How the weight is calculated

No black box. Everything on a report comes from two public datasets, and the score is one formula you can reproduce.

The weight

Every domain in the Common Crawl host graph (cc-main-2026-may-jun-jul) has a position when ranked by harmonic centrality — roughly, how close it sits to every other site through links. The weight maps that position onto 0–100 on a logarithmic scale:

weight = 100 × (1 − log10(position) / log10(102,661,340))

Position 1 scores 100. Each factor of ten in position costs the same number of points, because moving from #10 to #100 matters about as much as moving from #10,000 to #100,000.

Referring domains

The number of distinct registrable domains with at least one host linking in, from the same crawl. Subdomains of one site count once.

Chrome traffic

The Chrome UX Report publishes, every month, the origins that had enough real Chrome users to be measured, with a coarse popularity bucket. We count in how many of the last 66 monthly releases a domain appears. A site can have many links and no audience; this is the part links cannot fake.

Expiring domains

Registrability and prices come from the registrar at cost. Before a domain is listed it passes the same checks we use for our own purchases: link farms (hundreds of near-identical domains sharing an identical link count), toxic niches, spam-shaped names, blocklists, and link profiles whose links have already died are excluded. Domains rejected only for commercial reasons are listed with the reason.

Limits

Common Crawl does not see the whole web, so small sites can be missing or under-counted. Chrome UX Report only includes sites above a traffic threshold. Drop dates are estimates.