The 40-second answer
The AEO best practices that move citation rates fall into four groups: fix crawler access so engines can reach you, structure content answer-first so passages are extractable, add schema so parsing is reliable, and build authority plus freshness so engines trust and prefer you. Access and structure deliver the fastest gains; authority compounds over time.
There's a lot of AEO advice online, most of it a flat list with no priority. Below are 15 practices grouped by leverage, so you know what to do first. Each maps to a mechanism engines actually use in 2026.
Group 1: Access — let engines reach you
Access gates everything. These deliver the fastest wins because the fix is binary.
1. Allow retrieval bots. Permit OAI-SearchBot, PerplexityBot, Claude-SearchBot, and ChatGPT-User. These power live AI search; blocking them removes you from answers entirely. Remember, 41% of B2B sites still block at least one major bot, at an estimated 18-34% citation cost each.
2. Separate training from retrieval decisions. You can block GPTBot, Google-Extended, and CCBot (training) while allowing retrieval bots. Don't block everything by reflex.
3. Render content server-side. Many crawlers read little or no JavaScript. If your content only appears after client-side rendering, it may be invisible.
4. Add an llms.txt index. A root-level Markdown map to your best content. Adoption is still ~10%, so it's a low-effort edge.
Group 2: Structure — make passages extractable
Engines cite passages, not pages. Structure so your best information is easy to lift.
5. Lead with a 40-60 word answer capsule. Put the direct answer first, before any setup. This is the highest-leverage structural move.
6. Use question-shaped H2s. Match headings to how people ask — questions, definitions, comparisons. They map directly to AI prompts.
7. Make sections self-contained. Each passage must stand alone; engines extract them individually. Restate the subject, avoid "as above."
8. Build a sequential heading hierarchy. H2 > H3 > H4 without skipped levels. Pages with this structure see a 2.8x citation lift over unstructured equivalents.
9. Format facts as tables and lists. Comparison tables, numbered steps, and stat lines extract far more cleanly than prose.
Group 3: Schema — make parsing reliable
Structured data lets engines parse with confidence. LLMs grounded in structured knowledge can be far more accurate than those working from raw text.
10. Add FAQPage schema to Q&A sections. FAQPage-marked pages show a 41% citation rate versus 15% unmarked, and keeping answers to 40-60 words optimizes extraction.
11. Use JSON-LD, validated. It's the format Google, Bing, Perplexity, and ChatGPT rely on for structured signals. Validate before shipping.
12. Match schema to content type. Article, HowTo, Product, and Organization schema where they fit — don't force irrelevant types.
Group 4: Authority and freshness — earn the citation
Engines cite sources they trust and recency they can verify. This group compounds.
13. Earn mentions on engine-favored domains. ChatGPT leans encyclopedic (Wikipedia is 47.9% of its top citations), Perplexity leans Reddit (46.7%), Google AI Overviews lean YouTube (23.3%). Seed authority where each engine looks.
14. Publish original, citable data. Give other pages a reason to cite you. Original stats become the fact engines lift.
15. Keep content fresh. 83% of AI citations for commercial queries come from pages updated within 12 months; Perplexity cited content from the last 30 days at an 82% rate. Set a refresh cadence.
Priority at a glance
| Group | Practices | Speed of impact | Effort |
|---|---|---|---|
| Access | 1-4 | Fast | Low |
| Structure | 5-9 | Fast | Medium |
| Schema | 10-12 | Medium | Low-medium |
| Authority + freshness | 13-15 | Slow, compounding | High |
Start at the top. Access and structure are fast and cheap; authority is where the long-term moat is built.
The mistakes that undo the work
Blocking every AI bot — a 2023-era reflex that removes you from AI search.
Schema on unreachable pages — parsing help for content engines can't fetch.
Answer buried below the fold — your citable passage never gets extracted.
Publish-and-forget — freshness decays; citations follow.
FAQ
What's the single most important AEO best practice?
Allowing retrieval bots. If engines can't reach your page, no other practice matters. Access is the gate.
Do these practices differ by AI engine?
The structure and schema basics are universal. Authority differs — ChatGPT favors encyclopedic sources, Perplexity favors Reddit, Google AI Overviews favor YouTube.
How much does FAQPage schema help?
Meaningfully. FAQPage-marked pages show around a 41% citation rate versus 15% for unmarked pages, especially with 40-60 word answers.
Is answer-first structure really necessary?
Yes. Engines extract passages, and a front-loaded 40-60 word answer is the passage most likely to be lifted and cited.
How do I prioritize these 15 practices?
Access and structure first (fast, cheap), schema next, then authority and freshness (slow but compounding). See the priority table above.
How do I know which practices I'm missing?
Run an AEO audit — it checks access, structure, and schema across your pages and returns a prioritized list.
Audit against these 15
SEO AEO Specialist checks your site against these access, structure, and schema practices and prioritizes the gaps. Start free at €0 (50 pages/month), get a full report for €9, or track a domain at €19/mo. Back to the /aeo hub.
AEO Audit: What It Covers and How to Run One (2026)
See where you stand
SEO AEO Specialist runs a free AI-visibility audit and hands you the exact fixes. One-off report €9, unlimited €19/mo.
Run your free audit →