NewFastlook now supports Google AI Overviews & Perplexity citations.Explore resources
FAQs

How To Get Cited By Llms

Written by: Content & GEO Research

Fastlook Team

Posted: 5 min readUpdated:

Most teams assume that a page which ranks well will be quoted by ChatGPT, Perplexity or Google's AI Overviews. The research says otherwise, and the gap is where the work sits. This guide explains how to get cited by LLMs, separating the retrieval step from the citation step, and shows which page shapes and sourcing habits the verified data supports.

Quick answer

An LLM citation is a link that an answer engine attaches to a specific claim in a generated answer, so the reader can open the source. A mention is a brand or domain named in the text without a link. A backlink is a hyperlink from another website that passes search signals, and a backlink has no bearing on whether a language model chooses to quote a page.
Topic
how to get cited by llms
Last updated
Nov 16, 2026
Read time
5 min
How To Get Cited By Llms — illustrated banner

How to get cited by LLMs: a page-level overview

Citation is the outcome of a short chain of steps. An answer engine first retrieves candidate pages, then selects passages from those pages, then attaches a link to the claims it actually uses. Ahrefs describes this as a core RAG pipeline. The pipeline breaks a prompt into sub-queries, retrieves pages for each sub-query, splits those pages into passages, and selects the most relevant passages. Citations are attached to claims traceable to a source. Each stage filters out most candidates. A page can be retrieved and still never be cited. Citation therefore depends on three questions:

  • Can the answer engine reach and read the page?
  • Does the page's structure match the question?
  • Are the page's claims specific enough to attribute?

This guide takes those questions in order. It treats the research figures as the limits of what is known, not as a playbook.

Sources & further reading

Frequently asked questions

What is an LLM citation, and how does it differ from a mention or a backlink?

An LLM citation is a link that an answer engine attaches to a specific claim in a generated answer, so the reader can open the source. A mention is a brand or domain named in the text without a link. A backlink is a hyperlink from another website that passes search signals, and a backlink has no bearing on whether a language model chooses to quote a page. According to OpenAI's help documentation, ChatGPT search answers can include citations that readers can open. However, source selection is not fully predictable.

How do ChatGPT, Perplexity, Gemini and Google AI Overviews choose which sources to cite?

Each engine retrieves from a different index before selecting passages. According to Surfer, ChatGPT uses Bing's index, Google uses its own index, Perplexity crawls its own index, and Claude uses Brave's index. Selection then favours passages that answer the sub-query directly. According to Ahrefs' analysis of 1.4 million ChatGPT prompts, only about half of retrieved URLs end up cited. Retrieval alone therefore does not decide the outcome.

Why doesn't ranking well in Google guarantee an AI citation?

Ranking and citation select on different signals. Surfer's study of 10,000 keywords found that 67.82% of pages cited in Google's AI Overviews did not rank in the top 10 for the query or any of its fan-out queries. Surfer's analysis of about 5 million cited URLs also found domain authority had almost zero correlation with AI citations. A lower-ranked page with a direct, specific answer can therefore be cited where a top result is not.

Does the opening of a page matter for AI citation?

The opening carries disproportionate weight. Surfer reports that 38% of AI Overview citations come from the first 100 words of a page. Answer-first structure, where the direct answer appears in the opening sentences of the page and each section, gives retrieval systems a clean passage to lift. The first paragraph should state the claim a reader would quote, not introduce the topic.

Which content features make a page more likely to be quoted?

Quoted and sourced statistics make a page more likely to be cited. According to Surfer's research, adding quotations improved visibility in generative engines by about 41%. Adding statistics improved visibility by about 33%. However, both figures come from Surfer's own study and reflect its scope. They indicate direction rather than a guaranteed result. Attributable statistics with a named source give an engine a verifiable claim to cite.

Does schema markup or structured data determine whether an AI engine cites a page?

No published source establishes that schema markup causes a citation. According to Google's AI features documentation, no special AI-only markup or machine-readable file is required for AI Overviews or AI Mode. Pages must be indexed and eligible to show a snippet in Google Search. Structured data helps other systems parse a page. However, its effect on citation is unproven and should not be treated as the lever.

What technical conditions decide whether an AI system can retrieve a page?

Retrieval requires that a page is crawlable, indexed, and carries its answer as visible text. According to Google's documentation, AI Overview eligibility is tied to normal indexing and snippet eligibility in Google Search. Content hidden behind scripts, gated behind a login, or present only in images cannot be passed to a retriever. Engines with separate indexes, such as Perplexity's own crawler and the Bing index used by ChatGPT, also require the page to be accessible to those specific crawlers.

Do citation patterns differ between platforms?

Yes, citation patterns differ between platforms. Onely's research across 31 rounds of best-agencies questions found that 67.5% of 963 cited pages were lists of providers. According to Onely, Perplexity cited lists 82.2% of the time, Gemini 77.4%, Google AI Mode 67.9%, and ChatGPT 23.8%. ChatGPT leaned more on the provider's own site, at 60.5%. A page shaped as a comparison list may therefore suit one engine more than another.

How can a brand measure its AI visibility over time?

Measure AI visibility against a fixed set of prompts that matter to the brand, run on a schedule. Record which domains each engine cites. Track two things separately: whether the brand's domain is cited, and whether competitors are. According to the attribution-gap study of about 14,000 real conversations with search-enabled AI models (arXiv 2508.00838), 61% of AI answers drew on more websites than they credited. Citation counts therefore understate actual use. Pair citation tracking with referral data. According to SE Ranking, **AI referral traffic was about 0.32% of all website traffic in early 2026**.

Is there a single fix that guarantees citations?

No single fix guarantees citations. A page that is retrievable, answer-first, specific, and sourced improves its chances, but no combination of these guarantees selection. According to OpenAI, source selection is not fully predictable. According to Ahrefs, only about half of the URLs ChatGPT retrieves end up cited in the final answer. That gap shows how much of the outcome sits outside the page. The realistic goal is to be a strong candidate across many prompts, and to measure the result rather than assume it.

Related in this topic

Is your brand in the answer?

See how AI engines read your site.

Run the free Agent-Ready Check, or start Fastlook with 3 free pages. No card, no sales call.

  • ChatGPT
  • Perplexity
  • Gemini
  • Claude
  • Copilot
  • AI Overviews

Be the answer AI finds. Your AEO growth partner.

A founder in a black turtleneck, editorial black and white portrait with a warm orange glow