Post-PMF

Company stages used to be named by headcount and funding round alone. "Just raised a Series A, 45 people" tells you the size of the company and nothing about what it should be doing this quarter. Two companies of exactly that description can owe their next six months to opposite kinds of work.

So the same argument kept going in circles. The CEO wants to hire five salespeople with the money that just landed; the product side says there is nobody sticking around yet. Both sides have evidence, and they are measuring different things. This is where six months end with nothing but a larger payroll to show for it. Build a sales organization while what you sell is still moving, and when the product changes, the pitch that organization learned gets thrown out with it.

Post-PMF exists to name that dividing line in one word, so a team can say which question it is currently allowed to be working on. The stretch where demand hasn't been found and the job is still searching is pre-PMF. The stretch where demand is confirmed and the job is scale and efficiency is this one. They aren't two steps of the same activity — the work in each is close to opposite, which is why calling the stage wrong is expensive rather than merely inaccurate.

The clearest marker is that the question changes. Meetings stop being about what to build and start being about how to get what already exists to more people, faster and for less. Four changes show on the surface.

1. The numbers being watched change. In the earlier stretch you asked whether anyone stays; now every channel gets a price tag. Customer acquisition cost (CAC) is what one customer costs to bring in, and lifetime value (LTV) is the estimate of what that customer pays before leaving. A channel that spends $25 to earn $150 and a channel that spends $100 to earn $150 get treated differently.
2. Headcount grows and roles split. Where one person handled marketing, sales and support together, you get separate people for channels, contracts and onboarding. Going from 45 people to 70 is ordinary here.
3. Meeting agendas shift from features to operations. How many incidents happened last week and how long support took to reply now come before the question of when the next feature ships. With more people using the thing, an outage costs more, and a slow reply reaches someone who is already paying.
4. Customers get bigger and their demands change with them. Permission management, access logs and security review — none of which a twenty-person team ever asked about — arrive before the contract at a 300-person company.

The evidence that you're in this stretch comes from the retention curve rather than from a feeling. The curve plots how many of the people who signed up in the same week are still around as weeks pass, and the signal is the line settling at some height instead of continuing down. If 26 of every 100 are still there in week four and the line holds from there, you can assume the same share will hold for the people you bring in next.

Moving on to real scaling means proving, in numbers, that money going in produces output coming out. Three things have to hold.

1. One customer has to be profitable on its own. Unit economics is the profit and loss on a single customer, and the relationship holds if a customer who cost $25 to acquire stays about nine months and pays $150. If acquisition runs $150 instead, adding people only enlarges the loss.
2. One channel has to produce the same result twice. A single good month can be the month's own doing. A channel qualifies as somewhere to pour money only once doubling the budget doesn't send the cost per customer sharply up.
3. More people has to mean more output. If four salespeople can close nearly twice what two of them closed, growing the team is a defensible call. Where that proportion breaks, the problem is the way the thing gets sold, not the headcount.

This threshold isn't a line you cross once. It's usually a stretch spanning several quarters, and moving the market you sell to — from twenty-person teams up to larger companies — changes what is being asked of you and puts you back in the searching stretch. That place where deals stall regardless of the retention you built up earlier is what the word chasm points at.

"We're post-PMF now, so let's put this quarter into channels rather than features."

Demand is settled, so the team's hours should move to scale.

"Calling us post-PMF is a stretch. The curve hasn't flattened yet."

The share of users who stay has not yet settled at any level.

"Now that we're post-PMF, this week's review starts with the incident count."

An announcement that operations, not new features, is the first agenda item.

At Pagedeck the call landed once paid upgrades came in the same shape two quarters running. Mia brought the graph showing week-four retention flattening at 26 percent, and Ravi attached a table showing the cost of a free signup running anywhere from $25 to $90 depending on the channel.

Grace decided there to put the next quarter's budget behind the $25 channel and to hold new feature requests for a quarter while response times and permission management got attention first. What made that possible was having the flattened curve as evidence first.

The most common mistake is reading it as a state you keep once you reach it. The language of crossing a threshold makes going backward sound impossible. But the label attaches to a pairing of product and market, not to the product by itself. Decide to sell a product built for small teams to 300-person companies and the pairing changes, which means the checking starts over. Miss that and you staff up sales the way the earlier market rewarded, then watch every deal stop at a pre-contract security review that nobody on the team is equipped to answer.

The other one is hearing it as "product work is done, marketing from here." Agendas really do move from features to channels, so it looks that way. What changes isn't how much gets built but what gets built. Architecture that holds when usage goes up tenfold, a screen where an admin hands out permissions, access logs an auditor can read — things nobody requested in the earlier stretch — move to the top of the list. Bring in people without that switch and the new users leak straight back out, pulling down the height the curve used to flatten at.