What Are Referring Domains?

The word sits close to the HTTP header "Referer", which tells a server the page a visitor came from. RFC 1945, the HTTP/1.0 specification of May 1996, already named one use of that header: a server can build "lists of back-links" from it. RFC 9110 of June 2022 notes that "the field name is misspelled". None of the tool documentation says the SEO metric was named after the header, so the link between the two is a family resemblance of words, not a documented origin.

In web analytics the same words mean something else. In Adobe Analytics, "Referring domain" is a dimension that reports which domains visitors click through to reach your site. A link on the other site only shows up there once someone clicks it.

A link tool crawls the web, stores every link it finds to your site and then removes duplicates at the level of the linking website. What counts as one website differs slightly from tool to tool, and that is where the totals start to drift apart.

Ahrefs gives its own example: two links from the New York Times still make one referring domain. Subdomains are folded into the main domain. In Ahrefs' help, three subdomains of the same domain, hosted on three different servers, produce one referring domain and three referring IPs. Google Search Console groups its links report the same way: protocol, subdomains such as www. or m., and subdirectories are stripped, and everything lands under the root domain.

The second question is which links count at all. Ahrefs, Moz and Semrush include nofollow links in the referring domains list and let you filter them out. Ahrefs makes one exception: its Domain Rating only counts referring domains with at least one followed link. Semrush calls a domain "follow" if it sends at least one follow link, and "nofollow" if it sends none.

The third question is time. A link that was live last year may be gone today, and each tool keeps lost links for a different period or in a separate index. That is why the same site shows different totals in different tools on the same day. Ahrefs says so itself: "we all crawl the web differently."

Source: help pages and glossaries of Ahrefs, Majestic, Moz, Semrush and SE Ranking (accessed 10.11.2026).

Take a page that receives links from three websites. Site A links five times from three of its pages. Site B links once. Site C links once from blog.c.com and once from www.c.com.

That makes 8 backlinks. Tools that fold subdomains into the root domain, as Ahrefs, Majestic and Search Console document, report 3 referring domains. A tool that treated blog.c.com as its own website would report 4. Site A's five links carry no more weight in this count than site B's single link.

No two tools report the same figure, so the useful question is what each one counts:

Source: Ahrefs Help Center and Academy, Google Search Console Help, Majestic Glossary, Moz Help Hub, SE Ranking Knowledge Base and API documentation, Semrush Knowledge Base (accessed 10.11.2026).

Search Console is the only source in this table that shows what Google itself has recorded. It is also the least complete: Google calls the report a sample, not a full list.

The metric shows how many separate websites link to you, which a backlink total cannot. A thousand links from one partner site and a thousand links from a thousand sites look the same as backlinks and very different as referring domains.

It has three blind spots. It counts spam domains the same way as a trade association or a newspaper. It says nothing about whether the linking site is on your topic. And it only covers what the tool's crawler has found.

Ahrefs published one data point on the relationship with traffic in December 2023. Of the roughly 20 million pages in its index with no referring domains, 2,997 received more than 1,000 search visits a month. Ahrefs adds that its sample leans towards the "quality side of the web" and that correlation does not imply causation.

Source: ZDS glossary entries; Moz, Domain Authority (18.03.2025); Ahrefs Academy, backlink metrics (accessed 10.11.2026).

Google does not publish a metric called referring domains. What Google says about links is more general. Its documentation on crawlable links states that "Google uses links as a signal when determining the relevancy of pages and to find new pages" (last updated 10.12.2025). Its guide to ranking systems lists PageRank as part of the core systems.

Semrush's knowledge base goes further and says Google weighs a high number of referring domains "as a ranking signal". Google's documentation contains no such statement, so that sentence is Semrush's claim.

Google's spam policies, last updated on 28.08.2026, define link spam as links created "primarily for the purpose of manipulating search rankings", for example bought links or excessive link exchanges. Those domains count as referring domains in every tool. According to the Search Console help, Google typically ignores links from known spam sites.

• Open Google Search Console, go to Links and look at Top linking sites. That list is free and comes from Google's own data.
• Choose one tool for your regular checks and set the scope: root domain, subdomain or a single page.
• Write down the date of every export, because lost links drop out on each tool's own schedule.
• Look at new and lost referring domains once a month, not just the total.
• Check unknown sites. If one looks like spam, ask for removal first. The Search Console help names disavow as the step for when everything else has failed.

• Compare figures only within the same tool and with the same scope.
• Read the followed and nofollow shares separately. A total that grows only through nofollow links is not the same as one that grows through followed links.
• Set new referring domains against lost ones, so a single month with a gain is not mistaken for a trend.
• Check where new domains come from. Ten domains from one network can be a problem, and Majestic's subnet count helps to spot that.

• Putting the Ahrefs figure next to the Semrush figure and treating the gap as a change.
• Reporting backlinks when the report says referring domains.
• Reading "lost" as "removed". Moz says it only means its crawler failed to fetch the link again.
• Setting a target number of referring domains as if Google published one.
• Counting new domains from a link network or an expired domain as a win.

This entry was written in November 2026 with AI assistance. The counting rules come from the help pages of Ahrefs, Majestic, Moz, Semrush, SE Ranking and Google Search Console, read on 10.11.2026, and Google's statements from its own documentation. It was reviewed before publication and contains no client data.