A client sent us a link report last month. Forty new links, average DR 52, every one of them followed. Not one came from a page that had ranked for anything in the previous twelve months.
That report cost real money and it will do almost nothing. It is the clearest example I have of why the way most people judge backlink quality is a decade out of date.
What Google actually says about links now
Start with the official position, because it has shifted and most link guides have not caught up.
Google's spam policies define link spam as links created "primarily for the purpose of manipulating search rankings", and list buying or selling links for ranking purposes, excessive exchanges, widget links, footer links spread across sites and low-quality directory links as examples (Google Search Central). Paid links are fine if they carry rel="sponsored" or rel="nofollow". Google says sponsored is preferred for paid placements and ugc for comments and forum posts (Google Search Central).
Then there is the weighting. At Pubcon in September 2023, Gary Illyes said people "overestimate the importance of links" and that they had not been a top three ranking factor "for some time" (Search Engine Land). In April 2024 he went further, saying Google needs "very few links to rank pages" and had "made links less important" over the years, then walked it back on social media within hours. That same spring Google's spam policy text changed from describing links as "an important factor" to simply "a factor" in determining relevance (Search Engine Land).
Our read: links still matter, but the marginal link matters less than it did, which means the cost of a bad one has gone up relative to what it buys you. That is the whole argument for being picky.
The data still says links correlate with rankings
Google saying links matter less is not the same as links not mattering. The correlation studies have not changed much.
Backlinko's analysis of 11.8 million results, last updated in April 2025, found the first result has on average 3.8 times more backlinks than positions two to ten, and three times more referring domains (Backlinko). Ahrefs data reported by Search Engine Land shows pages that already rank keep collecting new followed links at roughly 5 to 14% a month, which is part of why those correlations are so stubborn (Search Engine Land).
Correlation is not cause. Pages rank, then get linked to because they rank. But when you are the page trying to break into a top ten where everyone has 200 referring domains and you have 12, that is the gap you have to close, and no amount of on-page work closes it.
The other shift is AI surfaces. Ahrefs looked at 75,000 brands in May 2025 and found branded web mentions correlated with AI Overview visibility at 0.664, versus 0.218 for raw backlink count and 0.295 for referring domains (Ahrefs). Being talked about is now doing work that used to belong to being linked to. Keep that in mind when a link and a mention cost the same.
How we actually judge a link in 2026
This is the order we run through when a prospect lands on the list. The order matters, because most people start at the bottom.
1. Is the page about your topic, and is the site about your topic?
Relevance at both levels. A link from a page about crypto tax to your crypto tax guide beats a link from a general "finance news" site, which beats a link from a lifestyle blog that happens to publish a sponsored money post.
The test we use is simple: would this page link to us if nobody was paying? If the answer is obviously no, Google's systems are good at spotting the same thing, and the link is at best ignored.
Relevance is also where niche-specific judgement comes in. For an iGaming site, a link from a regulator, a licensing consultant or a genuine affiliate review carries a completely different weight from a link on a "top 10 casinos" page that exists to sell placements. Same follow status, same DR, completely different link.
2. Does the linking page itself rank and get traffic?
This is the check that separates people who have done link building from people who have read about it.
Open the page in Ahrefs or Semrush. Does it rank for anything? Does it get organic visits? Does the site as a whole have traffic that roughly matches its DR? If a site is DR 60 and gets 300 visits a month, something is wrong with it, usually a history of selling links or an expired domain rebuilt on old authority.
A link from a page with no traffic is a link nobody will click, on a page Google has decided is not worth showing. We would rather have a DR 30 site with 20,000 real monthly visits than a DR 60 ghost.
3. Is the site itself being linked to?
Look at who links to the linking site. Real publications, industry bodies, and sites you have heard of are a good sign. A referring domain profile made of other sites that sell guest posts is a network, and Google's link spam systems have been trained to identify sites "used for the purpose of passing outgoing links" since the December 2022 link spam update (Search Engine Land).
While you are there, look at what the site links out to. If the last ten outbound links go to a casino, a CBD store, a payday lender and a locksmith, the site is a marketplace and every link on it is discounted accordingly.
4. Where does the link sit?
Editorial links inside the body of a piece, surrounded by text that is about the thing you do, are the ones worth having. Author bios are weaker. Sidebars and footers are close to worthless, and Google's spam policy calls out "widely distributed links in the footers or templates of various sites" by name (Google Search Central).
We also check whether the linking page is reachable from the site's own navigation and internal links. An article that sits in a "sponsored" or "partner content" subfolder with no internal links pointing at it is a page the publisher has quarantined, and Google can see that as clearly as you can.
5. What does the anchor say?
Natural anchors are the brand name, the URL, the page title or a phrase that describes what the reader will find. Exact-match commercial anchors ("best online casino bonus", "Cyprus tax lawyer") in every link are the oldest footprint in the book.
Our rule: the anchor should be whatever the writer would have typed if they were not thinking about SEO. If you find yourself negotiating anchor text with the publisher, you are already in the wrong conversation.
6. Would anyone click it?
Ask this last, because it summarises the others. A link on a page with an audience, in a spot a reader will see, with anchor text that tells them what they will get, sends people. Those people become brand searches, mentions, and further links. A link that meets none of those tests is a number in a report.
The quality backlink checklist
Run any prospect through this before you spend money or outreach time on it.
| Check | Pass | Fail |
|---|---|---|
| Topical relevance | Page and site are about your subject | Site is general, page is a one-off sponsored post |
| Page traffic | The linking page ranks and gets organic visits | Page has no rankings, or the site's traffic is far below its DR |
| Linking site's own links | Linked to by real publications and industry sites | Linked to mostly by other guest-post sites |
| Outbound profile | Links out selectively to relevant sites | Links out to unrelated commercial niches |
| Placement | In the editorial body, on a page reachable from site navigation | Footer, sidebar, author bio, or a hidden "partner" folder |
| Anchor | Brand, URL, title or descriptive phrase | Exact-match commercial keyword |
| Disclosure | Editorial link, or paid and marked sponsored |
Paid, followed, undisclosed |
| Real audience | A reader would see and click it | Nobody visits the page |
A link that passes six or more of these is worth pursuing. One that fails page traffic or relevance is not worth having at any price.
What no longer matters, or never did
DA as a target. John Mueller has said, more than once, that Google does not use Domain Authority "at all" for crawling, indexing or ranking (Search Engine Journal). DA and DR are third-party estimates of link equity and useful for sorting a prospect list, but "get me links above DR 50" is a brief that produces the report at the top of this post.
The follow/nofollow obsession. Google treats nofollow, sponsored and ugc as hints and generally does not follow them, but a nofollow link on a page real people read is worth more than a followed link on a page nobody visits (Google Search Central). We stopped filtering prospects by follow status years ago. A mention in a national paper is a mention in a national paper.
Raw link count. The linking-domain metric that correlates in every study is referring domains, not total backlinks. Ten links from one site is one vote.
Panicking about toxic links. Google's stated approach since the 2022 link spam update is to neutralise spammy links so that "any credit passed by these unnatural links are lost", not to punish you for links you did not build (Search Engine Land). Disavow is for links you built and regret, or for a manual action. It is not a monthly hygiene task.
Where the 2026 updates fit
Two things people asked us about this summer, so for the record: Google confirmed that neither the June 2026 nor the August 2026 spam update targeted link spam (both via Search Engine Roundtable). If your rankings moved in either window, links are an unlikely cause. We wrote up what the August update actually hit separately.
The more relevant change is that Google's spam definition now covers attempts to manipulate its generative AI responses, which we covered in can you buy your way into AI citations. The same relevance and audience tests that make a link good for rankings make a mention good for AI visibility. There is no separate rulebook.
Examples of links that pass
To make this concrete, here is what a good link looks like in the niches we work in most.
For a crypto exchange, a link from a developer's technical write-up that cites your documentation for a specific protocol, or coverage in a crypto trade publication that has its own readership and its own links from mainstream press.
For a law firm, a listing on the bar association site, a citation from a university law blog analysing a judgment you commented on, or a local news story quoting a partner.
For any business, a .edu or .gov link earned by giving those institutions something worth pointing to. We wrote a separate guide on how to get .edu backlinks because the approach is different from commercial outreach.
What these share: a real audience, a real reason for the link to exist, and a page that would have been published whether or not you were in it.
How to run this on your own profile
Export your backlinks from Search Console or Ahrefs, sort by referring domain, and score the top 50 against the checklist above. Most sites find that a small share of their links do nearly all the work, and that the "toxic" links they were worried about are mostly just irrelevant links Google already ignores.
Then do the same for the top three competitors on your main term. The gap between their passing links and yours is your link building brief. If you need somewhere to score prospects as you go, our link prospecting tracker is built for exactly this pass/fail vetting.
If you are being offered links that fail these tests but come with a low price, that is a separate question, and we answer it in should you buy backlinks. If you would rather someone else did the vetting, that is what our backlink building service does, and a free SEO review will tell you whether links are actually your bottleneck before you spend anything on them.
Frequently asked questions
What makes a backlink high quality?
A quality backlink sits on a page that is about your topic, that Google already ranks and sends traffic to, and that has real links of its own. It is placed in the editorial body, the anchor text reads naturally, and the linking site links out selectively rather than to anyone who pays. Domain-level scores like DA or DR are a filter at best, not the test.
Does Domain Authority matter for backlinks?
Not to Google. John Mueller has said repeatedly that Google does not use Domain Authority at all for crawling, indexing or ranking. DA and DR are third-party estimates of link strength, useful for sorting a long list of prospects, but a high score on a site with no traffic and a paid-links habit is worth very little. Judge the page, not the number.
Do nofollow backlinks help SEO?
Google treats nofollow, sponsored and ugc as hints and generally does not pass ranking credit through them. That said, a nofollow link on a page real people read can send customers, build brand mentions and lead to followed links elsewhere. We tell clients to stop screening prospects by follow status and start screening them by whether anyone will actually see the link.
How many quality backlinks do I need to rank?
There is no fixed number. It depends on the query and what the pages already ranking have. Google's Gary Illyes has said the company needs very few links to rank pages, and industry studies still show top results carrying more referring domains than the rest. Look at the top ten for your target term, count their referring domains, and treat that as the bar rather than a generic target.
