How Open Page Rank is calculated
Every score comes from one open, public source, the Common Crawl web graph, and the same idea behind the original PageRank algorithm: a domain is important when other important domains link to it. Here is exactly how we turn billions of links into a single 0-10 number.
We start with Common Crawl, not our own crawler
Common Crawl is a nonprofit that crawls the public web and releases the data for anyone to use. Alongside the raw pages it publishes a web graph: a map of which sites link to which. Building on an open, independent source means our scores are transparent and reproducible, and they never depend on any private or proprietary data set.
The latest graph connects over 118 million registered domains through several billion links.
We score whole domains, not individual hosts
A single site is usually spread across many hosts: the root domain (example.com), www, and subdomains like blog. or shop.. Scoring each one separately would split a site's authority across several fragments.
So we collapse every host into its registered domain using the Public Suffix List, the same rules a browser uses to know that example.co.uk is a site but co.uk is not. That gives one node per domain, and every link between two domains becomes one edge in the graph.
We run PageRank over the whole graph
PageRank treats every link as a vote, but votes are not equal. A link from a domain that is itself widely linked to counts for far more than a link from an obscure one. That definition is circular, a domain's importance depends on the importance of the domains linking to it, so it is solved across the entire graph at once rather than domain by domain.
The result is a single importance value for every domain, earned purely from the structure of real links across the open web.
We add the breadth of who links to you
PageRank captures the quality of incoming links. We pair it with a second signal, the number of distinct domains that link to you, which captures their breadth.
This count is authority-weighted: only referring domains that carry real weight of their own are counted. A domain cannot pad its breadth by collecting links from a thousand throwaway sites.
We combine both into a 0-10 score
The PageRank value and the authority-weighted referring-domain count are combined into a single Open Page Rank score from 0 to 10, shown to two decimals. Higher means more authoritative.
The scale is logarithmic, like the web itself: the distance from a 2 to a 3 is small in absolute links, while reaching a 9 or 10 takes the kind of link profile only the largest sites on the internet have. A perfect 10 is reserved for that small handful of global domains.
What the number means
A rough guide to how scores map onto the web.
The most-linked sites on the web: major platforms, global brands, top institutions.
Well-established sites with a strong, genuine link profile.
Typical active sites, the broad middle of the web.
New, small, or lightly-linked domains.
Manufactured authority is filtered out
Some domains try to fake authority by building or buying links from networks of low-value sites. We look at the shape of each domain's link profile to tell earned authority from manufactured authority, and domains that show the tell-tale patterns of a link network are scored down toward zero. The aim is a number that reflects authority a domain actually earned.
Comparable every month since 2018
We keep every release back to January 2018, so each domain has a full monthly history. Scores are anchored from one release to the next, so a 6.0 in 2019 means the same thing as a 6.0 today and you can read a genuine trend. Common Crawl's graph is monthly from 2024 onward; earlier gaps are interpolated and clearly flagged as estimates, never presented as measured points.
Refreshed with every crawl
Each time Common Crawl publishes a new web graph, about once a month, we ingest it, recompute every domain's score, and append a new point to the history. Scores stay current with the open web.
What the score does and does not capture
Open Page Rank measures authority as it appears in the crawled open web. A few honest caveats follow from that:
- Brand-new or fast-rising domains can read low for a release or two, until their new links are picked up by the next crawl.
- Coverage follows Common Crawl. A site that blocks crawlers, or is very new, may be thinly represented.
- It measures link authority, not traffic, revenue, or content quality.
See it for any domain
Check domains free in your browser, browse the top 10 million, or pull scores and history through the API.