Do AI Overviews Use Backlinks? What the Data Actually Says (2026)

Every SaaS team watching AI Overviews eat their clicks asks the same question before renewing a link budget: if Google now answers the query itself, do the links we buy still do anything?

It is a fair worry. The old deal was simple, you earned links, you ranked, you got the click.

When an AI answer sits on top of the results and takes the click, it is reasonable to ask whether the link underneath still matters.

The short version is that it does, just not in the way headlines suggest.

AI Overviews do not have a separate "backlink score" you can game, and links are no longer the single strongest signal. But AI Overviews are not a separate search engine.

They are built on the same index and the same ranking systems where backlinks have always counted, so links still shape what gets pulled into an answer.

Here is exactly how, with the real numbers.

Key takeaways

• Yes, indirectly. AI Overviews are generated from Google's normal index and ranking systems, and Google has confirmed there is no special "AI SEO" separate from regular SEO. Backlinks influence AI Overviews the same way they influence rankings: by getting pages found, indexed, trusted, and ranked well enough to be used as a source.
• In Ahrefs' study of 75,000 brands, link signals correlate positively with AI Overview visibility: Domain Rating at 0.326 and referring domains at 0.295, both higher than raw backlink count at 0.218. Brand mentions were stronger (0.664), but links are not zero.
• Links do the plumbing. A page that never gets indexed cannot be cited in an AI Overview, and links are still how pages get discovered, crawled, and trusted enough to qualify.
• Backlinks and brand mentions are not rivals. The same outreach that earns a link on a credible site also earns the mention AI systems reward, so you build both from the same work.
• For SaaS, the winning move is unchanged, only sharper: fewer, more relevant links on real-traffic sites, plus content worth quoting.

The short answer

Yes, AI Overviews use backlinks, but indirectly. There is no dedicated link score inside the AI Overview.

Instead, Google builds each AI Overview from its core index and ranking systems, and backlinks are still part of those systems.

So a link helps an AI Overview the same way it helps a normal ranking: it gets your page discovered and indexed, it lifts your authority, and it helps the page rank well enough to sit in the pool of trusted sources the AI Overview draws from.

Backlinks are no longer the top-correlated signal (brand mentions are), but they remain part of how a page becomes eligible to be cited at all.

For the full comparison of the two signals, see backlinks vs brand mentions for AI search.

How AI Overviews actually pick sources

To know whether links matter, you first have to know where an AI Overview gets its information.

It does not invent an answer from nothing, and it does not run a separate crawl of the web with its own rules.

An AI Overview is grounded in Google Search. It uses the same crawler (Googlebot), the same index of pages, and the same core ranking systems as the normal blue links.

When you ask a question, Google runs a technique it calls query fan-out, breaking your question into several related searches, then builds the summary from pages that surface for those searches.

In practice, the sources an AI Overview cites are drawn heavily from pages that already rank well for the underlying queries.

Google has been unusually direct about this.

In July 2025, Google's Gary Illyes said there is no separate discipline of AEO or GEO you need to rank in AI features, because they run on the same systems as Search.

As Search Engine Journal reported on July 24, 2025, the official line is essentially that good SEO is AI search optimization.

That single fact answers the headline question: if AI Overviews are built on the same ranking systems as Search, and backlinks are a factor in those systems, then backlinks feed AI Overviews too.

That is the mechanism. Now for how strong the effect actually is.

What the data says about backlinks and AI Overviews

The most-cited study on this is Ahrefs' analysis of AI Overview brand visibility factors, published May 26, 2025.

It looked at 75,000 brands and measured how strongly different signals correlate with a brand appearing in Google AI Overviews, using Spearman rank correlation:

Signal	Correlation with AI Overview visibility
Branded web mentions	0.664
Branded anchors	0.527
Branded search volume	0.392
Domain Rating (DR)	0.326
Referring domains	0.295
Branded traffic	0.274
Number of backlinks	0.218
URL Rating (UR)	0.180

Read that table carefully, because the popular takeaway skips the useful part. Yes, branded mentions (0.664) correlate about three times more strongly than raw backlink count (0.218).

But three of the top signals are link-derived: branded anchors (0.527) come from links, Domain Rating (0.326) is built from your link profile, and referring domains (0.295) are literally a count of linking sites.

Every one of those correlates positively with AI Overview visibility, and DR and referring domains both beat raw backlink count.

That is the same lesson experienced link buyers have repeated for years: the quality and diversity of your links matter more than the sheer number.

Fifty referring domains from real, relevant sites do more than five hundred links from one network. The Ahrefs data says AI Overviews reward exactly that shape of link profile.

(Two cautions on this study: it is a correlation, not proof of causation, and it measures Google AI Overviews specifically, not ChatGPT or Perplexity. Anyone applying it to every AI tool is guessing.)

Why links still matter inside an AI answer

Correlation tables are one thing. The plainer argument for links is mechanical, and it does not depend on any single study.

A page has to be indexed before it can be cited. An AI Overview can only pull from pages Google has crawled and stored in its index.

If your page is not indexed, it does not exist as far as the AI Overview is concerned.

Backlinks are still one of the main ways new pages get discovered and crawled in the first place, which is why getting your links and pages indexed is the floor, not a nice-to-have.

A link a model never sees cannot help you, and neither can a page it never finds.

Authority still decides which pages are in the running. AI Overviews draw from pages that rank, and ranking still runs on the systems where links carry authority.

Google's Gary Illyes said in 2024 that Google now needs "very few links" to rank a page well (reported April 22, 2024).

Notice that is an argument for fewer, better links, not for none. You still need enough authority, from the right sites, to be in the pool the AI Overview selects from.

For the wider evidence that links keep pulling their weight, see do backlinks still matter for SaaS SEO.

So the chain is: link, then crawl, then index, then rank, then eligible to be cited. Remove the link and you weaken every step after it.

Links are how brand mentions get made

Here is the part that dissolves the "should I switch my budget from links to PR" debate. The data says brand mentions are the stronger signal for AI Overviews.

Fine. Now ask the practical question: how does a SaaS brand actually earn credible, on-topic mentions?

You do not get mentioned on a respected industry publication by wishing for it.

You get mentioned because you ran a digital PR campaign, published original research a journalist cited, earned a guest placement, or made a "best tools" roundup.

Every one of those activities produces a mention, and almost always a link at the same time, from the same outreach. The link and the mention are two outputs of one effort.

That is why the smart reading of the AI Overview data is not "stop building links." It is "put your links on credible, relevant sites, because that is where the mentions count too."

We make the full version of that case in backlinks vs brand mentions for AI search and, for the AI-citation angle specifically, in generative engine optimization for SaaS.

Why being cited matters more than ever

This question feels urgent because AI Overviews are no longer a fringe feature.

Semrush's AI Overviews study (updated December 15, 2025, across more than 10 million keywords) tracked the trigger rate rising from 6.49% of searches in January 2025 to a peak of 24.61% in July, then settling near 15.69% in November as Google tuned where it shows them.

A large and growing share of searches now show an AI answer above the normal results.

And those answers take the click. Pew Research Center analyzed 68,879 real searches from 900 US adults (published July 22, 2025, on March 2025 data).

When an AI summary appeared, people clicked a normal result just 8% of the time, versus 15% when there was no summary, and only 1% clicked a link inside the summary itself.

Put those together and the stakes are clear.

If a buyer researches your category through an AI Overview and your brand is not one of the sources it trusts, you are invisible at the exact moment they are deciding, no matter how well you rank in the normal results below.

Being cited is becoming its own goal, and the pages that get cited are the ones that were indexed, ranked, and trusted, which brings you right back to links.

What a source AI Overviews trust looks like

If citations only go to credible, on-topic pages, the practical question is what makes a source credible enough to be pulled in.

We can answer that with real data, from the publisher sites we vet for SaaS link placements.

Across the SaaS publishers currently active in the Saaslinks catalog, the sites that pass vetting share a clear profile:

• 101 vetted SaaS publisher sites, with an average Domain Rating of 50. Fourteen sit at DR 70 or above, and most of the rest land in the healthy DR 30 to 69 range (the full spread runs from DR 21 to DR 91). These are established publications, not inflated metrics.
• 83% carry a spam score of 5 or lower. Low spam is not a vanity number here. It is the clean, trusted signal that keeps a site in the pool AI engines actually draw from.
• Together the catalog pulls more than 590,000 organic visits a month and ranks for over 212,000 keywords. That is real audience and real topical relevance, not parked domains dressed up with numbers.
• The catalog sits exactly where SaaS buyers need mentions and links: the biggest niches are SaaS, B2B, tech, martech, AI, e-commerce, and devtools.

That data points to the same conclusion the Ahrefs study does: relevance and credibility beat raw volume.

A DR 45 publication that genuinely covers your category, with real traffic and a low spam score, is a better home for a link (and the mention that rides along with it) than a DR 70 generalist an AI Overview has no reason to connect with your topic.

Topical fit is exactly what a SaaS-only catalog is built to deliver, which is why we show every metric with its source and refresh date instead of asking you to trust a number.

For how to read those metrics yourself, see how to judge a link before buying.

The SaaS playbook for AI Overview visibility

Here is the practical sequence. None of it is exotic, which is the point: AI Overviews reward good SEO, done with more discipline.

1. Get the page indexed and eligible. If Google has not crawled and indexed the page, no AI Overview can cite it. Fix internal linking, and treat indexation as the first checkpoint, not an afterthought.
2. Earn fewer, better, relevant links. Put topical fit and real traffic ahead of headline DR. This is what lifts your pages into the pool AI Overviews select from, and it is the same work that earns the mentions AI answers reward.
3. Make your pages worth quoting. AI engines favor content they can lift cleanly. The Princeton and Georgia Tech GEO study (arXiv, November 2023) found that adding statistics, direct quotations, and cited sources can raise visibility in generative engines by up to roughly 40%. Original data, clear structure, and clear sourcing are the cheapest AI-visibility wins you have.
4. Build a named presence on purpose. Podcasts, expert roundups, community answers, and "best X tools" lists all connect your brand to a topic in the model's view, with or without a link.
5. Measure the right thing. Track rankings and referral traffic, but also watch whether your brand shows up in AI Overviews for your core questions, and which sites get cited when it does. Then work to be on those sites.

The bottom line

Do AI Overviews use backlinks? Yes, indirectly, and that "indirectly" is the whole story.

AI Overviews are not a separate engine with their own link score.

They are built on the index and ranking systems where backlinks have always mattered, so a link still earns its keep by getting your page found, indexed, trusted, and ranked into the pool of sources an AI answer can cite.

Brand mentions may be the stronger single correlate, but you earn those on the same trusted sites, through the same outreach, that produce your links.

That is the bet Saaslinks is built on. We do not compete on catalog size, because size was never the signal.

We list SaaS-relevant publications with real, source-dated organic traffic and low spam scores (the kind of credible sites AI answer engines actually pull from), and we back every placement with a 30-day indexation guarantee, because a link or a page a model never sees cannot help you.

To see how that works in practice, here is how a link-building marketplace works.

In the AI-answer era, the job is the same as it always was, only sharper: be the credible, well-linked, often-mentioned source your category keeps citing.