Field Notes · GEO
How to Get Your Brand Cited by AI
AI answer engines like ChatGPT, Perplexity, and Google AI Overviews now sit between your buyers and your website. Here is how to become the source they quote, not the link they skip.
A growing share of buyers no longer start with ten blue links. They ask ChatGPT, Perplexity, Claude, or Google's AI Overview, read the synthesized answer, and act on it. When that answer names a brand, that brand wins the consideration before a single click happens. Getting named is the entire game, and it is a discipline, not an accident. The work of earning those mentions is called Generative Engine Optimization, or GEO.
GEO is not a rebrand of SEO. SEO earns a ranked position on a results page. GEO earns inclusion inside an answer that cites a handful of sources. The two overlap by roughly 40 percent. The other 60 percent, passage structure, entity consistency, and third-party corroboration, is new work. What follows is the version we run for clients, condensed into five steps.
What "getting cited" actually means
An AI answer engine does not rank pages. It retrieves passages, synthesizes them into an answer, and attributes some of them to sources. To be cited, your content has to be three things at once: reachable by the engine, structured so a single passage answers a single question, and corroborated elsewhere so the model trusts it.
That means the unit of optimization is no longer the page. It is the passage. A page can rank without any one sentence being quotable. An answer engine needs the quotable sentence.
Step 1. Make sure the engine can reach your site
This is the step most brands skip, and it is the one that quietly disqualifies them. If an AI crawler cannot fetch your content, nothing else you do matters. You do not exist to that engine.
- Check your robots.txt and firewall for blocks on GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, and Google-Extended.
- Confirm your important content is in the HTML response, not painted in afterwards by JavaScript the crawler never runs.
- Separate training crawlers from live-answer crawlers. Blocking training is a valid choice. Blocking the live-answer fetchers means you are invisible inside the answers themselves.
Run one test: fetch your key pages with each AI user-agent and read what comes back. Most sites we audit are accidentally blocking at least one major engine.
Step 2. Write content as citable passages
Answer engines prefer content that reads like a reference entry: one clear claim, one credible basis, one sentence worth quoting. Lead with the answer, then support it. Bury the answer three paragraphs down and the model will quote a competitor who put it first.
- Open each section by answering the question in its heading directly, in one or two sentences.
- Use question-shaped headings that match how buyers actually ask.
- Keep claims specific and verifiable. "Faster" is ignored. "Mobile LCP under one second" gets quoted.
Step 3. Make your brand one unambiguous entity
Models cite brands they can identify with confidence. If the engine cannot tell that your website, your Google Business Profile, and your LinkedIn page describe the same organization, it hedges, and a hedged model names someone clearer instead.
- Define an Organization in your structured data with a stable identifier, then reference it from every page.
- Add a sameAs graph linking your site to your authoritative profiles: your Google entity, your LinkedIn page, any industry directory you appear in.
- Keep your name, focus, and description identical across all of them. Consistency is the signal.
This is the least glamorous step and the highest leverage. Entity consistency is what lets a model say your name without qualifying it.
Step 4. Earn corroboration the model already trusts
An engine is more confident citing a claim that appears in more than one place. Your own site asserting something is a start. The same fact appearing in a publication, a directory, or a credible third-party source is what turns an assertion into a citable fact.
You do not need a press release. You need your core claims to exist, accurately, on the sources these models already crawl. Industry directories, quotable media, and authoritative reference pages move citation rate more than another blog post on your own domain.
Step 5. Measure citation rate, not rankings
You cannot improve what you do not measure, and rankings do not measure this. Build a fixed set of prompts your buyers would actually ask, run them across ChatGPT, Perplexity, Gemini, and Claude on a schedule, and record three things: whether you were named, how you compared to competitors in the same answer, and how the model framed you.
That citation rate, tracked monthly, is the real scoreboard. A keyword position tells you nothing about whether the AI recommended you.
None of this is magic, and none of it is fast. Citations on Perplexity and Claude can move within weeks because they re-crawl aggressively. Google AI Overviews move slower. What compounds is the entity work and the corroboration: once a model is confident about who you are, every new piece of content inherits that trust.
Getting cited is the new getting ranked, and the brands that treat it as a discipline are pulling ahead of the ones still optimizing only for a results page that fewer people read each quarter. If you want to know whether the engines name you today, that is exactly the audit we start every GEO engagement with.
Questions people ask
Is GEO just SEO with a new name?
No. SEO optimizes for a ranked list of links. GEO optimizes for inclusion inside a synthesized answer that cites a few sources. The tactics overlap by around 40 percent. Passage structure, entity consistency, and third-party corroboration are the new 60 percent.
How long until my brand gets cited?
It depends on the engine. Perplexity and Claude re-crawl aggressively and can reflect changes within weeks. Google AI Overviews move more slowly. Entity and corroboration work compounds, so the first citation is usually the hardest and the rest follow faster.
Can I just add schema and be done?
Schema helps a model understand and trust your content, but it does not create citable substance on its own. Schema, plus reachable and passage-structured content, plus entity consistency, is the combination that earns citations. Any one of them alone underperforms.
How do I know if AI already cites my brand?
Build a set of prompts a buyer would ask, run them across the major engines, and record whether your brand appears, how it compares to competitors, and how it is described. That is your baseline citation rate. Re-run it monthly to see whether your work is moving it.