The product-market fit conversation in most early-stage startups goes one of two ways.
Version one: "We have it." Said with confidence in pitch meetings, usually based on some combination of user growth, retention metrics, and the feeling that the team is moving fast. Investors probe and the founder defends.
Version two: "We're still working toward it." Said with various levels of honesty. Sometimes accurate. Sometimes a way to avoid the harder conversation about whether the current direction is actually going to get there.
Both versions treat PMF as a binary state — something you either have or you don't. The framing is wrong. It's the wrong mental model, and it leads to the wrong decisions at the wrong moments.
Here's how I think about it now, after building Realm and watching dozens of companies work through this question.
Product-market fit is better understood as a spectrum with at least three distinct stages, each with its own diagnostic signals and its own appropriate actions. Confusing one stage for another — or treating later-stage signals as earlier-stage evidence — is how companies waste years.
Stage 1: Problem-market fit
Before you can have product-market fit, you need evidence that the problem you're solving is one that a specific market actually has. This sounds obvious. It isn't.
The signal for Stage 1 is not product usage. It's the character of conversations with potential customers. You're looking for:
- People who describe the problem spontaneously, without prompting from your framing
- Specific stories about how the problem costs them something real — time, money, opportunity
- Evidence that they've already tried to solve it and the solutions they've tried are inadequate
What Stage 1 doesn't look like: people who agree the problem sounds important, who would theoretically want a solution, who find your demos interesting. Polite engagement is not Stage 1 evidence.
The practical test: can you describe the problem in a sentence, name the specific kind of person who has it, and point to three conversations where they described it to you in their own words, unprompted? If not, you're pre-Stage 1.
Stage 2: Solution-problem fit
This is where most product-market fit conversations go wrong. Stage 2 is about whether your solution actually addresses the problem you identified in Stage 1 — at a level of quality that changes behavior.
The signal for Stage 2 is not retention alone. It's the nature of engagement with the solution. You're looking for:
- Users who complete the core workflow without guidance from your team
- Users who would describe the product to a colleague without being asked
- Evidence that the product is integrated into an actual workflow rather than evaluated as a demo
The critical diagnostic question for Stage 2: if you removed access to the product tomorrow, would users be frustrated or would they be inconvenienced? Frustrated means the product is integrated into something important. Inconvenienced means they'll find a workaround.
For Realm, the Stage 2 signal wasn't a metric. It was a support ticket. A developer had built an app using Realm, shipped it to production, and his app was now being used by tens of thousands of users. He had hit a behavior that was on the edge of what we documented. His message wasn't a complaint — it was methodical, technically precise, and clearly written by someone who needed this to work because real people were depending on it. He wasn't evaluating the product. He was operating it. That was the moment I felt we had crossed the line. Not because the number was large, but because the character of the engagement had changed. The question had gone from "is this interesting?" to "I need this to keep working." We started seeing that pattern repeatedly in the weeks that followed — developers writing in not to report bugs but to explain how they were using Realm in production and what they needed from us to keep it reliable.
Stage 3: Market fit
Stage 3 is where the growth dynamics change. If Stage 1 and Stage 2 are about finding and proving a specific user's problem-solution fit, Stage 3 is about whether that fit scales across a market — whether the market itself has a shape that supports the kind of distribution you need.
The signal for Stage 3 is the cost of acquiring the next customer relative to the cost of acquiring the first ten. If the cost drops as you grow, your distribution model is working — there's a market shape that amplifies your efforts. If the cost stays flat or increases, you may have product-market fit with a specific cohort and a distribution problem at scale.
Stage 3 is where retention metrics become central, but in a specific way: not just absolute retention, but whether cohort retention curves flatten (suggesting a stable engaged user base) or continue declining (suggesting the problem-solution fit is eroding over time).
The reason the binary framing is dangerous is that it collapses stages that require different actions.
A company at Stage 1 needs to be doing customer discovery. Full stop. Not building product. Not pitching investors for Series A. Not hiring a head of marketing. Customer discovery.
A company at Stage 2 needs to be refining the core workflow until the engagement quality I described above is consistent — not expanding features, not building integrations, not preparing the growth infrastructure. The core.
A company at Stage 3 needs to be building distribution. The product is working. The question is whether the market shape supports scaling it.
Companies fail at each transition by misreading which stage they're in. Stage 1 companies raise on the strength of compelling conversations and build products that don't get used. Stage 2 companies over-invest in growth infrastructure before the core workflow is sticky. Stage 3 companies mistake distribution problems for product problems and rebuild features instead of channels.
The practical implication for fundraising: investors have different expectations at each stage, and fundraising against the wrong signals is a cause of enormous founder frustration.
A Series A investor is generally evaluating Stage 2 → 3 transition evidence: do you have a working core workflow, and is there early evidence that the market shape supports scaling it? Trying to raise a Series A on Stage 1 evidence — compelling customer conversations, strong team, large market — will produce a lot of interested meetings and no term sheets.
Not because the investors are wrong. Because you're presenting evidence for the wrong stage.
The diagnostic question before any fundraise: which stage of PMF evidence do I actually have, and which investors are optimizing for evidence at that stage? The mismatch between those two answers explains most failed fundraising processes better than any pitch feedback will.