Dataset: Dead mental health startups, 2000-2026, coded on 18 fields
4 hours ago
- Dataset includes 542 mental health startups that left the market between 2000 and 2026, with 18 fields covering business model, payer, funding, outcome, and key mistakes.
- Who pays (consumer vs institution) is the strongest predictor of survival: consumer-pay startups died 53% of the time vs. 21% for institution-pay; a medical co-founder had no impact on exit rates.
- Classification axes (product type, entity type, care mode, clinician involvement) were assigned by LLM agents, with written rationale provided for each label to allow verification.
- The data is a graveyard, not a random sample, so comparisons are relative; funding data is partial, and acquisitions may include fire sales, not successes.
- A full 406-page report with seven findings and detailed appendix accompanies the dataset, which is freely downloadable under CC BY 4.0.