Product/Market Fit
A market that wants the product pulls it out of you. Everything before that is pushing.
The Model
Product/market fit is not a score. It is a change of regime, and the two regimes behave so differently that you can identify which one you are in without measuring anything.
Before it, growth is proportional to effort. You push, something moves; you stop pushing, it stops moving.
Every unit of revenue has a unit of exertion attached to it, and the ratio does not improve.
After it, demand arrives faster than you can serve it. Usage grows in weeks you ran no campaign, and support queues.
People buy through a bad onboarding, forgive a missing feature, and tell somebody else.
Rachleff's underlying claim, taken from Valentine, was that the market is the dominant variable. A strong team in a weak market loses. A strong market pulls an average product forward and makes the team look better than it is.
Which reorders the founder's job. The question is not how do I make this product better. It is whether this is a market that pulls.
Why operators get this wrong
Founders diagnose fit using metrics that respond to spend, which means the diagnosis is circular.
Here is the shape it takes. Revenue is growing fifteen percent a month, and acquisition spend is growing eighteen.
Logo churn is six percent a month, so the entire customer base turns over inside eighteen months.
That is an acquisition treadmill wearing the clothes of fit. The growth is real and it is entirely purchased, and the moment spend flattens, revenue flattens with it. Everyone in the company will describe this as traction, because the top-line chart is genuinely up and to the right.
The second failure is where you look for the signal. Founders study the customers who left, because leaving feels like the thing to explain, and then build features for people who have already decided.
The signal lives in the cohort that renews without being asked. That cohort is almost always narrower than the founder wants it to be — one segment, one use case, one type of buyer.
Narrowing to it feels like shrinking the business, which is precisely why it does not happen, and why so many companies stay eighteen months from fit indefinitely.
Applied
Body
The training equivalent is the programme you keep doing without negotiating with yourself every morning.
Most people push a programme. Willpower goes in, compliance comes out, and the arrangement holds until the first genuinely bad week at work — at which point the programme is revealed as something that was being sustained rather than something that was working.
When the fit is right, the shape changes. The movements are ones you do not dread, the time slot survives contact with your actual calendar, and the load is one your joints tolerate without a negotiation. You find yourself training on the bad weeks too, not because discipline improved but because the friction is low enough that the bad week does not break it.
The diagnostic is the same one that applies to a business. If adherence requires motivation every single time, the thing does not fit. Changing the programme beats increasing the discipline, and the reverse almost never works for long.
Business
Measure pull, not push. Three instruments, all cheap.
What share of new customers arrived with no paid touch at all. Whether cohort revenue retention exceeds one hundred percent twelve months in. And whether more than forty percent of users say they would be very disappointed if the product disappeared — the question Sean Ellis put into circulation, and still the fastest read available.
None of those can be bought, which is exactly why they are worth watching.
Then the harder discipline, which is what to do before the instruments read positive: refuse to scale.
Hiring salespeople, raising money and buying traffic all amplify whatever is currently true. If what is currently true is that the market is indifferent, amplification does not fix it — it makes the discovery slower, more expensive, and more painful to act on, because by then there is a team whose jobs depend on the current answer.
The companies that die of this rarely die of a bad product. They die of scaling a good product into a market that was never pulling.
AI Leverage
Building got cheap. The constraint moved from production to demand, decisively, in about two years.
Anyone can ship the product now. The scarce thing is knowing in advance that somebody will pull it, and that has not become cheaper at all.
So point the tools at the demand side first, where they are underused.
Synthesise two hundred support tickets into the five recurring jobs customers are actually hiring you for. Cluster the free-text reasons on cancellations, which nobody reads because there are four hundred of them.
Draft twelve positioning statements and run them as ad copy for three hundred euros before a line of code is written.
That last one is the highest-value use available to an early company and it takes an afternoon.
The failure mode of this decade will not be teams that cannot build. It will be teams that build eleven things nobody asked for, because building stopped hurting and nothing replaced the pain as a filter.
The Drill
Email the last twenty customers who stayed past ninety days. One question, from your own address, sent twenty separate times.
What were you doing before this, and what would you use if we disappeared tomorrow.
Do not use a survey tool. A survey gets you a rating; a reply from a person you emailed personally gets you their sentence, and their sentence is the deliverable.
Roughly ten will answer. Read all ten in one sitting and look for the job that appears in the same words more than three times.
That sentence is your positioning, and you did not write it. If no common sentence emerges from ten replies — if every answer describes a different job — you have your answer about fit, and it is more useful than the answer you were hoping for.
Twenty emails is under an hour of work. Send them this week, before the next roadmap meeting rather than after it.
Stoic parallel
Marcus wrote that what stands in the way becomes the way. The obstacle is not an interruption of the work; it is information the work required, arriving in a form nobody enjoys.
A market saying no is exactly that — the highest-quality information available about the market, delivered free.
Most founders spend two years explaining why it does not count. Too early. Wrong channel, and the buyers do not understand it yet.
Epictetus put the discipline underneath it: do not demand that things happen as you wish, but wish them to happen as they do.
Later readers gave the idea a Latin name, amor fati, and made it sound romantic. It is not romantic — it is a rule about assent.
An accurate impression has arrived. Your only real choice is whether to agree with it.
The founder who cannot hear no about this product cannot find the thing the market would say yes to, because both answers come down the same channel and he has stopped listening on it.
One model per week.
Applied to training, business, and AI leverage. No fluff.
Related models
- Arguing from First PrinciplesReduce the problem to what is irreducibly true, then rebuild without borrowing anyone else's conclusions.
- Second-Order ThinkingAnd then what? The first consequence is obvious and priced in. The second one is where the money is.
- Pareto PrincipleResults are not spread evenly across inputs. Measure the distribution before you decide where the effort goes.
Origin: Andy Rachleff, who coined the term from Don Valentine's investing thesis at Sequoia; popularised by Marc Andreessen's 2007 essay