Intelligence from real outcomes.
Failradar is a startup failure intelligence platform. We analyze real outcomes from 5,727 startups to help founders, investors, and educators learn what actually kills companies — and how to avoid it.
Why this exists
Most startup advice is survivorship bias in disguise. You hear about the 1% that made it — not the thousands that didn't. The result? Founders keep making the same fatal mistakes, in the same categories, with the same patterns.
Failradar flips that lens. Instead of studying what winners did right, we study what losers did wrong — systematically, across thousands of startups since 2005. The data tells a different story than the headlines.
Our thesis is simple: the fastest way to de-risk a new venture is to understand the specific ways similar ventures have already failed. Not anecdotes — data.
Data & methodology
Failradar's dataset is built from two primary sources:
- 5,727 startups — tracked from public batch directories, with status classifications (Active, Inactive, Acquired) cross-referenced against multiple signals including domain status, social presence, and press coverage
- 1,749 supplemental entries — from loot-drop.io's curated collection of failed startup post-mortems with structured failure reasons
Every record passes through a Zod-validated ingestion pipeline that normalizes 239 raw categories into a clean taxonomy of ~70 sectors, standardizes batch identifiers, and computes per-category risk scores. The pipeline is deterministic — same input, same output.
Category risk scores are calculated as the ratio of Inactive (dead) companies to total companies within each sector, weighted by sample size confidence. Sectors with fewer than 5 companies are flagged as low-confidence. All analytics are pre-computed at build time and served as static pages — no runtime database queries, no API latency.
1,003
Confirmed dead
3,992
Still active
732
Acquired
70+
Sectors tracked
Built by
Failradar is built by a solo founder based in Saudi Arabia with a background in software engineering and data analysis. The project grew out of a personal frustration: while researching whether to pursue a startup idea, the only available “failure data” was scattered across blog posts, podcasts, and Twitter threads — nothing structured, nothing searchable, nothing quantified.
So we built what we wished existed: a structured, searchable database of startup outcomes with computed risk scores and pattern analysis. The kind of due diligence tool that should exist for anyone betting years of their life on a new venture.
Data attribution
Startup data compiled from startups.rip and public Y Combinator records. Supplemental failure data from loot-drop.io. Failradar does not host or reproduce original post-mortem content — we link to primary sources and provide computed analytics on top of structured data.
All category risk scores, survival curves, and pattern analyses are original computations by Failradar. If you use our data in research or reporting, please credit Failradar as the source.
Start exploring
See the data for yourself.
Browse every startup, explore failure patterns by sector, or run your own idea through the validator.