Backlinks have been the default shorthand for online authority for so long that it is tempting to treat them as the master switch for AI visibility. The evidence is more complicated. Links still help search engines discover, understand, and rank pages, and those outcomes can improve the odds that an AI system retrieves a page. But a backlink is not the same thing as an AI citation, and current studies do not support treating raw link volume as the one dominant lever across ChatGPT, Google AI features, and Perplexity.
For marketers learning what AI citations are, the useful distinction is simple: backlinks help build discoverability and authority; citation selection also depends on whether a page is accessible, relevant to the specific question, clear enough to extract, well supported, current enough for the topic, and consistent with other credible sources.
TL;DR
Backlinks still matter, especially high-quality links that contribute to authority and organic search visibility. They are no longer a sensible stand-alone proxy for AI citations. Recent industry studies associate AI visibility with domain authority, rankings, branded web mentions, query relevance, and source quality, while academic evidence shows that well-supported, clearly presented content can improve visibility after retrieval. The practical move is not to abandon link building. It is to stop measuring success by link count alone and build a broader authority system around useful content, genuine mentions, technical access, topical coverage, and repeatable citation tracking.
The short answer is no
There is no universal, platform-independent ranking factor called “the biggest citation driver.” Google says its generative search features are rooted in core Search ranking and quality systems, including retrieval-augmented generation and query fan-out. OpenAI says ChatGPT Search uses multiple factors to rank relevant, reliable results and that placement is not guaranteed. Both descriptions point to a pipeline: a page must be crawlable and retrievable before the system can decide whether it is useful evidence for a particular answer.
That is why backlinks remain valuable without being sufficient. Google still lists link analysis and PageRank among its ranking systems, but Google’s guidance for generative AI features (source) emphasizes the wider set of SEO fundamentals rather than a special link formula. The right conclusion is not “links are dead.” The right conclusion is “links are one part of retrieval and authority, not a guaranteed ticket into the answer.”
What the latest evidence says
Quality and authority beat raw backlink volume
A Semrush study of 1,000 domains (source) found a strong positive relationship between its Authority Score and AI mentions: Pearson correlation of 0.65 and Spearman correlation of 0.57. The gains appeared more clearly in higher authority tiers, which matters because Authority Score is a composite proxy, not a count of links. Semrush’s own takeaway was that quality, referring-domain diversity, and credibility matter more than collecting links in bulk.
Brand presence can correlate more strongly than backlinks
In an Ahrefs analysis of 75,000 brands (source), branded web mentions showed a Spearman correlation of 0.664 with brand visibility in Google AI Overviews. The number of backlinks showed a much weaker 0.218 correlation. Branded anchor text landed between them at 0.527. This does not prove that mentions cause AI visibility, and the outcome measured brand mentions rather than source citations. It does show why a link-only strategy misses important off-site signals.
Traditional rankings help, but citation selection reaches beyond page one
Ahrefs later analyzed four million Google AI Overview URLs and found that 37.9% of cited URLs also appeared in the first 10 search-result blocks (source) for the same query. Another 31.2% appeared in positions 11–100, while 31.0% did not appear in the top 100 for the direct query. Query fan-out offers a reasonable explanation: a page can be retrieved for a related subquery even when it does not rank for the exact wording the user entered.
Content evidence matters after retrieval
The foundational Generative Engine Optimization study (source) found that adding relevant citations, quotations, and statistics improved source visibility in its benchmark. That experiment evaluated content already available to the generative pipeline, so it does not prove that those edits cause a page to be discovered organically. A 2026 critical survey of GEO research (source) makes the same caution explicit: evidence is stronger for influencing selection after retrieval than for producing stable, cross-platform gains in organic discoverability or business outcomes.
How backlinks fit into the citation system
| Signal | What it can help | What it cannot guarantee |
|---|---|---|
| Relevant backlinks | Discovery, authority, ranking support, referral context | Selection for a specific AI answer |
| Brand mentions | Entity recognition and corroboration across the web | A source link or favorable description |
| Organic visibility | Retrieval across direct and fan-out queries | Citation when a better passage answers the subquery |
| Evidence-rich content | Extractable claims, statistics, definitions, and support | Discovery if crawlers cannot access the page |
| Freshness and access | Eligibility for current, time-sensitive answers | Authority or accuracy by themselves |
The table explains why link building still earns a place in the plan. A credible backlink can support several upstream conditions at once. But the final citation decision happens at the level of the query, passage, platform, and moment in time. Understanding how AI citations work makes it easier to diagnose which part of that chain needs attention.
What This Means for Marketers
Backlinks are still an authority asset. They are just a lousy single-number dashboard for AI visibility. If a campaign earns relevant editorial links, real brand mentions, referral traffic, and stronger coverage of a topic, keep doing it. If the campaign exists to increase a link counter with placements nobody reads, it is probably optimizing the receipt instead of the result.
A practical citation strategy for marketers
1. Protect the SEO foundation
Keep important pages crawlable, internally linked, indexable, fast, and available in clear HTML text. Earn relevant links where they improve discovery, authority, or qualified referral traffic. Avoid treating third-party authority scores as if they were platform rules.
2. Create evidence worth citing
Answer a defined question directly. Use precise definitions, attributable data, transparent methodology, useful comparisons, and dates where freshness matters. Cite original evidence rather than recycling claims through several blogs.
3. Build a wider web presence
Pursue genuine coverage, expert contributions, relevant communities, video, partnerships, and customer conversations that create consistent brand context. Some of that coverage will link. Some will not. Both can help people and machines understand what the brand is known for.
4. Measure citations separately from rankings and links
Track a stable set of prompts by platform, date, market, and intent. Record whether the brand was mentioned, whether the page was cited, and whether the citation supported the claim. Use the process in How to Increase AI Citations as a starting point, then compare changes over repeated runs rather than reacting to one answer.
5. Judge link opportunities by the whole outcome
Before pursuing a placement, ask whether it reaches the right audience, adds credible topical context, can send referral traffic, and strengthens the brand’s reputation. A useful editorial mention that also includes a link is better than a link built only to exist.
Frequently Asked Questions
Do backlinks help a page get cited by AI?
They can help indirectly by improving discovery, authority, organic visibility, and the context surrounding a brand or page. Current public evidence does not support a guaranteed one-to-one effect.
Are unlinked brand mentions useful?
Potentially. Large correlation studies associate broad brand mentions with AI visibility, especially in Google’s AI features. Correlation does not prove causation, and effects differ by platform and query.
Should marketers stop traditional link building?
No. Focus it on relevant, reputable placements that support real visibility and referral value. Stop treating raw volume as the goal.
What should be measured alongside backlinks?
Track referring-domain quality, branded mentions, organic visibility, crawl access, citation frequency, citation accuracy, brand mentions inside answers, and downstream traffic or conversions.
Can a page without many backlinks earn AI citations?
Yes. Query fan-out, passage relevance, freshness, source diversity, and evidence quality can bring less prominent pages into an answer. Results vary by platform, and no optimization guarantees selection.
The Practical Takeaway
Backlinks still deserve investment when they represent real relevance and reputation. They simply share the job now. Build pages that search systems can retrieve, answer engines can understand, credible sources can corroborate, and humans would choose even if an authority score never existed.