Market research from 1,000 failed startups.
Find the gap before you build. Validate before you commit.
90% of startups fail. Every postmortem tells the same story: a founder bet on an idea nobody validated, in a market they didn't understand, with a model that couldn't sustain them.
This is 1,000 of those ideas — extracted, categorized, tagged, and ranked. Not guesses. Not speculation. Real ideas that real people bet their careers on.
Use this to:
- Find underserved markets — categories with the fewest attempts signal the largest gaps
- Validate your idea — stress-test against similar bets that failed
- Spot antipatterns — recurring patterns that killed companies in your space
- Ship faster — know which model-effort-speed combos actually work
This repository is a self-contained skill for AI coding agents (OpenCode, Claude Code, Cursor, Copilot, Gemini CLI):
cp -r skill /path/to/your/skills/productfound-market-researcherThe skill turns any agent into a market researcher. No installs, no dependencies, no API keys. Ask:
"What's the most underserved category?" "Validate my idea for a fintech SaaS" "Compare these three ideas" "I've been building for two months — assess my market"
Seven analysis types, persona self-selector, risk flag detection, confidence scoring, JSON output mode.
| Analysis | What it does |
|---|---|
gaps |
Find underserved categories and first-mover opportunities |
validate |
4-axis stress-test (Gap Clarity, Model Fit, Effort Realism, Speed vs Runway) |
competitive |
Category density, dominant models, risk tags |
persona |
Match ideas to builder profiles |
trends |
Model/tag/effort pattern surfacing |
compare |
Side-by-side scorecard across multiple ideas |
assess |
Evaluate an existing product against the dataset |
india-radar |
Cross-sector scan of 10 India-specific opportunity spaces |
Full documentation: skill/SKILL.md
| Dimension | Details |
|---|---|
| Ideas | 1,000 AI-generated from real postmortem patterns |
| Categories | 28 (DevTools, Health, Fintech, Ecommerce, AI-Tools, LegalTech, etc.) |
| Business models | 18 (SaaS, Marketplace, Freemium, API-First, etc.) |
| Tags | 142 |
| Builder personas | 5 |
| Effort levels | Weekend Project — 6+ Months |
| Speed tiers | Quick Cash — Long Game |
Real failed startups from around the world with structured data:
| Region | Companies | Capital lost | Top failure reason |
|---|---|---|---|
| US | 18 | $23.7B | Unit Economics |
| India | 10 | $8.2B | Fraud / Governance |
| Europe | 5 | $20.9B | Execution |
| SEA | 3 | $850M | Unit Economics |
| LATAM | 2 | $3.6B | Unit Economics |
| Africa | 2 | $350M | Execution |
| Global | 1 | $200M | Fraud / Governance |
MIT license. Free. Open source.
GitHub · Issues · MIT License