What PageRank measures
PageRank models a reader who follows links at random. A page's score is the share of time that reader spends on it, which depends on how many pages link in and how much score those pages hold themselves. A single link from a heavily linked domain can outweigh hundreds from obscure ones.
Open PageRank runs that calculation over the Common Crawl link graph, a public dataset of links between domains, and scales the result to a decimal between 0 and 10 alongside a global rank. Because the input is a crawl of the public web rather than Google's index, the figures differ from anything Google holds, but they move in the same direction when a domain gains or loses good links.
Reading the results table
Each domain you entered gets one row with three columns. Domain is the normalised name after the checker has removed the protocol, path, port and any leading www. and lower-cased it. Open PageRank shows the score as a filled bar and a number with two decimals. Global rank is the domain's position in the dataset, written with a hash, or a dash when the dataset holds a score but no rank.
A domain the dataset does not know returns no score, and the row carries the message Not found in the Open PageRank dataset. New domains, domains with no inbound links in the crawl and some internationalised names fall into this group. It does not mean the domain is penalised. It means the crawl has no link data for it.
Input rules and error messages
Split your list with commas, spaces or new lines. The checker validates each name as a domain before calling the data source, so a name with a space, a missing top-level domain or a reserved suffix such as .local is rejected with Invalid domain(s) and the offending entries listed. More than ten domains returns Check up to 10 domains at a time, and an empty box returns Enter at least one domain.
The lookup needs a server-side API key for the Open PageRank service. If the key is missing on the server you are using, the tool says so plainly (The PageRank checker isn't configured on this server yet) rather than showing invented numbers. Other failures report the HTTP status the service returned, and a request that takes longer than 10 seconds is abandoned.
Using the score without misreading it
Compare domains within one niche rather than against the giants. A regional supplier at 3.2 is doing well if its direct competitors sit at 2.5, and chasing the 8.9 of a global publisher is not a plan. The calculation rewards links from already well-linked domains, so a handful of mentions from respected industry sites moves the figure more than many directory listings.
A high score does not guarantee traffic. The measure counts links, not visits, rankings or relevance, and a domain can be well linked for historical reasons while publishing nothing today. Pair the figure with what the domain actually is: the WHOIS Lookup shows how old it is and who registered it, and the Domain Scanner shows how it is set up.
- Evaluating an expired or aftermarket domain: check that the score is backed by links from real sites, not an old link network.
- Vetting a guest post or partnership: a higher score than yours suggests the link is worth having.
- Tracking your own domain: note the score and rank monthly. Movement in rank shows up before the rounded score changes.
Limits of the method
Scores are per domain and cannot tell you which pages carry the links or who links to you. A dedicated backlink index is needed for that. The dataset is refreshed on Common Crawl's schedule, so a link earned this week will not show for a while, and a domain that recently changed hands keeps the score its old content earned until the crawl catches up.
The checker sends only the domain names to the data source, stores nothing about your query afterwards and is rate limited per visitor. Results for a list are cached for a day. To see where a page's own outbound links go, and whether they carry nofollow, use the Website Link Analyzer.