Link relevance vs authority: which matters more?
Neither wins outright. Authority is the better confirmed signal, because court evidence ties PageRank to Google's quality score. Relevance decides whether a link makes sense at all. The working rule is to require relevance first and then choose the strongest source among relevant sites.
In short
- No public source gives the weight of relevance against authority, so any claim that one counts a fixed amount more than the other is a guess.
- A Google engineer interview published as a court exhibit in May 2025 confirms that PageRank feeds a quality score and that anchors feed a separate relevance score.
- The 2024 leak documents an anchor mismatch demotion and fields for a site's topical focus, which analysts read as support for topical relevance.
- Topical relevance is the first test of a link for 58.6% of senior SEOs, according to uSERP's September 2026 survey of 116 respondents.
- Ahrefs' January 2025 study found Domain Rating correlates with rankings at only 0.131, so a high authority score alone is a weak reason to want a link.
Neither relevance nor authority wins outright, and no public source gives their weights. Authority has the firmer evidence: court documents confirm that PageRank feeds Google’s quality score. Relevance has firm evidence in direction and none on size. In practice the two answer different questions. Relevance decides whether a link makes sense. Authority decides how much a sensible link is worth. The working rule most practitioners use is this: treat relevance as a filter, then pick the strongest sources that pass it.
The two words need pinning down before the evidence makes sense. Authority is how much trust the linking site and page have earned from their own links. Inside Google that is PageRank and related signals. In SEO tools it is an estimate such as Domain Rating or Domain Authority. Relevance is how closely the linking page, its section and its site match your subject. It also covers language and country. A link from a Polish cooking blog to a UK accounting firm is weak on every count, because the subject, the language and the country all point somewhere else. Notice that the two definitions describe different things about the same link, which is the first hint that asking which one “wins” is the wrong way to frame the choice.
The evidence for authority
A Google engineer interview, published as an exhibit in the US antitrust case and reported in May 2025, describes the PageRank signal as “a single signal relating to distance from a known good source”, used as an input to the quality score. The same exhibit calls that quality score “incredibly important” and “generally static across multiple queries”. Those last words matter for how you think about a link’s value. A score that stays much the same from one query to the next belongs to the site and the page, so whatever a link adds on the authority side is carried into every search the page competes in, and is not tied to one keyword.
The 2024 leak adds a site-level authority field and a field that stores the PageRank of a site’s homepage against every page. Those are documented fields. Their weights are unknown. So authority, in Google’s own sense, is confirmed. What is not confirmed is that a vendor’s score matches it. Ahrefs’ study of one million keywords, published in January 2025, found Domain Rating correlated with rankings at only 0.131. A plain count of referring domains scored 0.255. A third-party score that tracks rankings less closely than a simple count is a weak stand-in for the thing Google measures, which means the confirmed importance of authority cannot be used as a reason to buy on a vendor’s number. Domain Authority vs Domain Rating explains why the scores are estimates.
The evidence for relevance
The case for relevance rests on several weaker pieces that point the same way, starting with Google’s statements. Google is reported to say on its “How Search works” pages that it uses links to help determine what pages are about. Google’s John Mueller is reported to have said that the total number of links “doesn’t matter at all”, because Google can ignore links at scale and weighs relevance. The court exhibit supplies the mechanism. Anchors are one of three inputs to the relevance score, alongside the page body and clicks. A link from a page about your subject is far more likely to use words that describe your page, so a relevant source tends to deliver the anchor text that this score reads.
The leak adds detail and also a caution. It documents a demotion for anchors that do not match their target, and two fields that measure how tightly a site sticks to a topic and how far a page strays from it. It also stores the text around each anchor. Analysts infer from these that links from topically aligned sites and pages count for more. That reading is an inference. The leak shows fields, not how they are used. There is research precedent as well: a 2002 academic paper on topic-sensitive PageRank is reported to describe computing PageRank separately for each topic, which shows the idea of tying link value to subject has been around for a long time, without showing that Google runs it that way now.
Practitioners have made up their minds. In uSERP’s September 2026 survey of 116 senior SEOs, 58.6% named topical relevance as their first test of a link. In BuzzStream’s State of Digital PR 2026 survey, 87.2% rated site relevance as a link quality criterion. Surveys record what people believe and do, so they are evidence about practice and not about Google, but they do explain why the filter-first rule above is so widely used. Our reading: confidence that relevance matters is high. Confidence in any specific weighting is low. Anyone who quotes a percentage split between the two is guessing, since nothing in the statements, the exhibit or the leak supplies one.
How the two work together
The court exhibit helps here because it separates two scores. Links act on both, and it is easier to follow if you take one side at a time. On the relevance side, what the link contributes is anchor text and context that say what your page is about. That affects ranking for the topic named by the link. The contribution is strong when the linking page covers your subject and uses descriptive words. A mismatch has a cost here: the anchor may be demoted or ignored, analysts infer from the leak.
On the authority side, the link contributes PageRank that says the site can be trusted. That affects the standing of the site and page across all queries. The contribution is strong when the linking page sits close to well-trusted sites, and when the source is weak it passes little. A link from a strong, off-topic page may still add to trust in your site, while adding little to relevance for your subject. A link from a weak, on-topic page does the reverse. That is why the debate has no single winner. Each kind of link feeds a different score, and a profile built from only one kind leaves the other score short.
Four kinds of link
| Link | Example | What to expect |
|---|---|---|
| Relevant and authoritative | A respected trade publication in your field cites your study | The best case. Rare and usually earned |
| Relevant, low authority | A small specialist blog with real readers links to your guide | Worth having. It adds context and looks natural. Do not expect it to move a hard keyword alone |
| Authoritative, off-topic | A national news site links to you in a general story | Worth having for trust and for readers. Less help for a specific topic |
| Neither | A general “write for us” site with no focus links with a keyword anchor | Likely ignored. This is the typical cheap paid link |
The fourth row is where most wasted budget goes. Sites that publish on every subject for a fee can show a high authority score and no topical focus at all. The score draws the buyer in, and the lack of focus means the link has nothing to contribute on the relevance side, so the money buys a number on a report. The two middle rows are the ones to think hardest about, because they are the links a campaign can obtain in quantity, and the table says both are worth having for different reasons. How to vet a publisher gives the checks.
Relevance is judged at three levels, in this order: the page, then the section, then the domain. At page level the question is whether the article that links to you covers your subject. At section level it is whether the article sits in a part of the site about that subject. At domain level it is whether the whole site is about your field. Page-level relevance comes first. A finance article on a general news site is relevant to a finance company, even though the site also covers sport. A finance article parked on a pet blog is not, because nothing around it supports it. This order is a working hierarchy used by practitioners, not a published Google rule.
Relevance can also be adjacent. A site for accountants is a sensible source for payroll software, since the two subjects differ while the readers are the same people. Ask whether the linking site’s readers could plausibly want what you offer. That test is more forgiving than matching keywords and also harder to fool, because it asks about an audience and not about the words on a page. Assume a firm that sells standing desks: a site about working from home passes the reader test even though it never uses the firm’s product terms, while a general site that has published one article stuffed with those terms does not.
The digital PR version of the debate
The question is sharpest in digital PR. A funny survey with no connection to the brand can earn links from large news sites. Those links sit in the third row of the table, authoritative and off-topic, and the argument is over how much that is worth. Agencies including Digitaloft and BuzzStream take the position that off-topic viral campaigns earn links that do little, and that campaigns should sit where the brand’s expertise, the audience’s interest and the target topic overlap. That is a practitioner position. We found no on-record Google statement from 2025 or 2026 on relevance in digital PR.
The position is consistent with the two-score picture above, which is a reason to give it some weight even without a Google statement behind it. A story with no connection to the brand gives a journalist no cause to use anchor text or surrounding words about the brand’s subject, so whatever the link passes arrives on the authority side alone. A story built on the brand’s own field gives the article a reason to describe that field. The agencies’ advice amounts to asking for both contributions from the same piece of coverage, and the cost of following it is that on-topic ideas are harder to make newsworthy than funny ones.
A working method
Use this when you compare two possible links. The order follows the rule from the top of the page, with relevance as the filter and strength as the tie-break.
- Reject sites with no topical connection at page, section or domain level, unless the link is pure editorial press coverage.
- Check that the site is real. It should have organic traffic from your target country and rankings that match its stated topic.
- Among sites that pass, prefer the stronger source. Look at the linking page, not just the domain score.
- Check the context. The paragraph around the link should make sense for a reader.
- Mix both kinds over time. Profiles of sites that rank in your market are the best benchmark. Link gap analysis shows how to read them.
The exception in the first step exists because press coverage belongs to the third row of the table, where the link is worth having for trust and for readers. The third step repeats a point that is easy to lose: authority in Google’s sense attaches to the linking page’s position among trusted sites, so a strong domain score on a page nobody links to tells you less than it appears to. In regulated or crowded markets the pool of relevant sites is small, and that changes the trade-off, because a strict filter can leave too few sources to work with. Our iGaming hub shows what that looks like. Read anchor text for the relevance signal you control most directly. To compare the tools that report authority scores, see backlink analysis tools. Terms are defined in the glossary.
Common questions
Is a relevant link better than a high authority link?
A relevant link from a real site is the safer choice, and most SEOs test relevance first. Authority is still confirmed as a Google signal, so the best link has both. No source gives the exact balance.
Does Google ignore irrelevant links?
Google has not said so in those words. The 2024 leak contains an anchor mismatch demotion, and Google's John Mueller is reported to have said Google can ignore links at scale. Analysts infer that off-topic links count for less.
What does relevance mean for a link?
It means the linking page, the section it sits in and the site as a whole cover a subject close to yours. Page-level relevance comes first, then the section, then the domain.
Is a link from a big news site relevant?
It can be. A general news site covers many subjects, so judge the article and the section. A business article that cites your data is relevant even if the site also covers sport.
Is Domain Rating a measure of authority?
It is Ahrefs' estimate of the strength of a site's backlink profile. It is not a Google metric, it can be inflated, and it says nothing about the page that would link to you.

