How do you find Product–Market Fit?
War stories, hard metrics (including the “40% rule”), and a practical field guide.
Marc Andreessen’s famous line sets the tone: “The only thing that matters is getting to product/market fit.” He defined PMF as “being in a good market with a product that can satisfy that market,” and noted you can feel when it’s not happening-word of mouth stalls, usage crawls, sales cycles drag. Conversely, when it’s working, customers buy (or use) as fast as you can serve them. (Pmarchive)
This post distills how founders actually get there: the signals that PMF is near, the loops that push you through the fog, and the gritty stories (Slack, Airbnb, Dropbox, Superhuman) that show what it looks like in practice.
First principles: what PMF is (and isn’t)
PMF is value × market, not hype. As Andreessen (and Andy Rachleff, who coined the term) argued: great teams can still fail in bad markets, and mediocre products can wiggle through in great ones-but the repeatable success path is pairing a real market with a product that truly satisfies it. (Andreessen Horowitz)
PMF is felt in behavior, not vibes. Y Combinator’s motto-“Make something people want”-sounds simple; it’s also the best north star when everything is messy. Sam Altman and Paul Buchheit’s corollary: in the beginning, it’s better to have a small number of users who love you than many who kind of like you. (Paul Graham)
PMF is not one metric. It’s a tapestry of leading and lagging indicators: how much users would miss you, whether cohorts stick, how fast people tell friends, and-especially in B2B-whether expansion outpaces churn.
The scoreboard: reliable signals that PMF is close
The “very disappointed” (40%) test
Sean Ellis popularized a deceptively simple survey: ask active users, “How would you feel if you could no longer use this product?” If ≥40% say “very disappointed,” you likely have initial PMF. Rahul Vohra (Superhuman) operationalized this into a full system, and Hiten Shah’s 2015 open study of Slack users found 51% “very disappointed” at ~500k paying users-strong external proof of PMF. (Venture Hacks)Retention curves that flatten (and don’t drop to zero)
For nearly every software product, a cohort retention curve that levels off-instead of bleeding out-signals durable value. Amplitude calls retention the “ultimate product strategy,” and PostHog’s founder guide is blunt: “If your retention curve flattens… it’s a strong sign you have product-market fit.” (Amplitude)Power usage and habit formation
For consumer and prosumer apps, Andrew Chen’s “magic metrics” include cohort curves that flatten, DAU/MAU > 50% for daily-habit products, and a power-user curve with a visible engaged core. These are symptoms of PMF, not causes-but they’re consistent across winners. (LinkedIn)B2B expansion and net revenue retention (NRR)
In B2B, PMF shows up as teams adding seats and features without you begging. A useful rule-of-thumb from growth leaders aggregated by Lenny Rachitsky and Casey Winters: bottom‑up SaaS NRR ~100% is good; ~120% is great; enterprise NRR ~110% good; ~130% great. (Revenue.fyi)Fast, compounding word‑of‑mouth
Andreessen’s “you can feel it” list included press and customers pulling on you. Chen layers math on top: the viral factor matters (how many users each user brings) but is only one piece-still, channels like referrals and WOM comprising a big chunk of new users are common in PMF stories. (Andrew Chen)
War stories: what PMF looked like up close
Slack: a usage threshold that predicted near‑certain retention
Inside Slack’s early data, a “magic number” emerged: when a team had sent 2,000 messages in its history, 93% of those teams were still active. Stewart Butterfield: “After 2,000 messages, [they’ve] really tried Slack.” This wasn’t vanity; it told the team what activation to design for. (First Round)
Slack’s internal storytelling matched the metrics. In the now‑famous “We Don’t Sell Saddles Here” memo (2013), Butterfield reframed Slack around making work life simpler, more pleasant, more productive-a story the market could feel. (Medium)
Takeaway: instrument your own “activation threshold” and orient onboarding toward it. Slack optimized everything-from invites to notifications-to get teams to 2,000 messages fast. (GrowthHackers)
Airbnb: do things that don’t scale to unlock the must‑have
In 2009, with revenue stuck at $200/week, the founders grabbed a camera, flew to New York, and re‑shot listings. A week later, revenue doubled. That small, hands‑on fix clarified the value proposition and sparked a systematic professional photography program that improved bookings and trust at scale. (First Round)
Takeaway: PMF often hides behind an unscalable insight. Remove the biggest friction (in this case, poor photos) and the market’s real appetite surfaces. (First Round)
Dropbox: demonstrate the magic, then build
Before there was a working product, the team released a 3‑minute demo video, packed with in‑jokes for early adopters. The result: the waitlist jumped from 5,000 to 75,000 overnight. Demand was real; the product still had to catch up. (TechCrunch)
Takeaway: for deep tech, show the core value early to validate appetite-and then race to meet it.
Superhuman: turning a survey into a roadmap
Superhuman’s PMF system starts with the Ellis survey, then segments by persona to find the High‑Expectation Customer (HXC)-a term from Julie Supan’s brand work with Airbnb, Dropbox, and Thumbtack: “the most discerning person within your target demographic.” Superhuman focused on the HXCs’ words to prioritize the roadmap and systematically raised its PMF score. (First Round)
Takeaway: identify the small group who love you (or could), build for them, and let their language shape your positioning and features. (First Round)
A practical playbook: loops that push you toward PMF
1) Narrow the market (and name your HXC)
Write a crisp one‑liner for who you’re for and what pain you solve (not your feature list). Supan’s HXC definition is a great forcing function-this is the persona whose endorsement changes everyone else’s mind. If you can’t name yours, you’re not close. (First Round)
Tactic: Run 15–30 discovery calls with only prospective HXCs. Ask for stories about the last time the pain showed up. Ship a prototype or a workflow mock and listen for the “Where has this been all my life?” reaction-Steve Blank calls it the “pupil‑dilation” test. (Lenny’s Newsletter)
2) Define the activation moment that predicts habit
Find the behavioral milestone that separates dabblers from devoted users (Slack’s 2,000 messages; for your app it might be shared a doc with 3 teammates, took a ride in a new city, etc.). Instrument-then design onboarding to get people there fast. (First Round)
3) Measure the two PMF pillars: retention and love
Retention: Plot cohort retention and keep iterating until the curve flattens-that is, a meaningful % of users still active months later. (For many consumer products, Andrew Chen’s heuristics-flattening cohorts, strong DAU/MAU-are practical guardrails.) (PostHog)
Love: Run the 40% PMF survey on active users monthly. Segment results by persona and use verbatim feedback from “very disappointed” respondents to draft your next sprint. (It’s how Superhuman turned a metric into a roadmap.) (Venture Hacks)
4) Chase one distribution loop that amplifies pull
Don’t bolt on “growth hacks” before PMF; as Andrew Chen says, growth follows fit. That said, once users love you, lean into the loop that compounds-referrals, integrations, content SEO, or sales‑assisted pilots-so that more of your new users come from WOM and referrals over time. (Andrew Chen)
5) Keep your runway for iteration, not theatrics
YC’s “make something people want” pairs with another Graham-ism: stay cheap long enough to iterate. Most startups die before PMF because they run out of time. Keep cycles short, experiments honest, and your definition of done anchored in user behavior, not pitch‑deck milestones. (Paul Graham)
Benchmarks & heuristics (so you don’t optimize noise)
Ellis/40% rule: ≥40% “very disappointed” ⇒ strong early PMF. Slack’s 51% at scale is a useful reality check-even beloved products rarely score 70–80%. (First Round)
Consumer habit signals: DAU/MAU > 50% for daily‑use apps; cohorts flattening; power‑user curve with a heavy engaged tail. (LinkedIn)
B2B health: NRR ~100% good / ~120% great (bottom‑up), ~110% / ~130% (enterprise). Pair with logo retention and seat expansion. (Revenue.fyi)
WOM share: expect an increasing fraction of signups from referrals/WOM as fit improves (e.g., Chen’s marketplace case studies). (Lenny’s Newsletter)
Treat these as sanity checks, not dogma. For instance, during the pandemic, Hiten Shah’s survey found Zoom scored ~30% “very disappointed”-below the Ellis bar-despite obvious category growth. Surveys are snapshots; triangulate with retention and expansion. (Nira)
Common pitfalls on the road to PMF
Counting signups, not sticking users. Until retention flattens for a meaningful cohort, you don’t have PMF-no matter how pretty the acquisition graph looks. (PostHog)
Optimizing average feedback. Your HXC is not “average.” Build for the most discerning users who feel your product’s core benefit most strongly, then generalize. (First Round)
Treating velocity as victory. Shipping fast without learning fast just burns runway. The Slack, Airbnb, and Dropbox stories all hinged on specific insights (messages sent; photography; a demo video) that unlocked the curve. (First Round)
Buying growth before you’ve earned it. Chen’s admonition bears repeating: “Startups need product/market fit, not growth.” Paid acquisition can expose more people to a non‑fit-it can’t create one. (Andrew Chen)
A 30–60–90 day PMF plan you can actually run
Days 1–30: Define & discover
Write your HXC spec (who, where they hang out, exact pain, current workaround). Steal their words from interviews to update your landing page and onboarding. (First Round)
Instrument a candidate activation event (your equivalent of Slack’s 2,000 messages). Make sure you can seecohorts hitting it. (First Round)
Days 31–60: Prototype & prove
Ship two changes per week that specifically move activation and early retention (e.g., pre‑filled templates, better default notifications, a single‑player “sandbox” to demo value).
Launch your first PMF survey to active users; segment by persona; build the next sprint from the “very disappointed” group’s words. (Venture Hacks)
Days 61–90: Consolidate & concentrate
Run a cohort analysis. Do later cohorts flatten higher? If yes, keep pushing the same loops. If not, decide: move upmarket/downmarket, or pivot the wedge. (PostHog)
Pick one distribution loop to amplify genuine pull (referrals, integrations, or founder‑led sales to look‑alike accounts). Keep spending small until you see retention. (Andrew Chen)
Closing thought
You’ll know you’re close when two things happen at once: users come back on their own, and more new users come from existing ones. At that moment, all the slogans make sense: make something people want; make a few people love you; then broaden the circle. The stories are remarkably consistent-from Slack’s 2,000‑message threshold to Airbnb’s scrappy photography sprints, to Dropbox’s “show, don’t tell” video. Small, focused moves revealed a market that was already waiting. (First Round)