The AEO Checklist: A 40-Point Audit for Answer Engine Optimization in 2026

What Is an AEO Checklist?

An AEO checklist is a scored audit of your content's readiness to be cited by AI answer engines — ChatGPT, Perplexity, Claude, Google AI Overviews. You work through a defined list of factors known to influence citation rates, score your pages, fix the misses, and re-check.

This one has 40 points across five dimensions, in priority order: technical access → content structure → schema → authority → measurement. The order matters. Polishing schema on a page GPTBot can't crawl is a spa day for a page nobody visits.

To show what a real audit looks like, we ran this checklist on altyzo.com — the site you're reading. Our score is at the bottom, and it's lower than we'd like. That's the point.

TL;DR: A 40-point AEO audit across five dimensions — technical access (8 pts), content structure (10), schema (7), authority (8), measurement (7) — in priority order. altyzo.com scores 34/40 with authority as the open gap. Score your top 10 pages; below 25 means fix access first.

Dimension 1: Technical Access (8 Points)

These are single-point failures — the "car won't start" tier of problems. A page blocked to GPTBot or PerplexityBot makes everything below irrelevant, so fix these first, always.

1. robots.txt allows AI crawlers. GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, and Google-Extended all permitted. Blocking any of them makes your content invisible to that engine's retrieval layer. (Our own robots.txt allows all five.)

2. Indexed in Bing. ChatGPT Search retrieves from Bing's index, not Google's — and Perplexity leans on Bing heavily. Submit your sitemap to Bing Webmaster Tools and verify your priority pages appear.

3. No paywall or login gate on key content. AI crawlers can't authenticate. Anything behind a login is invisible.

4. Core Web Vitals pass. LCP under 2.5s, INP under 200ms, CLS under 0.1 (75th percentile, mobile). Crawlers don't wait around.

5. HTML is the primary delivery method. Server-rendered or statically generated content — not injected by client-side JavaScript. Test it: view source, search for your paragraph.

6. Canonical URLs set. One version of each page. Duplicate versions (www/non-www, http/https, trailing slashes) split the citation.

7. Sitemap submitted to both consoles. Google for AI Overviews, Bing for ChatGPT and Perplexity. (Winning the Overview itself: how to get cited in Google AI Overviews.)

8. No aggressive bot protection on content pages. Cloudflare challenges and CAPTCHAs block AI crawlers too. Whitelist the search bots.

Dimension 2: Content Structure (10 Points)

Access gets you retrieved. Structure gets you extracted. This is where most pages quietly fail — they read beautifully to a human and look like a wall of mush to a machine looking for an answer to lift.

9. Direct answer in the first 100 words. Lead with a 40–60 word answer capsule, then expand. Answer-first pages show up disproportionately in citation analyses — see RankSenseAI's synthesis of citation-factor research.

10. Question-based H2/H3 headings. Match real buyer prompts, not keyword fragments. The question is the heading; the answer is the paragraph directly under it.

11. Self-contained subsections. Each should make sense lifted out of context — because that's exactly what an engine does with it. (Not arbitrary "chunks" — Google's own guidance says there's no ideal piece size; the point is each section survives extraction, not that you slice for it.)

12. TL;DR or key-takeaways block near the top. Under 100 words. It gives the engine a ready-made answer and the skimmer a reason to stay.

13. Lists and tables for structured information. Engines extract structure more reliably than narrative. Lists for steps and enumerations; tables for comparisons.

14. The page covers follow-up questions. Engines fan queries out into sub-questions. A page answering the cluster outranks a page answering one slice of it.

15. Specific numbers, dates, and linked sources. Verifiable facts beat confident adjectives. The Princeton GEO study measured the lift directly: adding statistics raised visibility up to 41%, adding source citations up to 28%. Every claim you make should survive a click.

16. Content is fresh. Pages under three months old are up to 3x more likely to be cited than stale content, per RankSenseAI's 2026 report. Refresh priority pages every quarter — and show the updated date.

17. No JavaScript-rendered content blocks. Key content in the HTML source, not injected after render.

18. Clean heading hierarchy. One H1, H2s for sections, H3s for subsections, no skipped levels.

Dimension 3: Schema Markup (7 Points)

Schema is the most contested signal in AEO — the research genuinely disagrees on whether it lifts citations directly (RankSenseAI calls it the most disputed factor). Half the industry treats it as magic, the other half as theatre. It's neither: it's entity infrastructure. Do it, just don't expect it to carry the page. Our schema markup for AEO guide separates what works from what's cargo cult.

19. Article/BlogPosting schema on content pages. author, datePublished, dateModified, publisher, mainEntityOfPage.

20. FAQPage schema on FAQ sections. Maps directly onto how engines extract question-answer pairs.

21. Organization schema on homepage and about. name, url, logo, sameAs, foundingDate, founder — your machine-readable identity card.

22. Person schema on author bylines. Named humans with credentials and sameAs links. Authored content outperforms anonymous content.

23. HowTo schema on process content. For genuine step-by-step guides.

24. Everything validates. Run Google's Rich Results Test. Invalid schema is worse than none.

25. The five high-impact fields. author (Person + sameAs), mainEntityOfPage, datePublished + dateModified, publisher (Organization), articleSection. Most sites implement three and skip the rest.

Dimension 4: Authority Signals (8 Points)

Engines cite sources that other sources corroborate — it's the web's version of "nobody gets fired for citing IBM." This dimension is slow compounding work. Anyone selling you a shortcut here is selling you a bridge.

26. Author bylines with credentials and a bio link. Name, role, and a page that proves the person knows the subject.

27. Original data on the page. This is the single most defensible citation asset — a number only you can publish. It's also what earns the backlinks that build authority: Attrifast's 200-site AI traffic study gets cited constantly because nobody else has the dataset.

28. Third-party mentions on authoritative sites. Branded mentions correlate with AI visibility more strongly than backlinks do — 0.664 vs 0.218 in Ahrefs' 75,000-brand analysis. Reddit discussions specifically earn 3.4x higher citation rates than owned content.

29. Consistent brand entity across the web. Same name, same description, same facts everywhere. Build the Wikidata entry.

30. Topical clusters. A pillar page plus interlinked cluster posts. Depth on one topic beats breadth across ten.

31. Other sites cite your content. Which loops back to #27 — publish things worth citing.

32. Presence where engines source answers. Reddit, YouTube, G2, industry forums. Erlin's data puts Wikipedia mentions at 2.9x citation lift.

33. Your brand surfaces unprompted. Ask the engines about your category without naming yourself. If you don't appear, that's the gap.

Dimension 5: Measurement (7 Points)

You can't improve what you don't measure — and 78% of brands have factual errors in how AI engines describe them, so you're probably measuring less accurately than you think. Our AI visibility tracking guide covers the full framework.

34. Baseline citation rate. 20 target queries across ChatGPT, Perplexity, Claude, AI Overviews. Cited or not — write it down.

35. Weekly tracking, top 20 queries. Query, engine, date, cited Y/N, cited URL, which competitor appeared instead.

36. Monthly tracking, full query set. 30–40 queries. Single-month snapshots are noise; you want the trend.

37. Share of voice vs competitors. Of all citations on your queries, what fraction is yours? This is the number clients understand.

38. Mention rate tracked separately from citation rate. Mentioned ≠ linked. The gap tells you whether you have a brand problem or a structure problem.

39. AI referral traffic monitored. GA4 channel group for chatgpt.com, perplexity.ai, claude.ai — our attribution guide has the regex.

40. Quarterly re-audit. Engines retrain constantly. What passed in Q1 fails in Q3.

We Ran It on Ourselves: altyzo.com Scores 34/40

We sell AEO, so we get to be the first test subject. Here's our own audit, run on the site you're reading right now:

Two things worth stealing from that scorecard. First, authority is almost everyone's weakest dimension — it's also the one that compounds, so starting it beats perfecting anything else. Second, the checklist earns its keep by surfacing unflattering numbers. A tool that only ever tells you you're doing great is a mirror, not an audit — and we'd rather show you a real score than a polished one.

Frequently Asked Questions

What is a good AEO checklist score?

Treat 35+/40 as "citable," 25–34 as "structural work needed," and below 25 as "invisible." The thresholds are a practical framework, not a verified industry benchmark — your own baseline matters more than the absolute number.

How often should I run an AEO audit?

Quarterly for your top pages, monthly for your tracked query set. AI engines update their retrieval and citation behavior continuously — an audit is a snapshot, not a certificate.

What's the highest-impact single fix?

Depends on your score. If AI crawlers can't reach you (Dimension 1), nothing else matters. If they can, restructuring your top page to lead with a direct answer is the fastest visible change.

Does schema markup guarantee AI citations?

No. Research is split on whether schema directly lifts citations — RankSenseAI's report calls it the most contested factor. It builds entity clarity and disambiguation regardless, so implement it — just don't expect it to carry the page alone.

Running this across a client roster? The Altyzo platform automates the audit and tracking cycle for agencies — $30K license, self-hosted. Want it done for you instead? Altyzo Managed opens with the $7,500 Full-Surface Audit, credited to your first month.