Why 150 enriched profiles beat 650,000 hollow ones

Every cofounder matching platform eventually gets asked the same question by a prospective user: "How many people are on it?" The question sounds reasonable. Bigger pool, better odds. More fish, easier fishing. That intuition is almost completely wrong for the kind of search cofounder matching actually is, and understanding why is the single most important thing a founder can internalize before deciding where to spend their time.

The short version: the expected number of real introductions you get out of a matching platform is approximately the pool size multiplied by the fraction of profiles that contain real signal, multiplied by the fraction that are active, multiplied by the fit ratio. Three of those four numbers collapse to near zero in a large thin pool. So the product of them also collapses to near zero. You can have 650,000 users and produce fewer real introductions than a small, carefully curated pool of 150.

Let's walk through the math, then walk through why the industry still fights this truth.

The funnel of a cofounder search

For any matching platform, the real funnel a user experiences looks like this.

1. Total user count. The headline number.
2. Active profiles. Users who logged in in the last 90 days. In most older platforms, this is 5 to 15 percent of the total count.
3. Profiles with enough information to evaluate. Users with a real bio, real sources, and an identifiable skill set. In older, filter-era platforms, this is often 5 to 20 percent of the active count.
4. Profiles that match your actual need. A complementarity filter. For a sharp founder, this is usually 1 to 5 percent of the evaluable pool.
5. Profiles where the other party is open to starting right now. Not every serious founder is ready this quarter. Halve it again.

Now run the numbers on a large thin pool. Start with 650,000 total users. Take 10 percent active, 10 percent evaluable, 2 percent relevant, 50 percent currently available. That's 650,000 × 0.1 × 0.1 × 0.02 × 0.5 = 65 real candidates.

Run the same math on a small deep pool. Start with 150 total users. Take 80 percent active (because the platform enforces depth and activity), 100 percent evaluable (because the platform doesn't let thin profiles stay in the pool), 2 percent relevant, 50 percent available. That's 150 × 0.8 × 1.0 × 0.02 × 0.5 = 1.2 real candidates.

At first glance the big pool wins, 65 to 1.2. Except the 65 are hidden in a haystack of 650,000 profiles you have to browse, filter, and manually triage, and the match signal itself is noisy because the structured fields that were used to rank them barely correlate with actual fit. You will not discover more than a handful of the 65 in any reasonable amount of time. Most founders give up after inspecting 30 or 40 profiles.

Meanwhile, the 1.2 candidates in the small deep pool are surfaced directly by a matching engine that has read each profile's real sources. Every one of them is worth a message. Your conversion from "introduction surfaced" to "actual conversation" is a different order of magnitude.

The quality-adjusted comparison isn't 65 vs 1.2. It's more like 5 real introductions vs 1 real introduction. And those are similar numbers, with the small deep pool winning on time spent, dignity preserved, and signal-to-noise.

That's the math on day one. Where it gets interesting is what happens as each pool grows.

What happens as each pool grows

A large thin pool that scales without enforcing quality gets worse, not better. Every new user dilutes the pool. The percentage of evaluable profiles drops as the inactive tail lengthens. The sensation of browsing becomes more like a graveyard, because every additional year adds more abandoned accounts than live ones. Serious users leave. Search quality drops. The flywheel spins the wrong direction.

This is why older platforms with nominally enormous user counts produce so few introductions. The raw number keeps going up. The real pool stays flat or shrinks. The number you're quoted is a vanity metric.

A small deep pool that scales with quality enforcement gets better in almost every dimension. Each new user adds signal because the platform only lets them into the active pool if they have real sources connected and a real bio. The percentage of evaluable profiles stays near 100 percent. Matching gets more precise because the matching engine has more examples to work with. Even the chance of a specific rare complementary match (a GTM-heavy healthtech founder, a researcher with shipping experience) goes up faster than pool size, because quality growth concentrates people who take the platform seriously enough to connect their actual work.

The flywheel direction matters more than the raw count at any given moment.

The industry pattern that keeps repeating

Knowing all this, why does the industry keep building large thin pools? Three reasons, and they're worth understanding so you can spot them.

First, pool size is the easiest number to optimize and the easiest number to show investors. A platform that optimizes for signups can accelerate fast because signups are cheap. Thin profiles are free. Matching quality is invisible. Press writes about "hundreds of thousands of founders". The venture narrative gets built on the headline, the engineering ships the features that drive the headline, and by the time anyone cares about introduction quality, the product is locked into the thin-pool architecture and can't easily change direction.

Second, the filter-era technology genuinely could not do better. Without language models that could read real sources and without embeddings that could represent complementarity, the only way to match people was to give them a filter panel. A filter needs categories. Categories need simple profile fields. Simple profile fields select for thin profiles. The whole architecture was a response to a technological limit, and when the limit lifted, the architecture stayed because it was already the business model.

Third, founders initially ask the wrong question. "How many users do you have" is a proxy for "is this platform worth my time". It's the wrong proxy, but it's the obvious proxy, and it rewards platforms that prioritize growth over depth. The industry trained its users to ask the size question, because size was what the industry could sell, and the feedback loop closed.

This pattern is not unique to cofounder matching. Dating apps went through the same cycle, hit the same wall, and are now rebuilding around verified signal and smaller, more curated pools. Recruiting platforms went through the same cycle, hit the same wall, and are now (in the higher end of the market) rebuilding around deep candidate profiles rather than keyword search. The cofounder matching category is a few years behind but moving in the same direction.

See our live data on the Finnish cofounder ecosystem →

What a quality-first platform actually does differently

A platform that takes pool quality seriously looks different in specific ways. A few things to look for:

It gates the search pool. Users do not become visible to others until they have connected real sources, written a real bio, and passed a depth threshold. Users who let themselves go stale drop out of the pool automatically. This is a painful feature to build because it shrinks the headline number, and it's the single clearest signal of a platform that prioritizes introduction quality.

It reads real sources, not self-reported fields. Connecting a GitHub, LinkedIn, personal site, or company website should meaningfully change what the platform knows about you. If a platform's profile is just dropdowns you filled in yourself, it is a filter-era product regardless of what it claims.

It matches on complementarity, not preference. A quality-first platform should be able to identify the cofounder you need, not the cofounder you asked for. These are usually different. If the platform is just filtering by your stated preferences, it's automating your bias rather than correcting it.

It measures what happens after the match. Conversations with depth, companies formed, long-term outcomes. A platform that can tell you that users with high complementarity scores exchange three times more messages than users with low scores is a platform that has a feedback loop. A platform that only reports signups and "matches" (first clicks) has no feedback loop and cannot improve the thing that matters.

It is comfortable with being small. This is the most counterintuitive one. A quality-first platform will tell you exactly how many active profiles it has, including the ones that failed the depth gate, and will not try to hide the number behind an aggregate. Honesty about the pool size is itself a quality signal.

What this means for the founder deciding where to spend time

If you're a founder deciding where to spend your cofounder search time, ignore the headline pool number. It is the least predictive single metric in the category. Instead, ask three questions.

Can I see a sample match from this platform? Does it actually read my work, or does it just echo back what I filled in?

Is the platform willing to tell me how many profiles pass its quality gate, not just how many users it has?

Does the platform's matching change meaningfully when I connect more real sources, or is it stuck on the dropdowns?

A platform that answers those three questions well is worth your month. A platform that refuses to, or hides behind a headline user count, is probably not worth your week.

The cofounder you want is not waiting for you in the largest pool. They are waiting in the deepest one. Those are almost never the same place.