It's not keyword density. Schema, FAQ depth, Exa-verified sources, and information architecture that pre-answers follow-up questions — that's what AI platforms cite. This is the exact brief Hema uses to write every generated article, and every article you should be writing.
Why AI cites some articles and ignores others
Google ranks content based on hundreds of signals developed over 25 years. AI citation is newer and simpler — but the signals are very different. AI platforms cite articles that are structurally transparent: articles where the answer to a specific question is immediately findable, where the information hierarchy is clear, and where the content is supported by structured data that the model can extract as facts.
An article that ranks #1 in Google can be completely invisible in AI search if it buries its key claims in long paragraphs without structured FAQ markup, or if it uses schema incorrectly, or if it answers the main question but ignores the five follow-up questions the AI is likely to generate next.
The good news: the changes required to make content AI-citable are specific, mechanical, and implementable. Here are the 7 mandatory elements.
The 7 mandatory elements of an AEO-ready article
1. Direct answer in the first 50 words
AI models extract the most likely citation text from the opening of an article. The first 50 words should directly answer the title's implied question. Not "In this article, we will explore..." — the actual answer. If your article is titled "What is Query Fanout?" the first sentence should define it.
2. Article + FAQPage JSON-LD schema
Both schema types are required. Article schema tells AI models what the page is, who wrote it, and when it was published. FAQPage schema marks up your FAQ section so every question and answer pair is individually parseable as structured data — allowing AI to extract specific Q&A pairs rather than guessing from unstructured prose.
3. Author entity with verifiable credentials
AI platforms weight content from named, credentialed authors more heavily than anonymous content. Include an Author schema entity with name, job title, and a sameAs URL (your LinkedIn or personal site). For regulated industries, add the relevant credential type.
4. Exa-verified or authoritative source citations
AI models are trained to prioritise content that cites authoritative sources. Every factual claim should link to an official document, peer-reviewed paper, or high-authority publication. Hema's Blog Writer uses Exa AI to find current, authoritative sources during the research phase — this is not decoration, it directly affects whether AI platforms choose to cite the article.
5. FAQ section that pre-answers AI-generated follow-ups
The FAQ section is not a vanity section. It's where you address the questions AI is likely to generate as fanout from your main article topic. Before writing the FAQ, run the article's core topic as a prompt in Hema and check the fanout questions. Write FAQ answers for those exact questions.
6. H2/H3 heading hierarchy that mirrors likely queries
Use headings that are phrased as questions or direct answer statements — not marketing headlines. "How Does Query Fanout Work?" is better than "Understanding the Power of AI Signals." AI models use heading text as an index of what the article covers.
7. Minimum 1,200 words of substantive content
Content under 800 words rarely has sufficient depth for AI to confidently cite. The minimum for a credibly cited article is 1,200 words. Hema's Blog Writer defaults to 1,600–2,200 words for most article types.
FAQ writing guide
The most common mistake in AEO FAQ writing is answering hypothetical questions. Don't write FAQs based on what you think users ask. Write them based on what AI actually generates as follow-ups to your core topic.
Process:
- •Run your article's main topic as a prompt in Hema
- •Check the Query Fanout section for the 5 follow-up questions AI generates
- •Write FAQ answers for those exact questions — they are what the AI predicts your readers will ask next
- •Each answer should be 40–100 words: direct, specific, and self-contained
- •Add FAQPage JSON-LD schema wrapping each Q&A pair
How to find the domains AI already trusts in your category
Source selection is one of the most important decisions in AEO content creation — and the most under-discussed. AI platforms don't cite all sources equally. They weight sources based on their own training data, which means some domains are consistently cited more than others in specific topic areas.
Hema's Sources tab shows you exactly which domains AI is citing in your category, with citation counts and per-platform breakdown. Before writing any article, check the Sources tab for your relevant prompts and identify the top 5 cited domains. Cite those domains in your article. You're creating a citation chain that AI already trusts.
Go to Analytics → Sources. Filter to your most important prompt group. The domains in the top 10 are the publications AI considers authoritative for your category. Cite them directly in your articles — and pursue coverage in the ones you haven't been mentioned in yet.