There is a version of this question that gets asked every few years, usually right after some algorithmic update or industry shift and it almost always leads to the same comfortable answer – Yes! backlinks still matter. That answer has held up really well for nearly two decades of SEO evolution and it is tempting to give it again here.

But the question being asked now is meaningfully different from the one being asked in 2015 or even 2022 because the system doing the evaluation now has changed.

When someone types a query into ChatGPT or Perplexity and asks which analytics tool a growth team should consider or what the best onboarding platforms are for mid-market SaaS companies, no PageRank calculation is running in the background. The model is not crawling links – it is pattern-matching across a training dataset and a retrieval layer that weights signals in ways that diverge from Google’s algorithm in some deeply important respects.

That divergence is not a reason to abandon link building – it is a reason to understand more deeply what backlinks actually do in an AI search context, what they no longer do, and what has moved into the role they once monopolized.

Why the Question Feels More Urgent Than It Did Two Years Ago?

Part of what makes this worth examining closely is that the stakes have shifted.

As long as AI-generated answers were a novelty or a small part of discovery, the answer barely mattered for GTM strategy. But now, AI search is far from being considered just a novelty and is growing so fast that forward-looking teams are forced to treat AI visibility a thing to act on right now and not just keep it as a future consideration.

More importantly, the category of AI search is broader than people initially assumed.

It includes AI Overviews embedded in Google’s standard results page, conversational interfaces like ChatGPT and Perplexity where retrieval happens in real time and the foundational training data layer that informs how models respond even when no live search is involved. These three layers behave differently  and backlinks interact with each of them in different ways. Treating them as a single system is a mistake and can push you in the wrong direction.

What Backlinks Actually Do in AI Search?

The most precise way to describe the current relationship is this – backlinks are still a meaningful input but they are increasingly an indirect one.

In the retrieval-augmented layer (RAG) that Perplexity and ChatGPT search use when they crawl the live web, domain authority matters. Pages from high-authority domains get retrieved more often which means they get considered more often. Research published by Search Engine Journal found that 73% of SEO practitioners believe backlink quality directly affects visibility in AI search results and there is enough supporting data to take that perception seriously rather than dismissing it as industry inertia. The number one organic result still carries roughly three times more backlinks than pages ranked two through ten and that gap matters when AI systems are filtering for trustworthy sources.

But here is where this understanding starts to break down.

BrightEdge’s analysis of citation patterns in Google AI Overviews found that the overlap between top ten organic rankings and AI Overview citations dropped from roughly 76% in late 2024 to somewhere between 17% and 38% by early 2026, depending on the dataset. The pace of that change is something that was never seen before – in under 18 months the assumption that ranking well in traditional search was sufficient to earn AI visibility stopped being defensible. The two systems now select sources through different mechanisms even when they appear to be answering the same query.

The data on correlation further reinforces this.

Ahrefs’ analysis of 75,000 brands found that brand mentions correlate with AI Overview visibility at 0.664  while backlinks correlate at 0.218, roughly a 3:1 difference. What that gap actually reveals is something most SEO frameworks have not yet articulated cleanly –  backlinks are a navigation signal and brand mentions are a trust signal  and AI systems are built to optimize for trust – not navigation. A crawler follows links to find content while  a language model follows mentions to form a judgment – those are fundamentally different operations and thinking them as one is what leads teams to invest in the wrong layer.

We call this the navigation-trust gap – the growing divergence between what traditional link building was designed to do and what AI retrieval systems actually reward. The wider that gap becomes in a given category, the less predictable traditional SEO authority becomes as a proxy for AI visibility.

The Mechanism Behind That Gap

To understand why brand mentions outperform backlinks as an AI signal, it helps to think about what large language models are actually learning from during training.

LLM models are not just absorbing documents and cataloguing links between them – they are learning patterns of credibility from the full context of how sources are discussed, cited, and referenced across the corpus. A brand that appears across independent analyst reports, practitioner forums, comparison posts and editorial coverage is being described as a credible entity by many different voices. That distributed reinforcement is given more importance by the model in a way that a cluster of backlinks on a single domain does not.

When we looked at what Search Engine Land’s cross-industry citation analysis actually reveals beneath the headline numbers, the finding that stands out is that organic keyword breadth correlates with AI visibility at r=0.41  stronger than backlinks at 0.37. That gap is small in absolute terms but significant in what it signals – AI engines are not primarily rewarding authority accumulation –  they are rewarding coverage patterns.

A site that ranks for 200 queries across a topic cluster is more likely to be cited than a site with twice the backlinks but half the topical range. What this means in practice is that pages ranking sixth through tenth with deep subject coverage get cited more than twice as often as pages ranking first with thin coverage around the same term. The reflex of equating link profile strength with AI visibility is not just outdated –  it is measuring the wrong thing entirely.

What also changes in this model is the cost of absence.

If a competitor is consistently described across community discussions, comparison resources, and independent reviews while your brand is not, that imbalance shapes how AI models understand the category regardless of your domain rating and starts surfacing  them in its results considering their presence as credible.

For B2B SaaS teams specifically, this cost is higher than it appears on a standard analytics dashboard because in categories where the feature set gap between players is small, AI visibility often ends up being decided less by product quality than by how consistently and clearly the broader web describes what you do and who you do it for. In this case, a brand with a tighter, more frequently reinforced description across third-party sources will be recommended over a technically superior product that the model cannot confidently summarize.

Where Backlinks Still Have Real Leverage?

The clearest remaining function of backlinks in an AI search context is foundational trust.

The same Ahrefs analysis of 75,000 brands found that domain authority correlates at r=0.18 with AI citation probability – a figure that sounds weak until you consider what it actually represents at the lower end.

A site with no backlink profile and no external references gives a retrieval system almost nothing to work from when forming a credibility judgment. There are no third-party signals to cross-reference, no pattern of mentions to interpret, no evidence that anyone outside the brand itself has found the content worth referencing. What this tells us is that backlinks function as a credibility floor in AI search rather than a ranking ceiling  – they do not determine how visible you become but their absence can quietly disqualify you before the content is ever evaluated.

But, most of the teams optimizing for AI visibility are focused on the ceiling which is the right instinct but ignoring the floor is how otherwise strong content gets filtered out at the retrieval stage before any of the higher-order signals even come into play.

The second place where backlinks retain meaningful value is through the pages they help rank.

In the hybrid layer where AI systems pull from live web results and combine retrieval with generation, getting a well-constructed piece of content into the organic top 20 still improves its odds of being retrieved and considered. Backlinks that put the right pages in the right positions within traditional search continue to matter but indirectly because the relationship now is sequential –  links help pages rank, ranking helps pages get retrieved, retrieval creates citation opportunity.

The third is what might be called the earned-media architecture.

A backlink from an editorial source is not just a link – it is a brand mention with a navigation signal attached and the mention is often the more valuable part. When a journalist or analyst references your product in a category explanation or comparison, the association between your brand and that problem space gets reinforced in both the traditional search index and the language model’s understanding of the category. That is why digital PR and earned coverage have become more central to AI visibility strategies, not because the links themselves are more powerful, but because the editorial context surrounding those links now carries independent weight.

The Part of the Playbook That No Longer Works

For years, a significant portion of link building strategy involved accumulating volume –  guest posts, directory submissions, syndicated content, link exchanges between non-competing properties. Some of this was always of marginal value for traditional SEO but for AI search, it does not.

The reason is structural as  AI retrieval systems are not counting links but are evaluating whether a page can be trusted to answer a specific question clearly and completely. A guest post on a mid-tier industry blog that contributes a referring domain to your profile does nothing to change how a language model interprets your content if the page it points to is thin, generic or inconsistently positioned.

What we are seeing across brands that come to GeoRankers for visibility audits reinforces this pattern consistently. Companies with backlink profiles built heavily around volume tactics –  high referring domain counts, broad anchor text distribution, wide syndication networks are not showing up in AI-generated answers at rates that reflect that investment.

Meanwhile, brands with leaner link profiles but tighter topical coverage, cleaner product documentation and stronger third-party mention density are being cited regularly. The gap between what a backlink audit shows and what an AI visibility audit shows for the same brand is one of the more revealing diagnostics we run  because the two profiles can look completely inverted – strong on links, weak on citations or the reverse.

What Has Moved Into the Structural Role Backlinks Occupied

The most useful reframe here is to stop asking whether backlinks matter and start asking what job they were doing that now needs to be done in a different way.

Backlinks in traditional SEO served three functions simultaneously – they signalled authority (someone credible enough to have a linked-to domain endorsed this page), they built navigability (the crawl graph could discover and return to this content) and finally, they created category association (if high authority CRM content links to your page about CRM migration, that context rubs off).

In AI search, those three functions have partly disaggregated.

Muck Rack’s analysis of over one million AI citations found that 94% came from non-paid, third-party sources rather than brand-owned content – which means the authority function has shifted decisively to earned media and independent validation.  Category association is now built through topical depth, clear ICP documentation and the consistency with which your brand appears alongside specific problem descriptions across the web.

What this creates is a situation where brands that built authority primarily through link acquisition now need to audit whether that authority is legible to AI systems or whether it is only legible to Google’s algorithm – those are not the same thing!

A Practical Orientation for Teams Thinking Through This

Rather than a checklist, the more useful exercise is a diagnostic.

Run a set of category-level queries across two or three AI interfaces without including your brand name and see where you appear, how you are described and whether the description reflects your current positioning. Then, look at where your main competitors appear and what language the models use to describe them. That gap, between your AI-facing narrative and your intended positioning is usually not a backlink problem – it is a brand mention problem, a content depth problem or a consistency problem, which is to say that the same brand name being described differently across the sources the model has ingested.

Backlinks remain part of the solution because they contribute to domain credibility and help high-quality content get discovered by retrieval systems. But treating them as the primary lever for AI visibility is like trying to solve a distribution problem by improving your warehouse – the warehouse still matters but it is not where the constraint lives anymore.

The strategic question worth finding an answer to is not whether to build links but whether the content and brand presence those links point to is doing enough work to be understood and cited in a search layer that now evaluates sources on very different terms than the one SEO was built around.

Frequently Asked Questions

Do backlinks influence AI search results at all?

Yes, but indirectly. Backlinks contribute to domain authority, which affects how often pages get retrieved by AI systems that pull from live search. They also help high-quality content rank in traditional search, and ranking improves the odds of retrieval. The mechanism is sequential rather than direct, and the correlation between backlink volume and AI citation is significantly weaker than the correlation between brand mentions and AI citation.

What signals matter more than backlinks for AI visibility?

The strongest predictors are off-site factors that backlink tools were not built to measure  – third-party brand mentions in editorial and community contexts, topical coverage breadth, cross-platform presence on review sites and industry forums and author credibility signals.

Should I stop link building and focus on AI search optimization instead?

No. Traditional SEO and AI visibility optimization are not conflicting objectives  but complimentary layers. A robust backlink profile shows the domain’s reputation and AI systems inherit that through retrieval. What changes is the allocation –  less focus on link volume and more on the editorial quality of coverage, the depth of content that links refer to and the broader consistency of your brand presence across third party sources.

Why do some low-authority sites get cited more often than high-authority ones in AI answers?

Because AI algorithms analyze content for extractability, clarity and topical specificity, not just domain authority. A site that has a strong direct-answer format, well-structured headers, and specialized credentialed authorship can outrank a higher-authority competitor with thin or generically optimized material. The retrieval mechanism encourages clarity and reliability of content, not only the reputation of the hosting site.

How do I know if my backlink profile is helping or not helping my AI visibility?

Run a dual audit: check your organic rankings for category-level terms (where link authority contributes) and separately run a set of natural-language queries across ChatGPT, Gemini, and Perplexity to see where your brand appears in AI-generated answers. If you are doing well in traditional search but rarely appearing in AI responses, the problem is typically content structure, topical depth or third party brand presence, not links.