GEO / AI Search

How to Get Your Brand Cited by ChatGPT, Claude and Perplexity

12 min read By Sebastian, Neurobird
A single lime answer node lit up among dimmed search result cards, representing a brand being cited by an AI engine
Short answer

To get cited by AI engines, do five things: let their crawlers in, mark your content up with schema, write answer-shaped passages with named statistics, make your brand a verifiable entity, and keep your content in raw HTML. Generative Engine Optimization (GEO) is the discipline of doing all five so your brand becomes the answer ChatGPT, Claude, Gemini and Perplexity quote.

Your buyers no longer scroll ten blue links. They ask ChatGPT, Claude or Perplexity a question and read the answer. If your brand is not inside that answer, you do not exist for that buyer. This guide is the practical playbook we use at Neurobird to make brands citable, step by step, with the evidence behind each move.

What "getting cited" actually means

An AI answer engine works in two stages. First it retrieves a set of passages from pages it has already crawled. Then it writes an answer and attaches citations to the passages it leaned on. Getting cited means your page is both retrieved (the engine can reach and understand it) and selected (your passage is the clearest, most credible answer to the question). Generative Engine Optimization, or GEO, is the practice of winning both stages.

This is a different goal from classic SEO. SEO chases a ranking position. GEO chases citation share: how often your brand appears as a quoted source across the answers your buyers see. A page can rank poorly on Google and still be cited heavily by Perplexity, and the reverse is true too.

Step 1: Let the AI crawlers in

You cannot be cited by an engine that was never allowed to read you. The single most common GEO mistake is a robots.txt that blocks the very bots you want citing you, often left over from a 2023-era "block the scrapers" reflex.

AI bots split into three jobs: training crawl, search and answering index, and live fetch when a user asks the assistant to open a page. For visibility you want to allow the search and answering bots even if you have feelings about training. The ones that must never be blocked if you want citations:

EngineAllow these user-agents
ChatGPTOAI-SearchBot, ChatGPT-User
ClaudeClaude-SearchBot, Claude-User
PerplexityPerplexityBot, Perplexity-User
Google AI Overviews and AI ModeGooglebot
CopilotBingbot
Apple IntelligenceApplebot

One subtlety worth knowing: Google-Extended and Applebot-Extended are control tokens, not crawlers. Blocking Google-Extended does not remove you from AI Overviews, because those are powered by Googlebot. Add a Sitemap line to robots.txt and publish a real sitemap.xml so engines can discover every page.

Step 2: Mark your content up with schema

Structured data is how you hand an engine a clean, machine-readable summary of who you are and what each page says. Pages that carry complete schema show up materially more often in AI answers across 2025 and 2026 studies. The high-impact types for most businesses:

One hard rule: only mark up content that is actually visible on the page. Schema that describes things a reader cannot see violates Google policy and erodes trust.

Step 3: Write in the shape AI engines quote

This is where most of the citation lift lives, and it is the most researched. A peer-reviewed study from Princeton presented at KDD 2024 tested nine GEO methods across ten thousand queries. The techniques that moved visibility the most were concrete and writeable:

Adding cited statistics lifted visibility by about 32 percent. Adding expert quotations lifted it by about 41 percent. Citing outside sources added about 27 percent. Keyword stuffing actively hurt, at about minus 8 percent. (Princeton, KDD 2024)

Translated into how you write:

Step 4: Become a verifiable entity

Before an engine stakes a citation on your brand, it tries to confirm you are a real, consistent entity. This is the signal that has overtaken raw backlinks as the headline factor in 2026. You reinforce it on two fronts.

On your own site: an Organization schema with a rich sameAs array and a consistent name, address and contact across every page. Off your site, in rough order of payoff: a Wikidata entry (you can create one directly, and it is the knowledge-graph node your sameAs should point to), a Crunchbase profile, a company LinkedIn page, and listings on independent directories like G2 or Clutch. Third-party listings carry more weight than your own blog because they are independent verification.

Step 5: Keep your content in raw HTML

This is the technical trap that silently kills GEO for modern sites. Most AI crawlers do not run JavaScript. Analyses of hundreds of millions of GPTBot fetches in 2026 found no JavaScript execution: they read raw HTML only. Only Googlebot renders JavaScript.

If a sentence on your page is inserted by JavaScript after load, it is invisible to ChatGPT, Claude and Perplexity. They never see it.

So any copy that matters for citation, your definitions, your statistics, your answers, must be present in the HTML the server sends, not built client-side. Static sites and server-rendered pages have a real advantage here.

How to measure whether it is working

The KPI has shifted from "rank one" to "citation share." Track how often your brand appears in answers to your target questions across ChatGPT, Perplexity, Claude and Google AI Overviews, and watch your server logs for visits from OAI-SearchBot, PerplexityBot and Claude-SearchBot, which confirm the engines are actually reading you. Our own platform at geo.neurobird.com audits any site for exactly these signals and generates the fixes.

The honest part

Some numbers circulating in GEO marketing are unverifiable, including a widely repeated "FAQ schema gets cited 4.2 times more" claim with no traceable source. Treat those with caution. The evidence that holds up is the Princeton method study, the consistent finding that schema and entity signals help, and the well-documented fact that AI crawlers do not run JavaScript. Build on the proven levers and ignore the folklore.

How do AI engines decide which sources to cite?

They retrieve passages from pages they have crawled, then cite the ones that most clearly and credibly answer the query. Clarity, structured data, named statistics, and a verifiable brand entity are the main drivers.

Is GEO different from SEO?

It overlaps, but GEO optimizes for being cited inside an AI answer rather than ranking as a blue link. The big differences are crawler access for AI bots, structured data, answer-shaped passages, and entity authority.

Do AI crawlers read JavaScript-rendered content?

Mostly no. GPTBot, ClaudeBot, PerplexityBot and similar fetch raw HTML and do not execute JavaScript. Only Googlebot renders JavaScript, so anything injected by JavaScript is invisible to most AI engines.

How long does GEO take to work?

Crawler access and schema take effect within days to weeks of the next crawl. Entity authority and content depth compound over weeks to months as engines recrawl and third-party signals accumulate.

Want to know how AI sees your brand today?

Drop your domain into our live studio and get an AI-visibility audit in seconds, or book a call and we will map the GEO work for you.

Run the free audit