# About

<!-- Traction Radar · https://qiliang.pivlu.com/en/about/ · 中文: https://qiliang.pivlu.com/about/ -->

# About Traction Radar

Traction Radar does one thing every day: find small products that suddenly took off in public data, work out why, who built them and how they earn, then write down how you could follow.

## What we watch

- Apple App Store free, paid and grossing charts in 150 countries, one snapshot a day.
- GitHub repos created within 30 days whose stars are climbing fast.
- Hacker News top items of the day, to find triggers.
- Small teams only: developers with more than 15 apps and large companies are removed.

## How we write

- Nine fixed sections per investigation; every sentence is marked verified, inferred or not found.
- Numbers must match our own snapshot database; a verified sentence must carry a link.
- We do not estimate revenue; we quote only what a founder published, with the source.
- Investigations that do not hold up are withdrawn and listed on the home page and the investigations page.

## What we do not do

- No scraping of sources whose terms forbid it, no bulk copying of other databases.
- No ads, no paid placement.
- No invented numbers, reviews or user counts.

## How this differs from similar sites

Similar sites are dense but hard to read at first glance. We put today's products, the number and the reason on the first screen and keep the dense data below; English is as complete as Chinese; and we write counter-examples (how many of last week's risers are still there) and the openings hidden in bad reviews.

## For agents

Every page has a Markdown twin (replace the trailing / with .md), the index is at [/llms.txt](/llms.txt), and daily public data is at [/api/](/api/today.json).

## Contact

Corrections, withdrawal requests or data licensing: the Markdown twin at the bottom of every page shows the page address; the contact email will be added at launch.


---
Generated by scripts from public data; investigations are written by a model with web access and spot-checked by a person. Every sentence is marked verified, inferred or not found.
