Data & Methodology
Every number in FindStreamers is produced by a documented process. This page explains where the data comes from, how each signal is derived, how often it refreshes, and — just as important — what each signal can and cannot tell you.
1. Collection: public sources, monitored continuously
The database is compiled exclusively from publicly available information: official platform APIs and public channel pages on Twitch, Kick and YouTube. Automated monitoring runs around the clock — sampling live viewership on a rolling cycle, capturing channel metadata, stream titles and categories, and public channel descriptions. The product serves refreshed data every 6 hours; underlying collection is continuous. We never access private data, DMs, or anything behind a login.
2. One creator, one profile
Channels are deduplicated across platforms. When the same creator streams on Twitch and Kick, both channels attach to a single unified profile, so audience and pricing are evaluated per creator — not double-counted per channel.
3. Contact verification
- We only collect business contact details that creators have published publicly themselves — for example an email address in a channel bio.
- Email addresses pass automated deliverability checks before entering the vault, and are re-validated on a recurring cycle; each contact displays its current status in the product.
- The vault holds contacts in currently-working status only. A contact you reveal stays revealed at no further cost.
4. Audience geography & language
A creator's own country is a poor predictor of where their viewers live. We estimate audience geography from behavioural signals on the channel — when streams go live, viewership patterns, stream language and title language — then rank countries per channel and attach a confidence level. Treat it as a ranked estimate, not a measurement of every viewer.
5. Sponsorship detection
Stream titles and channel details are read continuously to detect brand appearances. Each detection is recorded as a sighting with first-seen and last-seen dates, deduplicated in time buckets, and rolled into tenure: how long a brand has appeared on the channel and whether it still does. A sighting means a brand appeared on the channel — the commercial terms behind it are not public and we do not claim to know them.
6. Rate modeling
Suggested prices are modeled estimates, not quotes. The model combines viewer numbers, audience quality and geography, platform, and comparable observed deals into an expected price, a low–high range, and a confidence grade. Use it as a negotiation benchmark and sanity check. Actual rates are set by creators and can differ from any model. We never publish a named creator's modeled rate on the public web — rates are visible only inside the product.
7. Hygiene and ethics
- Sanctioned regions are excluded from the database.
- Creator removal requests are honoured permanently, with a suppression list preventing re-collection.
- Identifiable minors are excluded. Creators are profiled only in the context of their public streaming activity.
- Access is paid, quota-limited and logged — the database is not publicly queryable and cannot be bulk-exported.
8. Known limitations
We would rather you know these than discover them: viewership sampling can miss short streams; geography estimates carry uncertainty, which the confidence level reflects; sponsorship detection can miss deals that never surface in public channel data; and rate estimates are strongest for mid-sized channels where comparable deals are plentiful, weaker at the extremes. When a signal is uncertain, the product shows the uncertainty instead of hiding it.
Questions about methodology: support@findstreamers.com.