For twenty years, authority in search had a working definition almost everyone internalised: PageRank. Links were votes, quality links were better votes, and the page with the strongest link profile won. That model was so dominant that many marketers still reach for it reflexively when thinking about AI search. But answer engines do not evaluate authority the way PageRank does, and treating the two as the same leads you to invest in the wrong signals. Answer-engine authority is closer to reputation and topical trust than to a tally of links. Getting this distinction right reshapes how you earn visibility in AI answers, so let us lay it out carefully.
The short answer
PageRank evaluates authority mainly from the link graph, scoring pages by how many quality pages link to them. Answer engines evaluate authority differently, leaning on topical and entity trust, corroboration across many sources, consistency, and how directly and credibly you answer, with raw link metrics a comparatively weak signal. The clearest evidence is that Ahrefs, studying 75,000 brands, found branded web mentions correlate strongly with AI visibility while backlink metrics correlate very weakly (Ahrefs). So answer-engine authority is not PageRank; it is closer to being trusted and talked about on your topic.
What PageRank actually measures
Recall what PageRank does. It models the web as a graph of links and treats each link as a vote, weighting votes by the authority of the linking page, so authority flows through links. Its genius was turning the link graph into a scalable authority signal, and it made link building the central authority-earning activity in classic SEO. Crucially, PageRank is fundamentally about links: the structure of who points to whom. That is its strength and, for AI search, its limitation, because answer engines are not primarily reasoning over a link graph when they decide whom to trust.
What answer engines mean by authority
An answer engine is trying to decide which sources to trust enough to base an answer on, and it draws that from a much wider set of signals than links. It looks at whether you are consistently associated with a topic, whether many credible sources mention and corroborate you, whether your information is accurate and current, and whether you directly and clearly answer the question. This is authority as reputation and topical trust, assembled from how the whole web discusses you, not from a link tally. Profound’s analysis of citation patterns shows engines favouring sources that credibly and directly answer, rather than simply the most-linked (Profound).
The evidence: mentions beat links
The data draws the line sharply. In the same 75,000-brand study, branded web mentions correlated with AI visibility in the range of roughly 0.66 to 0.71 across platforms, while link metrics like number of backlinks and URL rating showed very weak correlations (Ahrefs). Read plainly: the thing PageRank measures, links, is a weak predictor of the authority answer engines act on, while being mentioned, the thing PageRank largely ignores, is a strong one. This is not a subtle tweak to the old model; it is a different basis for authority, and it explains why link-heavy strategies underdeliver in AI search, a point developed in do ChatGPT links count as backlinks for SEO.
Topical and entity authority
A key part of answer-engine authority is topical: are you an authority on this specific subject, as understood from everything the model has read. A site can have a strong general link profile and still lack topical authority on a given question, and it can have modest links but deep, consistent, widely-referenced expertise on a narrow topic and be treated as highly authoritative there. Answer engines reward the latter more than PageRank ever did, which is why a focused specialist can out-authority a bigger, more-linked generalist on its own subject. Authority here is subject-specific, not a single site-wide score.
This is one of the most liberating differences for smaller players. Under a site-wide link-authority model, a newcomer is more or less stuck behind incumbents who have accumulated links for years, because the score is global and slow to move. Under topical authority, the game resets per subject: you can be a nobody overall and still be the recognised authority on one narrow topic, if the web consistently treats you as such. That is why depth in a niche beats breadth with links in AI search, and why the authority contest is far more winnable for focused specialists than the old PageRank world ever allowed.
Corroboration and consistency
Answer engines also weigh corroboration heavily: do independent sources agree about you and your claims. A brand that many credible sources describe consistently is easier to trust than one with a big link profile but a thin or contradictory reputation. This is why digital PR and genuine mentions do more for AI authority than link acquisition, and why consistency of how you are described matters. It is reputation logic, not link-graph logic, and it rewards being genuinely, verifiably what you claim to be across the web.
There is a subtle robustness to corroboration that a link count lacks. A link can be bought, exchanged, or dropped on a low-value page, which is why link-based authority has always been gameable at the margins. Corroboration across many independent sources is much harder to fake, because it requires actually persuading a lot of separate voices to say consistent things about you, which usually only happens when you genuinely are what they say. Answer engines lean on that harder-to-fake signal precisely because it is harder to fake, and that is part of why the authority they reward tracks real reputation more faithfully than a link total ever did.
Where PageRank-style authority still helps
Do not overcorrect into thinking links are worthless. Classic link-earned authority still helps answer engines, but indirectly. Backlinks lift your rankings in classic search, and pages that rank well are disproportionately the ones pulled into AI answers, with Ahrefs finding a large share of AI Overview citations come from top-ranking pages (Ahrefs). So strong PageRank-style authority feeds AI visibility upstream, through ranking, even though it is a weak direct authority signal to the engine. The relationship is indirect but real, and Google frames AI features as resting on your normal search presence (Google Search Central).
PageRank versus answer-engine authority
This table contrasts the two models.
| Dimension | PageRank authority | Answer-engine authority |
|---|---|---|
| Primary signal | The link graph | Mentions, topical trust, corroboration |
| Scope | Largely site-wide | Topic and entity specific |
| Links | Central | Weak direct signal |
| Mentions | Mostly ignored | Strong signal |
| How it is earned | Link building | Reputation, expertise, being talked about |
| Role in AI | Indirect, via ranking | Direct |
Why chasing links is the wrong move now
If you carry the PageRank model into AI search, you will over-invest in link building and under-invest in the signals that actually earn answer-engine authority. Links have their place for classic ranking, but pouring your budget into acquiring them expecting AI authority is optimising the weak signal while neglecting the strong one. The higher-leverage work is earning genuine mentions, building topical depth, and establishing consistent, corroborated expertise, the shift argued in classic SEO versus AI visibility. Reallocating from link chasing to reputation building is the practical consequence of this whole distinction.
How to build authority answer engines reward
Work the signals that matter. Develop deep, consistent topical expertise so you are clearly an authority on your subject, not a generalist. Earn mentions across the credible sources your audience and the models read, through genuine PR and being worth talking about. Keep your information accurate and current, and answer the specific questions in your area directly and clearly. Maintain a consistent description of who you are so corroboration is easy. And keep enough classic SEO strength that you rank, since ranking still feeds AI, following the citation playbook in how to get cited in ChatGPT.
The E-E-A-T overlap
If this sounds like E-E-A-T, that is not a coincidence. The experience, expertise, authoritativeness, and trustworthiness framing that Google promotes for quality content maps closely onto how answer engines evaluate authority: demonstrated expertise, a trustworthy reputation, and corroboration. Answer-engine authority is, in effect, E-E-A-T operationalised at scale from what the web says about you, which is why quality-and-reputation work pays off in AI even though it looks nothing like link building. Some engines lean on specific proxies, such as domain-level trust, which is explored for one platform in are Perplexity answers based on domain rating, but the general principle holds across engines.
A worked example
Say two companies compete on a niche topic. Company A has a large, general backlink profile built over years. Company B has fewer links but is the recognised specialist on the exact topic, mentioned consistently across industry sources and answering the key questions clearly. Under PageRank thinking, Company A looks more authoritative. But when an answer engine handles a question on that topic, it favours Company B, because the mentions, topical trust, and corroboration all point to B as the authority on this subject. Company A’s link advantage helps it rank generally, which offers some indirect benefit, but on the specific question, reputation beats the link graph. That inversion is the whole lesson.
Common misconceptions
The biggest misconception is that answer-engine authority is just PageRank; it is a mentions-and-trust model, not a link-graph one. The second is that backlinks are useless for AI, when they still help indirectly through ranking. The third is that a strong site-wide link profile makes you authoritative on every topic, when answer-engine authority is topic-specific. The fourth is confusing link building with reputation building, when the latter is what earns AI authority. Optimise for trust and mentions, and treat links as a classic-ranking input rather than your AI-authority strategy.
The bottom line
How do answer engines evaluate authority compared to PageRank? PageRank scores authority from the link graph; answer engines score it from mentions, topical trust, corroboration, and how well you answer, treating raw links as a weak direct signal. The data is clear that mentions predict AI visibility far better than backlinks do, so answer-engine authority is reputation, not a link tally. Keep enough classic authority to rank, because ranking feeds AI, but pour your real effort into being genuinely trusted, mentioned, and expert on your topic. Build the reputation, not just the link profile, and answer engines will treat you as the authority you are.
Frequently asked questions
How do answer engines evaluate authority compared to PageRank?
PageRank scores authority mainly from the link graph, treating links as weighted votes. Answer engines lean instead on topical and entity trust, corroboration across many sources, consistency, and how directly you answer, with raw link metrics a weak signal. So authority to an answer engine is closer to reputation than to a link count.
Do backlinks still matter for answer-engine authority?
Indirectly. Backlinks help you rank in classic search, and pages that rank well are more likely to be pulled into AI answers, so link-earned authority feeds AI visibility through ranking. But links are a weak direct signal of authority to answer engines, which weight mentions and topical trust far more.
What signals do answer engines treat as authority?
Being mentioned and discussed across many credible sources, consistent topical focus, corroboration between independent sources, and directly and accurately answering questions in your area. These reputation-and-relevance signals matter more to answer engines than the raw quantity or quality of backlinks that PageRank rewards.
Why does chasing backlinks underperform for AI visibility?
Because answer engines do not treat the link graph as their main authority signal. A large study found backlink metrics correlate weakly with AI visibility while mentions correlate strongly, so effort poured purely into link building buys little AI authority compared with earning genuine topical trust and mentions.
