Introduction
DataSocial is a TikTok data warehouse you can query directly. Think Dune or Flipside, for TikTok.
What is in it today (2026-09-27): about 4 billion creators, about 10 million sounds with video counts per day since January 2026 (with a gap from mid-May to September), and a first 35 thousand videos. The video and creator-history tables started filling on launch night and grow every hour; datasets shows every table’s live row count.
Two ways in
Section titled “Two ways in”| What it is | Use it for | |
|---|---|---|
Data API (/v1/data/*) |
Raw SQL (ClickHouse) on the tiktok database. Open now, free during launch. |
Discovery, rankings, trends, bulk exports. See Data API. |
Live API (/v1/live/*) |
A live read of TikTok, one request at a time. Coming soon. | Anything not in the warehouse yet, or when you need it this second. See Live API. |
Not sure where to start? Open the search page, pick a table, and run the starter query. Want an API key? Join the waitlist.
Credits
Section titled “Credits”- One account has one credit balance, bought as a monthly subscription tier, and spent by both APIs.
- SQL is charged by the work a query does. Premium columns (today only
tiktok.creators.email) cost more per row, and must be revealed per query, on purpose. - Live calls are charged per request.
- Prices, units and plan sizes are set at launch. Until then SQL is free, with per-visitor limits and no sign-up.
The tables
Section titled “The tables”All customer tables live in one database, tiktok. There are no duplicates and you never need FINAL.
- Entities (latest state): creators, videos, sounds.
- History (appended once per day or hour, never rewritten): creator_daily, video_snapshots, sound_daily.
- Derived (precomputed metrics; look here first): sound_trends, sound_countries, creator_stats, video_trends.
- Catalogue: datasets lists every table with its row count and freshness, live.
Metric definitions used everywhere: growth % = change ÷ previous value (with a floor, so tiny bases don’t top
the charts); a missing day is NULL, never zero; engagement rate = (likes + comments + shares + saves) ÷ views.