Who Actually Wins on Product Hunt — And Why Founder Support Networks Break Down

Most advice about launching on Product Hunt is anecdotal — a founder who did well writes up what they think worked. What's missing is the denominator. Nobody publishes the launches that didn't work, and nobody has visibility into what founders privately promise each other in the run-up to launch day.

We have an unusual vantage point on the second question. LaunchPact is a coordination platform where founders make explicit, tracked commitments to support each other's launches, and where fulfilment of those commitments is verified rather than self-reported. That produces something rare: a dataset of promises between strangers, with a recorded outcome for each one.

This report combines that internal dataset with a sample of Product Hunt leaderboard data covering the same period.

Dataset A — Product Hunt launches. 3,069 launches featured between 3 September 2025 and 14 July 2026, spanning 314 distinct days. Fields include vote count, comment count, daily rank, maker count, topic tags, and media composition.

This is not a complete census of Product Hunt. It is a leaderboard-weighted sample: a median of 8 launches per day, skewed toward the top of each day's ranking. 55.0% of sampled launches finished in the day's top 10, and the lowest-ranked launch in the sample placed 83rd. Product Hunt features substantially more products per day than this sample captures.

Dataset B — Support agreements. 3,534 resolved mutual agreements between founders on LaunchPact, comprising 7,068 individual commitments. An agreement is "resolved" when both sides have either fulfilled or failed to fulfil their commitment. We excluded 104 agreements terminated for administrative reasons — a launch being withdrawn, or a participant cancelling before launch day — because those reflect changed circumstances rather than a decision not to follow through. Including them would inflate the failure rate.

Fulfilment is verified by screenshot review rather than self-attestation, which removes most of the honesty bias that would otherwise make this kind of data worthless.

What Dataset B cannot tell you. These are founders who opted into a platform built around mutual support. They are, almost by definition, more predisposed to cooperate than founders in general. The follow-through rates below should be read as an optimistic ceiling for founder reciprocity, not an average. A founder asking for help in a Slack group with no tracking and no verification should expect worse.

We also cannot separate deliberate defection from ordinary human failure. A founder who intended to reciprocate and forgot is recorded identically to one who never intended to. We think that distinction matters a great deal for interpretation, and we have no way to measure it.

The relationship is monotonic and steep. Each step up in team size improves both median votes and top-10 probability, and the gap between the extremes is more than twofold on top-10 rate.

The obvious mechanism is distribution: more makers means more personal networks activated on launch day, more employees and investors and friends with a reason to show up. A five-person team doesn't need to be five times better than a solo founder to place five times more reliably — it just needs five sets of contacts.

This is worth stating plainly because it cuts against how launch outcomes are usually discussed. A solo founder who lands at rank 40 and a five-person team that lands at rank 5 have not necessarily built products of differing quality. They have differing headcount, and Product Hunt's ranking is sensitive to that in ways the leaderboard doesn't disclose.

We want to be careful about the causal claim here. Larger teams also tend to have more funding, more runway to polish a launch, and more launch experience. Maker count is partly a proxy for all of those. We can say the association is strong and consistent; we cannot cleanly isolate network size as the cause.

Launches including video placed in the top 10 roughly 14 percentage points more often. The same caveat applies with more force than in Finding 1: producing a launch video signals a team with time, budget, and preparation, and those attributes drive outcomes independently. Video is likely both a cause and a symptom.

Of 3,534 resolved mutual agreements:

• 1,993 (56.4%) completed — both founders followed through.
• 1,541 (43.6%) broke down — at least one founder did not.

At the level of individual commitments rather than pairs, the picture looks better: 5,175 of 7,068 commitments were fulfilled, a 73.2% individual follow-through rate.

The gap between those two numbers is the entire problem. Mutual agreements require both sides. A 73% individual rate compounds into a 56% pair rate, because each agreement has two independent chances to fail. Reasonably reliable individuals produce unreliable pairs.

This is the result we found most striking. Among the 1,541 agreements that broke down:

• 1,261 (81.8%) had exactly one founder follow through.
• 280 (18.2%) had neither follow through.
• Zero had both.

The intuitive story about broken commitments between strangers is that everyone quietly defects — both parties assume the other won't show, and neither does. That story is wrong. In more than four out of five failed agreements, somebody kept their word and got nothing back.

That reframes the problem entirely. The scarce resource in founder communities is not goodwill; goodwill is abundant and measurable. What's scarce is matched goodwill. Most failures are a coordination problem, not a character problem — and coordination problems are tractable in ways character problems are not.

It also means the cost of participation falls unevenly. Each collapsed agreement produced one founder who invested time and attention and received nothing measurable in return. Over the period studied, that happened 1,261 times.

We grouped founders by how many agreements they had entered and measured the fulfilment rate of their individual commitments.

First-timers fulfil barely a quarter of their commitments. Founders with 21 or more agreements fulfil nearly four-fifths.

Two mechanisms plausibly drive this, and we can't cleanly separate them. Selection: founders who don't follow through get filtered out — they lose access, or lose interest, and never reach the higher buckets. Learning: founders who stay develop reputation worth protecting and behave accordingly.

The distributional detail matters too. Eighty-five founders — the 21+ bucket — account for 4,795 of 7,068 total commitments, roughly 68% of all activity. A small, reliable core carries most of the volume. This is the shape of most healthy communities, and it means aggregate statistics are dominated by the behaviour of people who have already proven themselves.

The first-timer bucket is small (29 founders, 29 commitments) and its 27.6% figure should be treated as indicative rather than precise.

Among the 1,922 completed agreements where we could measure both timestamps, the median gap between the first founder honouring their commitment and the second reciprocating was 13 hours. 59.1% reciprocated within 24 hours.

The practical read: if a founder hasn't reciprocated within a day, they probably aren't going to. Reciprocity is a fast, same-cycle behaviour, not something that accrues slowly. Any system trying to improve follow-through has a window of roughly one day to work with.

Only 3.0% of fulfilled commitments included a comment on the supported launch (214 of 7,068).

Comments require writing something specific about somebody else's product, which means actually looking at it. Votes don't. A 3% comment rate against a 73% fulfilment rate suggests most support is genuine but shallow — founders show up, but they don't engage deeply enough to have an opinion worth typing.

For founders launching, this is the actionable version: comments are where the real signal is, and almost nobody is competing there.

Three things follow from the data, with varying confidence.

Solo founders face a structural disadvantage that has nothing to do with product quality. We're confident about the association and less confident about the precise mechanism. But a solo founder planning a launch should treat network mobilisation as a first-class problem rather than an afterthought, because the founders they're competing against on the leaderboard have four more people doing it by default.

Most support failures are coordination failures. The 81.8% one-sided figure is the strongest evidence in this report. Founders are, on the whole, willing to help each other. They are bad at ensuring help arrives in both directions. That's a solvable problem — through tracking, reciprocity windows, reputation, or simply making commitments explicit rather than implied.

Reputation systems appear to work, though we can't fully separate learning from selection. The gradient from 27.6% to 78.5% is large enough that we doubt it's entirely selection artefact, but we'd want a cleaner design to say so confidently.

This dataset and its findings are free to cite, quote, and reproduce with attribution to LaunchPact and a link to this page. Journalists, researchers, and founders writing about launch dynamics are welcome to use any figure here.

If you'd like a cut of the data we haven't published — by topic category, by day of week, by launch month — get in touch and we'll run it.