benchr
Ranking number one used to be the whole game. You hit the top of the results page, you got the click, you won. That deal is breaking. The prize now is being the sentence the AI quotes back to the user, because more and more the answer is the destination and the link is an afterthought.
This is a tactics guide, not a theory piece. Every number below traces to a named study. The throughline: the tricks that used to move rankings are mostly dead weight in generative engines, and a different set of moves actually earns you a spot inside the answer. Here they are, in order.
1. Answer the question first
Put the answer in the opening lines. Don't bury it under a 300-word windup. AI engines pull the cleanest, most extractable response they can find, and a page that states its answer up front is far easier to lift than one that meanders toward it.
The shape of your content matters as much as the timing. Ahrefs found AI Overviews trigger on about 57.9% of question-style queries versus only 15.5% of non-question results pages. So write in questions and answers. Use H2s that read like the things people actually type, then answer each one in the first sentence below it. That's the same pattern that drives zero-click search, where the answer lives on the results page and the user never leaves.
2. Cite, quote, and back it with stats
This is the single best-supported tactic, and it's the one most people skip. The Princeton-led GEO study tested a stack of optimization methods against a 10,000-query benchmark across generative engines. The winners weren't keyword plays. They were credibility plays.
Adding citations, direct quotations from relevant sources, and concrete statistics lifted a source's visibility by over 40% across queries. The three best-performing methods landed a roughly 30 to 40% relative improvement on the study's visibility metric. So when you make a claim, name the source and the number. Quote a primary document. Drop in a real statistic instead of a vague phrase. That's not padding; it's the lever.
3. Earn mentions across other sites
Your own pages aren't enough. AI engines weigh how often the wider web talks about you, and that signal beats almost everything you can do on your own domain. Across 75,000 brands, Ahrefs found branded web mentions had the strongest correlation with brand visibility in AI answers. The number of pages on your own site, the content-volume metric people obsess over, had almost no relationship at around 0.19.
So go get mentioned. Listicles, community threads, reference pages, video write-ups. The point is corroboration: when several independent sources say the same thing about you, an AI engine has more reason to repeat it. This is also why credibility cuts the other way against the made-up answer problem covered in our look at AI hallucinations in 2026; well-corroborated sources are the ones engines lean on when they're trying not to invent.
4. Build topical depth, not one long page
Length is a trap. Ahrefs measured a near-zero correlation, around 0.04, between word count and AI Overview citations. Padding a page to 4,000 words does nothing. What does move the needle is covering the cluster of related questions around your topic.
Ahrefs found that pages ranking across Google's related "fan-out" queries are 161% more likely to be cited in the final AI Overview than pages ranking only for the primary term, a correlation of 0.77. So instead of one bloated article, build several focused pages that each answer a distinct question in the cluster. Depth across the topic beats length on a single URL every time. If you've been told word count is a ranking signal for AI, that's part of why benchmarks stopped telling you what you think they measure.
5. Keep ranking, but don't bank on it
Classic rankings still help. They're just not the guarantee they once were. Ahrefs measured a moderate Spearman correlation of 0.347 between ranking in Google's top 10 and being cited among the top AI Overview sources. Useful, not deterministic.
The trend is loosening further. In July 2025, 76% of AI Overview citations came from top-10 pages with a median rank of position 2. An updated Ahrefs analysis across 863,000 results pages and roughly 4 million AI Overview URLs put that share closer to 38%. So Google is increasingly pulling AI sources from beyond page one. Strong rankings raise your odds; they don't lock in the citation.
The tactics, side by side
Here's the menu, with why each one earns you a place inside an AI answer rather than just a rank on a page.
| Tactic | Why it works for AI citation |
|---|---|
| Answer the question up front | Extractable answers get lifted; AI Overviews fire on 57.9% of question queries vs 15.5% otherwise |
| Add citations, quotes, and stats | The top GEO methods; lifted source visibility by over 40% across queries |
| Earn third-party mentions | Branded web mentions were the strongest AI-visibility correlate across 75,000 brands |
| Cover fan-out related queries | Pages ranking across related queries are 161% more likely to be cited (0.77 correlation) |
| Keep ranking on Google | Moderate help (0.347 correlation), but the top-10 share of citations fell from 76% to about 38% |
| Stuff keywords | Ineffective in the GEO study, sometimes worse than the baseline. Skip it. |
| Pad word count | Near-zero correlation with citations (~0.04). Length is not the lever. |
What about llms.txt?
Short version: don't count on it. The llms.txt file gets pitched as a way to hand AI crawlers a tidy map of your site, but there's no evidence it earns citations.
At Search Central Live in July 2025, Google's Gary Illyes said Google doesn't support llms.txt and has no plans to. John Mueller compared it to the long-ignored meta keywords tag, which tells you exactly how seriously the search side takes it. Server-log analyses show the major AI crawlers don't routinely request the file, and no major LLM provider has confirmed consuming it. It's cheap to add and won't hurt, so add it if you want. Just don't treat it as a lever. It's unproven at best.
The prize moved. It's no longer the top rank, it's the quoted sentence. Optimize for that.
Why this is worth the effort
Because the clicks are draining out of the old model. A Pew Research Center study in July 2025 found users clicked a traditional search result in just 8% of searches that showed an AI summary, versus 15% without one, and clicked a link inside the summary only 1% of the time. Ahrefs separately measured a 34.5% drop in clickthrough rate for the number-one result when an AI Overview was present.
So the link is no longer where the visibility lives. Being inside the answer is. If your audience leans social, the same logic applies to how assistants surface tools and brands, which is the backdrop to our pick of the best AI for social media. Optimize to get quoted, not just to rank, and you're playing the game that's actually being scored.
Frequently asked
What are GEO and AEO?
GEO is generative engine optimization and AEO is answer engine optimization. Both cover the same goal: getting your content surfaced and quoted inside AI answers instead of just ranking on a results page. The term GEO comes from the Princeton-led study by Aggarwal and colleagues (KDD 2024), which tested what actually lifts a source's visibility in generative engines. The headline finding is that credibility signals work and old keyword tricks don't.
How do I get cited by ChatGPT and Perplexity?
Lead with credibility and corroboration, not keywords. The GEO study found that adding citations, quotations from relevant sources, and statistics lifted a source's visibility by over 40%, while keyword stuffing did nothing and sometimes scored worse than the baseline. Separately, Ahrefs' study of 75,000 brands found branded web mentions had the strongest correlation with AI visibility, while the sheer number of pages on your own site barely mattered, around 0.19.
Does schema markup help me get cited?
Treat schema as a clean-structure best practice, not a proven lever. Google says structured data helps it understand a page, but there's no verified named study showing schema directly raises AI-answer citation rates. Add it because it keeps your content machine-readable and costs little. Don't expect it to earn citations on its own.
Is an llms.txt file worth adding?
There's no evidence it works. At Search Central Live in July 2025, Google's Gary Illyes said Google doesn't support llms.txt and has no plans to, and John Mueller compared it to the long-ignored meta keywords tag. Server-log analyses show major AI crawlers don't routinely request the file, and no major LLM provider has confirmed using it. Treat it as unproven at best, not a reliable way to earn AI citations.
Does ranking on Google still matter for AI answers?
It helps but doesn't guarantee anything. Ahrefs measured only a moderate Spearman correlation of 0.347 between ranking in Google's top 10 and being cited in AI Overviews. In July 2025, 76% of AI Overview citations came from top-10 pages with a median rank of 2, but an updated Ahrefs analysis put that closer to 38%, so Google increasingly pulls AI sources from beyond page one. Strong rankings raise your odds; they don't lock in a citation.
Changelog
- May 30, 2026 — Originally published. GEO study figures verified against Aggarwal et al. (KDD 2024); citation, click, and correlation numbers verified against Ahrefs and Pew Research Center reports.
References
- Aggarwal et al., "GEO: Generative Engine Optimization," arxiv.org, accessed May 2026.
- Aggarwal et al., "GEO: Generative Engine Optimization (HTML v3)," arxiv.org, accessed May 2026.
- Ahrefs, "Does Ranking Higher on Google Mean You'll Get Cited in AI Overviews?," ahrefs.com, accessed May 2026.
- Ahrefs, "76% of AI Overview Citations Pull From the Top 10," ahrefs.com, accessed May 2026.
- Ahrefs, "Update: 38% of AI Overview Citations Pull From The Top 10," ahrefs.com, accessed May 2026.
- Ahrefs, "Top Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)," ahrefs.com, accessed May 2026.
- Ahrefs, "How to Rank in AI Overviews," ahrefs.com, accessed May 2026.
- Ahrefs, "AI Overviews Reduce Clicks by 34.5%," ahrefs.com, accessed May 2026.
- Pew Research Center, "Google users are less likely to click on links when an AI summary appears," pewresearch.org, accessed May 2026.
- Search Engine Journal, "LLM Guidance Doesn't Transfer The Way SEO Guidance Did," searchenginejournal.com, accessed May 2026.