Product-Market Fit

The belief that a good product sells itself held up for a long time. So when something didn't sell, the cause got assigned to one of two things: the features were thin, or not enough people had heard about it. Both diagnoses were comfortable because both came with obvious work attached. Build more, or buy more ads.

The trouble was that companies kept appearing that applied both prescriptions and stayed exactly where they were. Feature lists longer than the competition's, ad spend climbing every month, and yet the day the ads stop the signups stop, while the people who already came through disappear within a month. Neither of the two diagnoses explains that state, because the work being done is pouring more water into a leaking bucket. Both prescriptions also hide the problem while they run: spend rises, the top-line signup chart rises with it, and the only number that would have shown the leak is the one nobody is plotting.

Pagedeck went through that stretch in its second year. Ravi pushed ad spend to eight million won a month and weekly signups crossed 300, and the meeting room had a chart climbing to the upper right. Four weeks after each of those weeks, though, around a dozen of the 300 were still around. What Mia brought to that month's review wasn't a feature request list, it was a question: she couldn't tell whether the company was collecting people or keeping them. Grace suggested turning the ads off for two weeks. With them off, weekly signups fell from 300 to 11 — and those 11 were people nobody had called.

Product-market fit (PMF) is the name given to that distinction. An investor wrote it up in 2007 and the term settled into the industry from there, though the use spread faster than the definition did. Teams reached for it as a way to check which side they were standing on before deciding what to do next, because in a state that only moves when it's pushed, hiring more salespeople and buying more ads both work purely to enlarge the loss.

This is also why the term sits on the strategy branch when the parent field splits up. Strategy is where a company decides which direction to go: setting a distant goal, narrowing the group of customers to go after, finding the gap in competing products. None of those judgments can verify itself. Whether the target market was chosen well only shows up in whether people from that market actually stay. PMF is the position from which those earlier choices get checked after the fact.

The terms under this one form two branches: what you check the fit with (measurement), and how life differs before and after fit (stage). The first three are the instruments, and the last two are the two worlds those instruments point at.

Measurement holds three, split between asking and watching.

1. Sean Ellis test — a survey that asks users directly. It asks how they would feel if they could no longer use the product, with three choices: very disappointed, somewhat disappointed, not disappointed. The convention is that fit exists when more than 40 percent pick "very disappointed." It's the fastest instrument because it can run in a few days, but it has a built-in limit: the people answering are the ones who stayed. Nobody sends the survey to the people who already left.
2. Retention curve — takes everyone who signed up in the same week as one group and plots what percentage of them is still around each following week. Start with 100 and land on 45 in week one, 30 in week two, and 24 in week four, and the curve drops steeply before easing off. It watches behavior rather than statements, which makes it more trustworthy than a survey, but the answer takes months to arrive.
3. Curve flattening — the point where that curve stops falling and lies sideways. This is the real signal. Flattening at 24 percent means a quarter of the people who came in found a place for themselves in the product, and it supports the arithmetic that the same share will hold for people brought in later. A curve that keeps falling to zero means nobody is being held, so no matter how much traffic gets added, total users stall at some ceiling.

The three get read together because each one covers for the others' blind spots. A good survey result with a curve that never flattens means a small group of survivors is enthusiastic. A flattening curve with a poor survey result can mean people are stuck for lack of an alternative. Either way, moving to the next stage on the strength of one instrument gets expensive later. If "very disappointed" comes back at 52 percent while week-four retention sits at 6 percent, the people who answered weren't the market — they were the handful who made it to the end. That 52 is measuring the temperature of a dozen devoted users, not the size of a market. Running all three costs little, since the survey takes days and the other two come out of data the product already records.

Stage is the branch that splits what a company should prioritize on either side of that line. There are two, and their work runs almost opposite.

1. Pre-PMF — demand hasn't been found yet, so searching comes first. The work here isn't telling more people about what exists; it's finding out why the few who stayed did stay. That's why ten interviews are worth more than eight million won of advertising. Growing a sales team or scaling ad spend at this stage leaves nothing but fixed cost behind, because when the product changes, the pitch that team was selling gets thrown out with it.
2. Post-PMF — demand is confirmed, so scale and efficiency come first. The question shifts from what to build to how to bring people in faster and cheaper. Hiring grows, each channel starts getting measured on cost, and reliability and speed take over the agenda from feature requests. Once the math says it costs 30,000 won to acquire someone who stays and pays 180,000 won, the next question isn't what to build but how much can be poured in before that math breaks.

The two aren't separated by a single threshold. Usually it's a stretch spanning several quarters, and crossing it once doesn't make it permanent. Move from selling to twenty-person teams to selling to 300-person companies and the requirements change enough to put the company back in the earlier stage. Permission management, access logs, and security review — things nobody at a twenty-person team ever asked about — come up before the contract at a 300-person company, and without those three the deal stops regardless of the retention already earned. That situation is exactly what the word chasm points at.

The two branches run in one direction. Measurement tells you the position, position sets the stage, and the stage decides where money and people go next quarter. Misread the instrument and the two steps after it are both wrong. A company still in the earlier stage that judges itself to be in the later one and hires five salespeople is left, six months on, with higher payroll and a product that doesn't sell. The reverse error is quieter but real too: a company that has fit and keeps treating every quarter as a search will spend a year rewriting things while a competitor takes the market it already proved.

"Let's hold the two extra sales hires until we've confirmed PMF."

Hiring now means putting people who sell in place before what they're selling is settled.

"Very disappointed came back at 31 percent. We're still below the line."

The survey missed the 40 percent benchmark, so expansion plans should wait.

"The curve stops dropping at week five. It flattens at 22 percent."

Roughly a fifth of arrivals found a place, so adding traffic now adds to the total.

The two weeks Pagedeck ran with the ads off were the turning point. Mia interviewed all 11 remaining users one at a time, and nine of them said the same thing: they used it for meeting notes, and no other tool let several people edit during the meeting itself. What Pagedeck had been putting on the front page until then was "all your documents in one place." The people who stayed had not come for that.

For the next six months Mia set the feature request list aside and worked only on the meeting-notes path. Ravi kept the ads off and wrote in three communities full of meeting-heavy organizations. Only after week-four retention climbed from 8 percent to 26 percent and held there for three consecutive months did Grace reopen the hiring plan. The screen that got the most time in the Series A diligence was this curve, not the revenue chart.

The most common reading treats it as a gate you pass once. The word "achieved" travels with it and there's a benchmark like 40 percent attached, so it looks like an exam. But the state describes whether a product and a market are engaged, and it comes apart when either side moves. A competitor solves the same problem more cheaply, the company shifts to selling to larger organizations, or the way people work changes — and the same product in the same place watches its numbers fall. A team that believed it passed and put the instruments away finds out after revenue bends. By then the headcount and costs added on the strength of that belief are attached to the problem, which makes reversing it expensive. Teams that keep the curve on a dashboard see the same change a quarter or two earlier, while it is still a question about the product rather than a question about payroll.

Reading rising revenue as proof of fit goes wrong just as often. Money arriving looks like the least ambiguous evidence available, so the reading is easy to reach. But revenue can be produced by the product or produced by people pushing. If a rep spends twenty hours on one deal and that customer doesn't renew a year later, the revenue demands the same labor again next year. The way to tell them apart is to look at what happens after the sale: what share arrived on their own, how many are still there a year later, and whether the ones who stayed are adding seats. That separates revenue that was pushed from revenue that came from fit. It's also why a figure like the quick ratio, which divides what was gained by what was lost, gets used at all. Identical revenue growth with a large leak out the back isn't growth, it's churn.