Judge backlinks by traffic and leads, not DR

Most link reports I am sent lead with Domain Rating. A month of outreach is summed up as twelve links at an average DR of 60, and that figure is treated as the result. It tells you nothing about whether anyone read the pages, clicked through, or bought anything. Links still matter. The mistake is measuring them with a number built for comparing link profiles, when the business needs to know what each link did.

This post sets out what the scores measure, where they mislead, and the tests I use instead. The wider picture of how links work, and where earning them turns into link spam, is in the links and digital PR guide.

Google uses links as a signal when deciding how relevant pages are, and to find new pages to crawl.¹ Bing's webmaster guidelines list external links from relevant websites under discovery, and say strong internal and external linking supports discovery, authority evaluation and grounding eligibility.² Grounding is how Copilot's answers are tied to web sources, so for Bing links feed AI answers as well as rankings.

They matter less than the link-building industry suggests. Gary Illyes of Google told Pubcon Pro in 2023 that people overestimate the importance of links, that he did not agree they are in the top three ranking factors, and that it is possible to rank without them.³

For AI answers, mentions look more important than links. Ahrefs studied 75,000 brands with a DR above 40 and correlated a range of factors with how often ChatGPT, AI Mode and AI Overviews mentioned each brand, using Spearman correlation. Branded web mentions correlated at 0.66 to 0.71. Number of backlinks correlated at 0.19 to 0.24, which Ahrefs describes as very weak.⁴ A correlation study cannot show cause, and Ahrefs sells the tools that measure both, but the gap is wide.

Ahrefs describes Domain Rating as showing the strength of a website's backlink profile compared with the others in its database, on a 100-point scale. A linking domain's rating is split equally among the domains it links to.⁵ The scale is logarithmic, so moving from 70 to 80 takes far more links than moving from 20 to 30.⁶

Ahrefs is open about the limits. DR is purely link-based and does not take the search traffic of a site into account. It does not account for link spam, and large amounts of low-quality backlinks may increase it. Ahrefs also notes that some black-hat SEOs found ways to inflate DR with link spam and made money from it.⁶ Moz is equally direct about its own score: Domain Authority is not a Google ranking factor and has no effect on the results.⁷

The vendors have moved towards traffic themselves. Semrush's Authority Score combines link power with estimated organic traffic and a set of spam factors, one of which is an imbalance between links and organic traffic.⁸ Ahrefs' own checklist for vetting a link source asks whether the site gets any organic traffic and whether it is topically relevant to yours.⁵

None of this makes DR useless. Across 344,956 sites with Search Console data, Ahrefs found median monthly clicks rising from 9 for sites rated DR 0 to 10 up to 81,530 for sites rated 90 to 100.⁹ Across the web, strong link profiles and traffic tend to go together. The problem is the individual site that has one without the other, and that is where links are sold.

BuzzStream analysed 257,267 sites from a major link marketplace that had DR, organic traffic and price data. 39.6% of them had an average monthly organic traffic of zero, and BuzzStream classed 96.2% of the guest post sites as low quality.¹⁰ Those are sites that exist to place links. A buyer screening by DR alone would pass many of them.

Google's link spam systems are aimed at exactly this inventory. Its December 2022 update used SpamBrain to detect both sites buying links and sites used to pass outgoing links, and to neutralise unnatural links, with any credit they passed lost.¹¹ A link bought on a site with a healthy DR and no readers can still show up in your DR tool while counting for nothing in Google. You paid for the number, and the number is the only thing that moved.

The page matters more than the domain. In Moz's 2021 correlation study of the top 20 results for 10,000 keywords, Page Authority and page-level links correlated with rankings at 0.311 and 0.273, against 0.172 for Domain Authority and 0.069 for domain-level links.¹² Those are modest figures, and Moz's author is clear that DA is not used by Google directly.¹² A link on a page nobody visits, on a site with a high score, is weaker than the score suggests.

For every link a campaign earns or a vendor proposes, I want answers to five questions, roughly in this order.

Some good links will fail the first two tests. A relevant link on a respected page can help rankings while sending few clicks, and that is fine. What I would not pay for is a link that fails every test except the DR column.

Search Console's Links report is useful for seeing which of your pages attract links and which sites link most. It shows a sample of links, its tables stop at 1,000 rows, it does not say whether a link is nofollow, and some links in it may since have been removed. Google also says it typically ignores links from known spam sites.¹⁷ Use it to find what earns links on your site. Do not use it to value them.

I would report links the way any other marketing channel is reported. For each link: the referral sessions it brought, the key events from those sessions, the audience of the linking page, and a line on relevance. Keep DR in an appendix if a client is used to seeing it. Over a few months that report shows which publications and which kinds of coverage bring customers, which is the information a links budget should be spent on.

It also changes what you go after. A piece of coverage in a trade title your buyers read, or a mention in a newsletter with a real list, will usually beat a guest post on a high-DR site that exists to sell guest posts. The first kind brings visitors and builds the mentions that assistants appear to rely on. The second kind mostly moves a number in a tool.