Mental Health Graveyard: 542 Dead Startups, Dataset Released
Get the Health newsletter
Daily health & science — research, biotech, public health, the studies worth knowing. Free.
- The Mental Health Startup Graveyard dataset catalogs 542 digital mental health organizations that left the market between 2000 and 2026 — shutdowns, bankruptcies, acquisitions, pivots and consolidations — with each company coded on up to 18 fields including business model, payer, funding, country, reason for leaving, clinical evidence and a 'key_mistake' paragraph
- Classification of four axes (product type, entity type, care mode, and clinician-in-loop status) was performed by LLM agents against a fixed taxonomy; the author ships written reasoning for every single label in a companion CSV so any coding decision can be checked rather than trusted
- The headline finding reports consumer-pay companies died at a 53% rate against 21% for institutional-pay companies — 'who pays predicts survival far better than anything on the founding team' — while a clinician co-founder 'moves the exit rate by nothing at all: 47% against 47%'
- The factual fields — funding, dates, country, outcome — were drawn from Crunchbase, CB Insights, Tracxn, public deadpool databases, app store removals, Ahrefs domain data, and trade press, not from the LLM classification pass
- The release ships as CSV, JSON, and a labels-rationale file under CC BY 4.0 with no form, email or signup; a worked-example Python snippet showing how the 53% vs 21% payer split is reproduced from the raw rows is included
- A companion 406-page report distills seven patterns from the data; the author says they assembled it while building Mentalium, a voice-first CBT diary, 'before putting years into a product in this niche'
- Stated limitations: the dataset is a graveyard, not a random sample, so shares compare groups against each other rather than estimate failure probabilities; funding is disclosed for 59% of companies; 67% of the sample is US and UK; groups under 25 observations show a direction rather than a precise value; 'an acquisition is not a success either: some of these deals are fire sales out of bankruptcy'
Why it matters: Founders, investors and acquirers in digital mental health now have a free, fully reproducible dataset to test claims about what kills companies in the niche, with the payer-type finding offering a concrete, citable signal: 53% death rate for consumer-payer models versus 21% for institutional-payer — though 67% of the sample sits in the US and UK, limiting international generalization and pushing observers to verify classifications with the shipped rationale file rather than trust them.
