Guide · from the data

When products die: 528 of 11,025 Product Hunt sites are gone, and how each one went

We checked the sites of 11,025 Product Hunt products twice, days apart. 528 are gone with proof: how they went, which launch years and markets lose the most, and what we didn't count.

Published
October 19, 2026
Numbers refreshed
November 3, 2026
Reading time
4 min
Share
Post on X
How the gone products went

528 of 11,025 Product Hunt products whose site we checked twice, days apart, gone with evidence that holds up.

Every product page here links to a live site, so we check that the site is still there. In October 2026 we checked 11,025 products launched on Product Hunt, across the 301 markets in Idea check: each site fetched twice, at least three days apart, with its domain looked up in DNS and in the registry. A site counts as gone only when both checks agree and there's proof of it — not just a timeout.

528 are gone: 5% of all we checked. That's a floor, not an estimate: the rules below leave out every case we couldn't prove.

Is this a failure rate?

Not quite. Failure is hard to see from outside: a product can stop selling and keep its site up for years, so this counts what can be checked, a site that's gone. And the sample leans towards survivors: these are products Product Hunt still lists as alternatives to a live one, and a product that died early is less likely to be on those lists. The Product Hunt Graveyard dataset ↗, which took every product that reached Product Hunt's daily top leaderboards in 2023–2025, found about a quarter dead by July 2026, counting a few signs we don't (an expired certificate, a domain-level 404). Read ours as a floor: at least this many are gone.

Most just let the domain go

352 of the 528 — 67% — lost the domain itself: it's no longer registered (234), or it no longer exists in DNS (118). No shutdown page, no redirect: most likely a renewal nobody paid, and the name went with it.

The rest still have a domain, but not a product on it. 71 domains are up for sale on a marketplace page and 19 more are parked. 66 sites were taken down by their host — a Vercel, Netlify or Bubble “not found”, an “account suspended”. And 20 domains now serve someone else's site: when we opened 15 of them by hand, 14 were gambling sites.

That last group is why a dead product's old link is worth removing, not just leaving: the address that once pointed to a maker's product now points to whoever bought it. On this site, a gone product no longer links to its domain.

The longer since launch, the more are gone

Of the products launched in 2023, 7% are gone; of 2024's, 7%; of 2025's, 3%; of this year's, 1%. That fits the biggest cause above: a domain is usually paid for a year or more ahead, so a product that stops in its first months still has a site until the renewal comes round.

Gone, by the year the product launched
20261%16 of 2,355
20253%49 of 1,659
20247%143 of 1,952
20237%132 of 1,775

Products launched on Product Hunt from 2023. Older years are left out: see below.

Older launches look healthier — 5% gone across everything before 2023 — but that's the sample, not survival. These products are in it because Product Hunt still lists them next to a live product, as an alternative; a product that died years ago is less likely to still be on those lists. So only the recent years compare fairly, and even they lean towards survivors.

Which markets lose the most

In Journaling and mental wellness apps, 15 of 163 checked are gone:

The markets with the most gone

Of the 48 markets with at least 60 products checked. Open a row for the market.

At the other end, 5 markets have none gone among their checked products, among them AI dictation and writing tools (113 checked), Mac productivity utilities (82 checked), Crypto and stock trading tools (70 checked). With 60 to a few hundred products per market these are small numbers; a gap of a few points can be chance. Every market's page in Idea check lists its gone products and how each went.

What we didn't count

Some sites didn't answer at all, twice. We nearly counted those, until famous sites turned up on the list: our checker had skipped the network proxy for plain http:// links, so a timeout said more about us than about the product. No answer isn't proof, so none of those count.

Nor do these, each of which looked like a death more often than it was one:

  • A missing www. The www. name is gone, but the bare domain still works — sometimes with the product on it.
  • A dead subdomain. app.example.com is gone while example.com lives on, often because the company was bought or changed product.
  • A page that says “not found” but answers 200, or a builder's placeholder page: too many live sites look the same.
  • An app removed from the App Store, or a link to a page deep inside a site that still runs.

And a gone site isn't a cause. Some of these products were acquired, renamed or folded into another; the check sees only that the address stopped working, so we don't say why.

Our own records

Of the 8,906 makers' products we track, 221 are gone by the same rules. Their pages stay, marked as gone and out of search: what a product was built with is still evidence of what makers chose. Their sites are checked again every month.