How do AI platforms decide what to cite?

ChatGPT, Perplexity, Claude, Gemini, and Microsoft Copilot all cite content differently. Three core factors drive those decisions, and each platform weights them differently. Understanding all three is where any effective ASEO strategy starts.

The three citation factors

AI platforms cite content based on three intersecting factors.

1. Content structure

AI platforms retrieve content in chunks, not pages. A page may contain one highly citable block and five uncitable blocks. A citable block opens with a direct declarative answer. It's self-contained, requires no surrounding context, contains verifiable facts rather than marketing language, and sits within the optimal length range for AI extraction.

Content that starts with "Welcome to our website" or "We are proud to offer" scores near zero on citability. Content that starts with "ASEO (AI Search Engine Optimisation) is the practice of..." scores highly.

2. Entity authority

AI platforms build knowledge of entities (brands, people, places, organisations) from crawl data, structured data, and training data. Brands with consistent, well-structured entity signals get treated as more authoritative and more citable. The key signals are:

• Schema markup (LocalBusiness, Organization, Person, FAQPage, Article)
• Consistent NAP data (Name, Address, Phone) across the web
• Third-party mentions on high-authority platforms (Reddit, YouTube, LinkedIn, Wikipedia)
• E-E-A-T signals: expertise, experience, authoritativeness, trustworthiness

3. Technical access

AI crawlers must be able to reach your site. A robots.txt that blocks major crawlers, a missing llms.txt, or absent structured data all cut citation probability regardless of content quality. Cited By AI® checks access for 15 distinct AI bots as part of the technical readiness audit.

Platform differences

Each major AI platform has distinct citation behaviour. A per-platform content strategy is more effective than a single approach:

Perplexity AI

Perplexity runs a live web search for every query. It favours recently updated, authoritative sources with clear factual content. Freshness signals and structured data matter more here than on most platforms. It's also the most citation-transparent — it shows its sources.

ChatGPT (OpenAI)

ChatGPT draws on training data and, with browsing enabled, live web searches. It favours content from established, frequently cited domains. Entity consistency across the web matters more for ChatGPT than for search-first platforms.

Claude (Anthropic)

Claude favours well-structured, factually precise content with clear declarative answers. Direct-answer formatting and self-contained blocks work well here. It's particularly sensitive to content that reads as authoritative rather than promotional.

Gemini (Google)

Gemini has deep access to Google's search index and knowledge graph. Traditional authority signals (E-E-A-T, structured data, organic rankings) carry over directly. If you rank well on Google, you're already partway there with Gemini.

Microsoft Copilot

Copilot is built on Bing search infrastructure and is heavily influenced by Bing's crawl data. Bing-specific technical signals, including schema markup and structured data, have a stronger effect here than on other platforms.

The citation feedback loop

AI citation and traditional authority signals reinforce each other. Content structured for AI extraction tends to rank well organically too. Third-party mentions that build AI entity authority also build traditional backlink equity. The most effective ASEO strategies build both at the same time.