If ranking no longer guarantees you'll be cited — and the data says it doesn't — then the obvious next question is: what does? What makes a large language model reach for one brand's page instead of another's when it builds an answer?
For once, this isn't guesswork. AI search is one of the few areas with genuine peer-reviewed research behind it, plus large-scale platform datasets that point the same way. Here's what the evidence actually shows.
The foundational study: GEO
The anchor is the GEO study — "Generative Engine Optimization" — by Aggarwal et al., from Princeton, Georgia Tech and IIT Delhi, published at KDD 2024. The researchers tested how different content changes affected a page's visibility in generative answers, across 10,000 queries and 25 domains, validated on Perplexity.
The results are unusually clean:
Lift in AI visibility by content change (GEO study)
| Content change | Change in visibility |
|---|---|
| Add quotations | +41% |
| Add statistics | +32% |
| Add citations / sources | +30% |
| Improve fluency | +28% |
| Keyword stuffing | −10% |
Source: Aggarwal et al., GEO, KDD 2024 (Princeton / Georgia Tech / IIT Delhi). Keyword stuffing shown as an illustrative negative — the paper found it reduced visibility.
The pattern is coherent: LLMs cite evidence, not keywords. Quotations from credible sources were the single most effective change (+41%); concrete statistics (+32%) and references to a page's own sources (+30%) followed; keyword stuffing — the classic SEO tactic — actively reduced visibility. Models are assembling a defensible answer, and they reach for pages that hand them quotable, attributable, numeric material.
Freshness is a ranking factor for citations
Content that sits untouched decays out of the citation set. AirOps' 2026 State of AI Search found:
- Pages not updated in 3+ months are 3× more likely to lose their citation.
- 70%+ of AI-cited pages had been updated within the last 12 months.
- Only 30% of brands hold visibility from one answer to the next.
- 48% of citations come from community platforms, and 85% of brand mentions originate from third parties.
That last pair matters: most of the conversation about a brand inside AI answers happens on pages it doesn't own. You can influence it, but you can't edit it directly — which makes knowing where you actually stand far more important than assuming your own pages are enough.
Authority and traffic still compound
None of this means domain strength stopped mattering. SE Ranking's analysis of 2.3 million pages (as reported by Superlines) found that high-traffic domains receive 3× more AI citations, and that domain traffic was the single strongest predictor of citation in their model (SHAP value 0.63).
So the picture isn't "authority is dead, just add quotes." It's both: authority gets a brand into the consideration set, and quotable, fresh, evidence-dense content gets it selected from it.
Where citations actually come from
OtterlyAI's AI Citations Report 2026 — built on over 1 million citations across ChatGPT, Perplexity and Google AI Overviews — adds two findings that reshape the picture:
- Community platforms (Reddit, Quora) take 52.5% of citations vs 47.5% for brand domains.
- 73% of websites have technical barriers that block AI crawlers — meaning a large share of brands are involuntarily invisible before content quality even enters the picture.
Put those together and the takeaway is sobering: a brand can have excellent content and still be invisible — blocked at the crawler, out-cited by a community thread, or simply not measured. The criteria are knowable; staying on the right side of all of them, across every engine, as they shift, is not a one-time job.
What this means for your business
Knowing what drives citations — quotes, statistics, named sources, freshness, authority, crawlability — is the easy part. The hard part is operational: which of your pages are cited today, which just dropped out of the answer, which third-party pages are speaking for your brand, and how all of that changes from one week to the next across ChatGPT, Perplexity and Google AI Overviews. That's a moving, multi-front target — not something a single content refresh solves.
That's the job CitationLab does. We monitor where your brand is cited across the AI engines, surface where you're losing ground and to whom, and translate the research above into the specific moves that actually move your visibility — so you get the outcome without having to build and run the machinery yourself.
The takeaway: AI citation isn't a black box. The research is consistent — quotes, statistics, named sources, freshness and crawlability win; keyword stuffing loses. The hard part isn't knowing what to do. It's knowing which of your pages are cited today and which just dropped out — and acting on it before your competitors do.
CitationLab shows you which pages and sources are carrying your citations right now, across every major AI engine — and we keep them there. See where you stand →
Sources
- Aggarwal et al. — GEO: Generative Engine Optimization, KDD 2024 (peer-reviewed; 10,000 queries, validated on Perplexity). Link · primary
- AirOps — 2026 State of AI Search. Link · primary
- OtterlyAI — AI Citations Report 2026 (1M+ citations). Link · primary
- SE Ranking (2.3M pages), as reported by Superlines. Link · aggregator
Frequently asked questions
What content factors are associated with more AI citations?
Does freshness affect whether AI cites a page?
Does domain authority still matter for AI citations?
Why might a brand be invisible to AI regardless of content?
Definitions used in this article
- GEO (Generative Engine Optimization)
- Generative Engine Optimization (GEO) is the practice of optimizing content so it gets surfaced, quoted and cited by AI answer engines. The foundational GEO study (Aggarwal et al., KDD 2024) measured which content changes raise a page's visibility in generated answers.
- Keyword stuffing
- Keyword stuffing is the practice of cramming a target search term into a page repeatedly to influence ranking. In the GEO study it reduced visibility in AI answers, because models reward evidence over keyword density.
- Freshness
- Freshness is how recently a page has been updated. It acts as a ranking factor for AI citations: pages not updated in 3+ months are markedly more likely to drop out of the citation set.
- AI crawlers
- AI crawlers are the automated bots — such as GPTBot — that AI companies use to fetch and index web pages for their models. If a site blocks them, its content can be invisible to AI answers regardless of quality.
- Domain authority
- Domain authority is the overall trust and strength of a website, often proxied by its organic traffic. High-authority domains receive disproportionately more AI citations, getting a brand into the consideration set before content quality decides selection.