When someone googles your brand – or asks ChatGPT who the experts in your category are – two things decide the answer: what independent sites say about you, and whether search and AI systems trust those sites enough to repeat it. Backlinks and digital PR are the two levers that shape that answer, and most advice treats them as one interchangeable “SEO” line item. That is the mistake. They do different jobs, they fail in different ways, and one of them is quietly worth more than it used to be.
My read is blunt: backlinks set your ranking floor, digital PR builds your reputation ceiling. A single page can rank on links alone. A brand cannot build trust on links alone. The companies getting this right in 2026 fund both but measure them separately – and they stop expecting one to cover for the other.
Backlinks and digital PR are not the same job
A backlink is a hyperlink on someone else’s site that passes ranking authority to a page on yours. Digital PR is the practice of getting credible, independent sources to describe you accurately; a link is one byproduct of that, not the point. The distinction sounds academic until you look at what each one actually moves.
| Backlinks (link building) | Digital PR | |
|---|---|---|
| Primary output | A hyperlink that passes ranking authority | Independent coverage and mentions (links are a byproduct) |
| What it moves | Classic organic rankings, crawl discovery | Brand trust, branded search, AI citations |
| How much you control | High – you pick the target page, often the anchor | Low – the editor or reporter decides |
| Typical failure mode | Manipulation patterns Google can detect | Nothing citable worth pitching |
| Time to visible effect | Weeks, if the link gets indexed | Three to six months of compounding |
Sources: Google Search Central’s spam policies, plus the Ahrefs and Backlinko studies cited below; the timing figures are practitioner norms rather than guarantees.
What the link data actually says
Short answer: backlinks still correlate with rankings, and the correlation is weaker than it was a decade ago. Both halves of that sentence matter.
Backlinko’s analysis of 11.8 million Google search results found that the number-one position averages 3.8 times more backlinks than positions two through ten. It also found something more uncomfortable for volume chasers: roughly 95% of all pages have zero backlinks, so the pages you are competing with are not winning on volume alone. Referring domains – the count of distinct sites linking to you – tracked rankings more closely than raw link totals.
Ahrefs reached a similar place from a different angle. Its January 2025 study of one million US SERPs put the correlation between referring domains and top-20 rankings at 0.255, with total backlinks at 0.248 and Domain Rating down at 0.131. Those are modest numbers, as correlations go. But Google weighs hundreds of signals, and a single factor explaining a quarter of the variance is not nothing.
Where the story gets genuinely interesting is what Google says out loud. In 2021, Google’s John Mueller said that “the total number [of links] essentially is completely irrelevant” to rankings, because the system weights each link individually and ignores what it does not trust. Search Engine Roundtable reported the remark at the time. I would not take that as “links are dead” – Mueller was pushing back on link-count obsession, not on link value. But it is a fair warning that the number in your backlink tool is a diagnostic, not a scoreboard.
The part that changed: mentions now predict AI visibility better than links do
Here is the finding that should reshuffle your budget. Ahrefs correlated 75,000 brands against millions of AI Overview responses and measured which signals predicted whether a brand got mentioned. The strongest predictor was branded web mentions – 0.664. Backlinks came in at 0.218. That is roughly three times the predictive power for mentions over links.
| Myth | What the evidence actually shows | What to do instead |
|---|---|---|
| More backlinks equal higher rankings | Correlation, not causation. Unique referring domains track rankings more closely than raw link count. | Chase relevance and one strong link over ten weak ones. |
| Nofollow links are worthless | Google has treated nofollow as a hint since 2019, and AI search engines ignore the attribute entirely. | Accept nofollow links from good sources. Stop pricing them at zero. |
| Domain Authority is Google’s metric | DA and DR are third-party scores. Google uses neither. | Use them as rough guides, never as targets. |
| You must disavow every toxic link | Google ignores most spam automatically, and “toxic” is not a Google concept. | Leave it alone unless you have a manual action to clear. |
| Buying links always triggers a penalty | The rules target manipulation and non-disclosure, not the transaction itself. | Vet relevance, and make sure any paid placement is properly qualified. |
Sources for the table: Google’s published spam policies and link-attribute guidance, Backlinko’s 11.8 million-result ranking study, and Ahrefs’ 75,000-brand visibility study. Dates and links are in the text and the methodology note below.
Source: Ahrefs, study of 75,000 brands, Spearman correlation, published May 2025. A December 2025 follow-up across ChatGPT, AI Mode and AI Overviews kept the pattern, and found YouTube brand mentions correlating at roughly 0.737 – the single strongest signal it measured.
Read the top three rows again: every one of them is an off-site factor, and every one describes how often your name appears elsewhere. Backlinks are also off-site, but they are the weakest off-site signal on the list. Notice, too, that mentions count whether or not someone hyperlinks you. Ahrefs found that brands earning the most mentions get up to 10x more AI Overview mentions than the next quartile down, while 26% of brands get none at all.
Why digital PR is really a corroboration layer
AI systems do not crawl your anchor-text distribution and judge it. They weigh whether independent sources describe you the same way. A company described consistently across six unrelated publications is one a model names with confidence. A company that only exists on its own domain is one the model hedges about or leaves out.
That is why I think the best mental model is not “PR earns links” but “PR builds corroboration, and links are a side effect.” Spotify Wrapped is the clean example: it is a data asset built from real user behaviour, distinctive enough that people share it and journalists cover it without being pitched. It generates coverage and links, but the reason it works is that it gives everyone something true and specific to repeat. You cannot buy that, and you cannot fake the consistency.
What actually earns coverage
Most digital PR campaigns fail before outreach starts, because there is nothing worth citing. Before you build a media list, build the thing a reporter can use:
- Original data. Survey your customers, aggregate your own product data, or analyse a public dataset from an angle nobody has taken. A single honest number nobody else has published will out-earn a quarter of opinion pitching.
- Source-request platforms. Reporters broadcast what they need on Connectively, Qwoted and Help a B2B Writer, often on deadline. Answering a few relevant queries each morning is the highest-return habit in the discipline – the writer already wants the story, you are just supplying the source.
- Trade press before national press. A relevant trade publication with a small but aligned audience earns a more useful link and a faster response than a broad mention nobody in your market reads.
- An angle bank. Keep a running list of story angles the way you keep a product backlog, so that when the news touches your category you have something ready the same day.
Package all of it for immediate use: one headline finding, two or three supporting figures, one clean chart, and a quotable line from a named spokesperson. If a reporter has to dig through a 40-page report to find the point, they will use someone else’s.
The link-building habits that now hurt you
Google’s spam policies are explicit about what counts as link spam: buying or selling links for ranking purposes, excessive link exchanges, and optimised anchor text in guest posts or press releases distributed across sites. Paid links are permitted only if they carry a rel="sponsored" or rel="nofollow" qualifier. This is not a grey area you can argue your way out of.
I would not abandon guest posting wholesale – a well-placed article in a genuinely relevant publication is still a reasonable way to earn content-based backlinks, the kind that exist because the piece is worth reading rather than because a fee changed hands. The pattern to avoid is scale: dozens of thin posts on unrelated blogs with commercial anchors is trivial for Google to detect, and the resulting harm is not recoverable by cleaning it up later. Google’s own spam-update guidance says ranking gains from spammy links cannot be regained once the system removes their effect.
How to tell whether it’s working
Link count is the wrong dashboard. Track the signals that map to reputation and to the AI-era correlation data:
- Referring domains (distinct linking sites), not total links.
- Branded search volume – how often people search your name directly. Rising branded search is the clearest sign PR is compounding.
- AI citation share. Put your buyers’ real questions through the major assistants monthly and record whether you appear.
- Unlinked mentions. Set up brand monitoring and, politely, ask publishers to attribute when a piece describes you without a link.
Where “mentions beat links” breaks down
The mention-first argument has a real weak point, and it is worth stating plainly. Ahrefs’ one-million-SERP research found that links correlate much more strongly for local queries – around 0.33 for referring domains – and more strongly for informational queries than for commercial ones. In competitive non-branded categories, links still look close to a requirement, not an optional extra. A separate analysis of 200 commercial phrases found that 96.3% of top-10 sites had over 1,000 unique domain backlinks.
So the honest position is not “links are dead.” It is that links and mentions do different jobs and should be measured on different timelines. Links defend rankings, especially for competitive and local search. Mentions win the narrative, and the narrative is what AI systems now repeat. If your category is competitive on classic Google search and you have no link acquisition plan, your mentions will not save your rankings. If you have a strong link profile and nobody talks about you, your AI visibility will stay at the bottom of that 75,000-brand list.
