Glossary

webrank — the popularity rank: 1 = most popular domain on the web, driven by

real-traffic-weighted evidence. Explainer.

webrank_score — the same popularity on a friendly 0–10 log scale (10 ≈ top of the web).
pagerank — the 0–10 link-authority score, PageRank-style.

pay-level domain (PLD) — the registrable domain: bbc.co.uk, not news.bbc.co.uk or

co.uk. Everything in webrank is normalized to PLDs.

domain popularity — how much a domain is actually used (visits, DNS lookups).
domain authority — how much the web links to a domain; a link is a vote.
PageRank — Google's original link-voting algorithm (1998); votes from authoritative

pages count more. Genericized: any link-recursion authority score.

referring domain / IP / subnet — distinct linking domains, their distinct IP

addresses, and the distinct /24 networks those IPs sit in. Each step up is harder to fake. Explainer.

backlink — a single incoming link. Raw counts are nearly meaningless; diversity

metrics (above) are what matter.

Common Crawl — the open web crawl whose inter-domain link graph (~140M domains) is

webrank's deepest source.

CrUX (Chrome UX Report) — Google's dataset of what real Chrome users load; our

real-traffic anchor for weight derivation.

composite rank — a rank combining several sources. Tranco averages equally; webrank

weights by measured source reliability. Why it matters.

traffic-anchored weighting — deriving source weights from agreement with real-traffic

data (CrUX): sources that predict traffic get more say. Method.

rank churn — day-to-day rank instability; composites suppress it, single lists suffer it.
citable snapshot — a frozen, never-changing monthly list (webrank-YYYY-MM) you can

reference in a paper. How to cite.

head / torso / tail — the top ~1M (all signals), ~1–10M (broad sources), and 10M+

(link-graph only) depth zones. Coverage.

DNS-based list — a rank from resolver logs (Umbrella, Radar): sees machines as well as

humans, which is why CDNs rank high on them.