# Our Data Methodology

> Every number on this site is one of two things: a live lookup from Twitch's public API at the moment you click, or a snapshot from our own public archive. Nothing is estimated, nothing is backfilled — and what we can't measure honestly, we don't publish.

Source: https://vodfetch.com/nl/methodology  ·  Free to quote and cite with attribution to vodfetch.

## Live lookups (the tool pages)

Follower counts, username checks, live status and clip lists are queried client-side — from your browser, straight against Twitch's public API, at the moment you click. Results are shown with a timestamp, nothing is stored, and the numbers are Twitch's own current totals, exact — not cached, not estimated.

## The snapshot archive

Alongside live lookups we run a snapshot collector against Twitch's official Helix API, twice a day: the top ~500 games, the top ~1,000 streams (overall and for German, French and Spanish), and exact follower totals for a curated roster of ~115 well-known channels.

Snapshots are committed as public, versioned JSON to our open-source repository — anyone can verify every data point and every collection day. The archive starts in July 2026; there is deliberately no data before that, and we will never backfill from other trackers or scraped sources. If a day is missing, it stays visibly missing — no interpolation.

## Known limits (read before citing)

Twice-daily sampling misses intraday peaks. Per-game viewer sums in the archive are approximations computed over the top ~1,000 streams, not the whole directory. The roster is curated, not exhaustive. And a methodological boundary: systematic collection uses only the official API; the unofficial web API is used solely for lookups a visitor actively triggers — the same way the downloader itself works.

## What we refuse to publish

Estimated sub counts (explained on the sub-counts page). Historical charts for periods we didn't measure. Net-worth and earnings guesses. Streamers' ages and other personal-life data. The rule behind all of it: if a number can't be traced to a public API response or a cited public announcement, it does not appear on this site.

## Corrections

Our editorial policy applies to data exactly as it does to text: when we get something wrong, we fix it everywhere and say so on the affected page. The About page has our contact.

## Frequently asked questions

### Where do vodfetch's numbers come from?

Two sources only: live lookups against Twitch's public API at the moment you click (timestamped, exact), and our own public snapshot archive collected via the official Helix API. Nothing is estimated or bought in.

### Why does your archive only start in 2026?

Because that's when we started measuring. Scraping or buying older data would mean publishing numbers we can't vouch for. Trackers that started in 2015 have longer history; ours is shorter but verifiable end-to-end — every data point sits in a public repository.

### Can I use your data?

Yes. Snapshots are public JSON/CSV in the open-source repository under the MIT license — please credit 'vodfetch.com Twitch snapshots'.

### Why do your numbers differ from TwitchTracker or SullyGnome?

Different sampling times and methods. Our follower numbers are Twitch's exact live totals at the shown timestamp; viewer figures depend on when and how often each site samples. Their numbers aren't necessarily wrong — but the methodologies differ, and ours is documented on this page.