The dataset
Every company here, with what was true about it on the day it was found and every dated observation since. Take it — no signup, no key, no rate limit.
What is in it
One row per published company. The columns that matter and are not self-explanatory:
- n
- Permanent log number. Assigned on first publish and never reused or re-sorted — it records order of notice, not rank.
- added
- The edition the company was published in. The claim this whole record makes is we saw this on that date.
- what
- A neutral one-line description. No opinion.
- why
- The opinionated note, capped at three sentences. This is the part that can be wrong, which is the part worth reading.
- signals
- Counted by hand from their own store and profile: stockists, products, reviews, entry price. Each carries the date it was taken.
- movement_since_found
- Change between the first measurement and the latest. Audience numbers appear only as a multiple, never as a raw count.
- asking
- What the company said it needs, in its own words.
What is deliberately not in it
No scores, no rankings, no votes, and no follower counts. Ranking would reward the legible companies and bury the weird ones, which is the opposite of the point. Raw audience numbers are stale within a week and publishing them would expose the selection rule to gaming, so growth ships as a multiple instead.
No revenue, headcount estimates or valuations either. Nothing here is a number that was not counted by a person on a stated date.
Terms
Use it, quote it, build on it. An attribution link back is appreciated and not required. The notes are opinion; the dates and counts are records. Nothing here is investment advice, and inclusion is never paid for or traded.