My Opinion: How ChatGPT and Perplexity Decide Which Sites to Cite

Every day, millions of people ask AI engines questions that used to go to Google. And every day, those AI engines decide which websites to pull from — and which to ignore.

If your site is not being cited, you are invisible to a growing share of your potential audience. Understanding how these systems work is the first step to changing that.

Google ranks pages. AI engines synthesize answers.

When you search Google for "best project management software," you get a list of pages ranked by relevance and authority. You choose which one to click.

When you ask Perplexity the same question, it reads dozens of sources, synthesizes the information, and gives you a direct answer — with citations. You may never visit most of those pages. You just read the summary.

This changes everything about how visibility works.

Before any AI engine can cite your site, it needs to be able to read it.

Most major AI systems use dedicated crawlers:

• GPTBot — OpenAI (ChatGPT). OpenAI publishes full GPTBot documentation, including how to allow or block it.
• PerplexityBot — Perplexity. Perplexity's crawler documentation is available here.
• ClaudeBot — Anthropic (Claude). Anthropic's usage policy covers ClaudeBot.
• Google-Extended — Google AI Overviews. Google documents Google-Extended in its Search Central help.

If your robots.txt file blocks these bots — even accidentally — you are invisible to those systems. This is one of the most common issues we find in audits.

AI language models work with entities — named things with defined attributes. Your business is an entity. Your products are entities. Your team members are entities.

The clearer your site makes these entities, the easier it is for an AI to understand, categorize, and cite you.

Ask yourself: if an AI read only your homepage, would it know:

• Exactly what your business does?
• Who your customers are?
• Where you operate?
• What makes you different?

If the answer is "probably not," your entity clarity needs work.

Schema markup is code you add to your pages that explicitly tells machines what your content means — not just what it says. Schema.org is the open vocabulary maintained by Google, Microsoft, Yahoo, and Yandex that defines every available schema type.

A page about a local restaurant might include schema that says: "This is a Restaurant, located at [address], open [hours], serving [cuisine type], with an average rating of [X]."

AI engines love this. It removes ambiguity. It makes your content machine-readable in a way that plain text simply is not.

Common schema types that help with AI citations:

• Organization or LocalBusiness
• Article or BlogPosting
• FAQPage
• Product
• Person
• HowTo

See our schema markup beginner guide for step-by-step implementation instructions.

AI engines are, at their core, answer machines. They are looking for content that clearly and directly answers the questions people ask.

This means:

• FAQ sections that mirror real user questions
• Headings structured as questions ("How does X work?", "What is the cost of Y?")
• Direct, concise answers in the first paragraph of each section
• Comprehensive coverage of a topic, not just surface-level summaries

The sites that get cited most often are the ones that answer questions thoroughly, accurately, and without burying the answer in fluff.

AI models are trained on the web. Sites that are frequently cited, linked to, and referenced by other authoritative sources carry more weight in the model's understanding of who the experts are.

This overlaps with traditional SEO link-building — but the mechanism is different. It is not just about PageRank. It is about whether the broader web treats your site as a credible source on a given topic. Moz's guide to domain authority explains how external citation signals are measured.

AI engines — especially those with real-time web access like Perplexity — favor recent, up-to-date content. A blog post from 2019 is less likely to be cited than one from last month, all else being equal.

Regular publishing and updating existing content signals that your site is actively maintained and current.

Some traditional SEO tactics have little or no impact on AI visibility:

• Keyword stuffing — AI models understand context, not just keyword frequency
• Thin content optimized for a single keyword — AI engines want depth and completeness
• Backlink schemes — manipulative link profiles do not translate to AI authority
• Meta keywords — completely irrelevant to AI systems

Here is what makes this moment so important: most websites have not been optimized for AI search at all.

The sites getting cited right now are largely getting cited by accident — because they happen to have good structured data, clear entity signals, and comprehensive content. They did not set out to optimize for AI. They just built good websites.

That gap is closing fast. As more site owners understand AEO, the competition for AI citations will intensify. The sites that act now have a significant first-mover advantage.

The fastest way to understand your AI search visibility is to get an audit that checks the signals that actually matter to AI engines — not just the traditional SEO metrics.

Our free audit checks 40+ SEO and AEO signals and gives you:

• Your AIReady™ Score (0–100) — learn exactly how it is calculated
• A breakdown of every signal checked
• A prioritized fix list ordered by impact
• Visibility analysis for 3 keywords in Claude

Get your free audit → See exactly where you stand and what to fix first.

Further reading:

• SEO vs AEO: what is the difference and why it matters
• Is your website visible to ChatGPT? How to check and fix it
• Perplexity SEO: how to get your site cited in Perplexity answers
• Schema markup: the beginner guide for site owners
• What is an AIReady™ Score and how is it calculated?

External references:

• OpenAI — GPTBot documentation
• Perplexity — PerplexityBot documentation
• Anthropic — ClaudeBot crawler policy
• Google Search Central — Google crawlers overview
• Schema.org — full schema type reference
• Moz — Domain authority explained