webrank vs Tranco
Tranco is the research standard for domain popularity lists, and it fixed the right problem: single lists (Alexa, Umbrella, Majestic alone) are unstable and easy to game, so Tranco averages several of them. webrank starts from the same idea and changes the one thing Tranco left on the table: the averaging itself.
The core difference: derived weights, not equal weights
Tranco combines its input lists with equal weight — it assumes every source is an equally good measure of popularity. Our data says that assumption is false: the sources disagree with each other in measurable, structural ways (link-based sources correlate with real user traffic at only ~0.3–0.5). So webrank derives each source's weight from data: sources are weighted by how well they agree with held-out real-traffic measurements (Chrome/CrUX). The result, on the same validation:
| weighting | predicts held-out real traffic (Spearman) |
|---|---|
| webrank (traffic-anchored weights) | 0.42 |
| consensus/eigenvector weights | 0.32 |
| equal weights (Tranco's method) | 0.29 |
webrank predicts real user traffic ~46% better than an equal-weighted composite. The full derivation, weight table, and validation protocol are on the methodology page — reproducible from public data.
Side by side
| Tranco | webrank | |
|---|---|---|
| combination | equal-weight average | data-derived weights, validated |
| numbers per domain | 1 (popularity rank) | 2 (popularity rank + 0–10 link authority) |
| coverage | ~4.5M (top-1M list) | 146M ranked domains |
| inputs | 4–5 lists | 8 sources + Common Crawl graph |
| freshness | daily | daily (monthly frozen snapshots) |
| citable snapshots | yes | yes (permanent monthly lists) |
| price / key | free, open | free, open, no key |
| per-source transparency | list of inputs | live per-source ranks in every API response |
What Tranco still is
Tranco is academic, widely cited, and battle-tested — if a paper's reviewers expect Tranco, use Tranco. webrank's claim is narrower and testable: if you want the list that best predicts real user traffic, an equal-weight average is measurably not it. And if you also care about the difference between popular and authoritative (CDN backends vs actual websites), one number can't express it — see the two numbers.
Try it: curl https://api.webrank.top/rank/tranco-list.eu · download the list · methodology · cite