---

license: cc-by-4.0 language:

• en tags:
• product-hunt
• startups
• survival-analysis
• web-archaeology
• ai-tools pretty_name: The Product Hunt Graveyard — top-launch survival 2023-2025 size_categories:
• 1K<n<10K

---

The Product Hunt Graveyard: top-launch survival, 2023–2025

As of July 18, 2026, 24.4% of the products that made Product Hunt's daily top leaderboards in 2023–2025 are hard-dead — their domain no longer resolves, sits parked for sale, serves an expired TLS certificate, or 404s at the domain level. Another 4.5% redirect somewhere else entirely (acquirers, rebrands, or the founder's next project).

Full write-up + methodology: https://rightaichoice.com/state-of-ai-tools/graveyard

Key findings

• 24.4% hard-dead (505 of 2,066 products with a definitive verdict; n=2,291 probed)
• AI products die at the same rate as everything else: 24.5% vs 24.4%
• A top-5 launch-day finish cuts the death rate ~3×: 9.0% vs 25.9% for rank 6+
• Writing tools are the deadliest tag (45.2% dead); Open Source the safest (8.9%)
• The 2025 cohort is already 20.0% dead after roughly a year

What's in the dataset

One row per product (2,291 rows):

column	description
slug	Product Hunt product slug
name	Product name
websiteUrl	The product website we probed
cohort	Year-month of the sampled leaderboard appearance
isAI	AI classification (PH topic tag or AI terms in name/tagline/description)
votesCount	Product Hunt upvotes at time of scrape
verdict	alive / dead / parked / redirected_away / unknown
evidence	The exact DNS error, HTTP status, redirect chain, or parking fingerprint behind the verdict

Methodology (short version)

Sample: every product on 59 complete Product Hunt daily leaderboards drawn across January 2023 – April 2025 (~20 featured launches/day, all ranks). Each website fetched with a browser user-agent, 15s timeout, retries, and a separate double-checked DNS lookup. Conservative rules: anti-bot walls, timeouts, vanished app-store listings, and deep links that 404 while the root domain works are all unknown — excluded from every percentage, never counted as dead. Same-name domain moves count as alive. A second stricter audit pass re-probed every 404-based verdict at the domain root and moved 75 borderline rows to unknown, lowering the headline from a preliminary 27.1% to the published 24.4%. A full independent re-probe of the entire cohort reproduced the headline exactly (99.4% row-level agreement).

License & citation

CC BY 4.0 — free to use, including commercially, with attribution.

RightAIChoice, "The Product Hunt Graveyard" (2026). https://rightaichoice.com/state-of-ai-tools/graveyard