All articles
link buildingAI searchdigital PR

Do Backlinks Still Matter When Buyers Use AI to Research Vendors?

Backlinks still matter, but brand mentions now correlate with AI citations 3x more. Here's what the data says and where to spend your effort.

On this page

The question lands in my inbox constantly: do backlinks still matter now that buyers ask ChatGPT which vendor to pick? Yes, but the job has changed. Raw link counts move Google less than they used to, and AI assistants barely read the link graph at all. They respond to the thing links were always a proxy for: being known, cited, and verified across the web. Recent research on brand mentions vs backlinks shows mentions track AI visibility roughly three times more strongly. Keep building links. Just stop treating the count as the goal.

What the evidence actually shows

Every major correlation study of AI visibility lands in the same place: how often your brand gets mentioned across the web predicts whether AI systems cite and recommend you. Backlinks barely register in comparison.

A brand mention is any reference to your company across the web (news coverage, Reddit threads, YouTube reviews, industry listicles) whether or not anyone bothered to hyperlink to you.

Ahrefs ran the biggest test of this. In May 2025, they analyzed 75,000 brands to find what correlated with visibility in Google’s AI Overviews. Branded web mentions came out on top with a 0.664 correlation. Backlinks scored 0.218, and referring domains (the metric link builders worship) scored 0.295 (Ahrefs, May 2025).

In December 2025, they repeated the exercise across ChatGPT, AI Mode, and AI Overviews. Same story. Branded mentions correlated between 0.656 and 0.709 depending on the platform, while the number of backlinks and URL rating were “very weak” across every AI system tested (Ahrefs, December 2025).

The raw gaps are striking. In the May study, brands in the top quartile for web mentions averaged 169 AI Overview mentions. The next quartile down averaged 14. And roughly 26% of brands had zero AI Overview mentions at all, completely invisible in a channel where about half of B2B software buyers now start their research in an AI chatbot instead of Google, per G2’s survey of 1,000+ buyers (G2, October 2025).

Semrush and Kevin Indig attacked the question from a different angle in October 2025, comparing backlink profiles against citations across five AI platforms for 1,000 domains. Their finding: link authority matters, but with a floor. Authority Score correlated 0.65 with AI mentions overall, yet the relationship with AI share of voice mainly kicked in past a minimum authority threshold. Below it, incremental links bought very little (Semrush, October 2025).

And Google’s own position? Unchanged. John Mueller said back in November 2022 that “over time, the weight on the links at some point will drop off a little bit,” while adding that links “will always be something that we care about because we have to find pages somehow” (Search Engine Journal, November 2022). Two years into the AI search era, that has aged well.

Ahrefs reports a correlation. The study can’t tell you whether earning more backlinks will get you cited more often. Big, trusted brands have lots of mentions and lots of links, so mentions may simply be a marker of being well-known rather than the mechanism that gets you cited. But when one signal correlates three times more strongly, and the pattern repeats across platforms, sample sizes, and independent teams, I pay attention.

Here’s one more data point that weakens the “links help AI indirectly through rankings” argument. In August 2025, Ahrefs put 15,000 prompts to both Google and four AI chatbots, then checked where the cited URLs ranked. Only about 12% of URLs cited by ChatGPT, Gemini, and Copilot also ranked in Google’s top 10. Roughly 80% didn’t rank for the original query at all. AI systems retrieve sources in a fundamentally different way than a SERP does (Ahrefs, August 2025).

Why do mentions beat links here? Language models don’t rank pages, they compose text. When an assistant builds a vendor shortlist, it’s reconstructing how humans talk about the category: who comes up, in what context, next to which competitors. A link is a machine-to-machine vote. A mention is the raw material of the answer itself. That’s also why unlinked mentions count. The model reads the sentence, not the href.

The two currencies of authority

Think of authority signals as two currencies: one Google still trades in, one AI systems trade in. They overlap less than most SEO budgets assume.

The split exists because the two systems do different jobs. Google ranks documents. It needs machine-readable votes to sort billions of pages. AI assistants answer questions. They need to reconstruct what humans already say about a topic. Different jobs, different signals, which is why a 500-link campaign and a well-known brand are no longer the same purchase.

SignalWhat it still influences (Google rankings)What it influences in AI answersEffort to earn
BacklinksStill a real, quality-weighted input; strongest on competitive commercial SERPsWeak direct effect on LLM citations; helps indirectly via rankings and crawl discoveryHigh
Brand mentions (linked or unlinked)Minor direct ranking valueStrongest correlate of AI visibility, 0.66 to 0.71 in Ahrefs’ December 2025 dataMedium
ReviewsLocal pack rankings and trustFeeds the sentiment and reliability judgments AI answers are built onLow to medium
Original researchEarns links naturally over timeGets cited directly; Princeton’s GEO paper found citations, quotations, and statistics boosted generative-engine visibility by up to 40%High
PR coverageBuilds authority and branded search demandPuts your name in the publications AI models quote mostHigh

Read that table right to left and the pattern is clear. The signals AI answers respond to are mostly things you earn by being present in a market, not by sending outreach emails asking for anchor text.

Backlinks aren’t dead money. They still pay out in three places, fewer than before, and not where most link builders are spending their days.

  1. Bottom-of-funnel rankings. The scary click-through numbers come mostly from informational queries. Ahrefs measured the CTR drop for the #1 result when an AI Overview appears at 34.5% in their April 2025 study, worsening to 58% by February 2026, but AI Overviews skew heavily toward informational searches. When a buyer types “vendor X pricing” or “best CRM for recruiting agencies,” they’re often still looking at ten blue links, and ranking there still wins the click (Ahrefs, February 2026).

  2. Discovery. Links remain how crawlers (including AI crawlers) find pages in the first place. Mueller again: “we have to find pages somehow.” A new site with no links pointing at it is a new site nobody finds.

  3. Referrals and borrowed trust. A link from a publication your buyers actually read does double duty: some clicks, plus an association that outlasts any single ranking. That value predates AI search and will outlast it too.

I want to be honest about the uncertainty here. Nobody has clean counterfactual data, AI referral traffic is still a small slice of most sites’ sessions, and I wouldn’t blow up a budget over any single correlation study. The direction is consistent across studies, though. That’s what I act on.

A note on that 58% figure too: Ahrefs itself points out that other teams’ measurements range from the high forties to the high eighties, depending on keyword mix and method. Read it as “the click you used to get is at risk,” not as a precise tax rate. And read the whole section as: rankings still matter most where AI answers haven’t taken over the query, which today means bottom-of-funnel.

Reallocate. Earn links as a byproduct of being worth mentioning, and put most of your effort into the mentions layer that AI systems actually read.

  1. Do digital PR with data, not press releases. Digital PR is earning coverage in real publications with something newsworthy, original data, sharp expert commentary, contrarian analysis. A proprietary survey gets you covered, linked, and mentioned. One campaign produces all three currencies in the table above, which is why it beats a monthly quota of guest posts.

  2. Show up where the models already look. Reddit, YouTube, niche communities, industry newsletters. Search Engine Land reports Reddit appears in about 3% of AI citations, and over 80% of those answers address decision-seeking prompts like brand comparisons (Search Engine Land, July 2026). A helpful, non-salesy presence in the threads where your category gets discussed is worth more than a niche-edit link on a random blog.

  3. Publish original research others reference. The Princeton GEO paper found that adding citations, quotations, and statistics to content boosted visibility in generative engines by up to 40% (Aggarwal et al., KDD 2024). Original data gives AI systems something concrete to cite and gives journalists something to link. It’s the rare tactic that serves Google rankings and LLM citations at the same time.

  4. Get into the comparison sets. G2’s research found AI chat is now the top source buyers use to build a software shortlist, ahead of review sites and vendor websites. Models compose “best X” answers from the listicles, marketplaces, and comparison pages they retrieve. Make sure you exist in those sets: keep review profiles current, comparison listings accurate, and category pages aware of you. Then run the ten prompts your actual buyers would run and note which brands and sources come back. That list is your gap analysis, and it costs nothing.

  5. Keep a floor of strong links. Don’t zero out the link budget. Competitive bottom-of-funnel terms still require authority to rank, and links still help crawlers find and contextualize your content. Link building in 2026 looks less like a spreadsheet of placements and more like a byproduct of being notable, but the floor still needs to hold.

The uncomfortable part

Here’s my opinion, and it won’t please everyone selling links: most link building was always brand-building wearing a technical costume. The good version of outreach (the kind that worked) was really “get credible people to vouch for you.” The bad version was gaming a proxy signal.

AI made the proxy visible. When rankings were the only scoreboard, you could hide behind Domain Rating and anchor ratios. Now the scoreboard asks a question with no tricks available: do people talk about you when they discuss this problem? If the answer is no, no volume of links will make a language model say your name, because models echo the corpus of human conversation, not the link graph.

That’s uncomfortable because mentions can’t be bought at scale the way links could. You can’t order forty authentic Reddit threads. You earn them with things worth talking about, or you don’t exist in the answer.

FAQ

Yes, for Google rankings, especially on competitive commercial terms where you’re fighting established domains. What’s changed is their share of the job. They’re a floor, not a strategy.

Directly, only weakly. Ahrefs’ data shows link metrics correlate far less with AI visibility than brand mentions, 0.218 versus 0.664 in the May 2025 study, and “very weak” again in December. Indirectly, through rankings, crawl discovery, and the coverage links come bundled with, somewhat more.

Brand mentions, review presence, original data, and third-party comparisons. The Ahrefs studies put branded web mentions at 0.66 to 0.71 correlation with AI visibility across ChatGPT, AI Mode, and AI Overviews, roughly three times backlinks.

Invest in being link-worthy, not in link counts. Digital PR, original research, and data-led campaigns earn links, mentions, and citations simultaneously. Buying placements can still move rankings in some verticals; it does almost nothing for AI answers.

Not mainly. Only about 12% of URLs ChatGPT cites rank in Google’s top 10 for the prompt, per Ahrefs’ August 2025 analysis. Being crawlable, clearly structured, and mentioned in the sources ChatGPT retrieves matters more than your backlink profile does.

Does this mean domain authority scores are useless?

No. They still help you rank, and Semrush’s data suggests authority works as a threshold for AI visibility, low-authority domains saw little payoff from extra links. Think of DR as a ticket to the game, not the final score. Enough authority gets you considered; mentions decide whether you get named.

Sources

Questions & answers

What is "Do Backlinks Still Matter When Buyers Use AI to Research Vendors?" about?

Backlinks still matter, but brand mentions now correlate with AI citations 3x more. Here's what the data says and where to spend your effort.

Who wrote this article?

Aditya Mallah is a growth marketer for SaaS, AI tools, and fintech. Full bio: https://adityamallah.com/about

About this article

The content on this blog is for educational and informational purposes. These are my own views and do not constitute professional, financial, legal, or business advice. I can be wrong; do your own research before making decisions. Third-party links are provided for convenience and do not imply endorsement.

Was this article helpful?

Get my growth notes

Notes on growth, AI, and finding customers. Delivered to your inbox.

All articles

B2B marketing 12 min read

If I Had to Grow a B2B Brand from Zero Today, Here's My 90-Day Plan

Here's the exact 90-day plan I'd use to grow a B2B brand from zero: fix positioning, pick one channel, build compounding assets, and run sales yourself.

GEO 12 min read

Best GEO Software for Brands: Comparing AI Visibility Tracking Platforms

Compare the best GEO software of 2026 by price, engine coverage, and limits, with picks for small budgets, mid-market teams, and enterprise brands.

SEO tools 13 min read

9 Content Optimization Tools for Improving Search Rankings and AI Discoverability

Compare nine content optimization tools by price, content features, and support for tracking AI search mentions.