting at 22 percent and reaching 58 percent by segmenting users, identifying their most passionate customer profile, and splitting their roadmap between deepening what worked and removing barriers for near-converts.
Why do some startups think they have product-market fit when they don't? False product-market fit occurs when vanity metrics like downloads, press coverage, or investor enthusiasm are mistaken for genuine market pull. Quibi launched with $1.75 billion in funding and 3.5 million first-month downloads but shut down in six months because only 500,000 users converted to paying subscribers. Juicero raised $120 million before discovering its product solved no real problem. The false consensus effect, a well-documented cognitive bias, leads founders and investors to assume their own enthusiasm reflects market demand. The corrective is measuring emotional dependency rather than surface engagement: not whether users downloaded the product, but whether they'd be devastated to lose it.
How did Superhuman measure and improve product-market fit? Superhuman CEO Rahul Vohra built a four-step engine after scoring 22 percent on the Sean Ellis survey. First, he segmented users and focused only on the personas that appeared most in the "very disappointed" group, which raised the score to 33 percent without changing the product. Second, he created a detailed profile of his ideal user, a busy professional named Nicole who lived in her inbox. Third, he split the roadmap evenly between doubling down on what passionate users loved and removing barriers that kept near-converts from fully engaging. Fourth, he resurveyed continuously, tracking the score weekly and quarterly. Within three quarters, Superhuman reached 58 percent, well above the 40 percent threshold.
Can you have product-market fit and lose it? Yes. Product-market fit is not a permanent state. Markets shift, competitors emerge, and user needs evolve. Instagram had product-market fit for photo sharing but had to continuously adapt as video, Stories, and Reels redefined how people shared visual content. The Ellis survey should be run regularly, especially after major product changes or market shifts, because the 40 percent threshold can erode if the product stops evolving with its core users. The founders who sustain fit are the ones who keep listening to the signal even after the dashboard confirms it.
Works Cited
Andreessen, Marc. "The Only Thing That Matters." Pmarchive, June 25, 2007. pmarchive.com/guide_to_startups_part4.html.
Bechara, Antoine, Hanna Damasio, Daniel Tranel, and Antonio R. Damasio. "Deciding Advantageously Before Knowing the Advantageous Strategy." Science, vol. 275, no. 5304, 1997, pp. 1293–1296.
Damasio, Antonio R. Descartes' Error: Emotion, Reason, and the Human Brain. G. P. Putnam's Sons, 1994.
Ellis, Sean, and Morgan Brown. Hacking Growth: How Today's Fastest-Growing Companies Drive Breakout Success. Crown Business, 2017.
Kahneman, Daniel, and Amos Tversky. "Prospect Theory: An Analysis of Decision Under Risk." Econometrica, vol. 47, no. 2, 1979, pp. 263–292.
Vohra, Rahul. "How Superhuman Built an Engine to Find Product/Market Fit." First Round Review, 2018. review.firstround.com.
Katzenberg, Jeffrey, and Meg Whitman. Open letter to Quibi employees and investors, October 2020.
Ross, Lee, David Greene, and Pamela House. "The 'False Consensus Effect': An Egocentric Bias in Social Perception and Attribution Processes." Journal of Experimental Social Psychology, vol. 13, no. 3, 1977, pp. 279–301.