aeo-content-audit
This skill should be used when the user asks to "audit content for AI search", "check if pages rank in ChatGPT", "run an AEO audit", "evaluate content for AI engines", "audit pages for Perplexity", "check AI search readiness", "assess content for answer engines", "review pages for AI citations", or any variation of auditing existing website content for answer engine optimization readiness.
AEO Content Audit
An AEO content audit evaluates how well your existing pages are structured for AI search engines to extract, cite, and recommend. It is not an SEO audit. SEO audits check rankings, backlinks, and technical health. AEO audits check extractability — can an AI engine read your page and lift a clean, accurate, citation-worthy answer?
Run an AEO audit before creating new content. Most sites have 50-100 existing pages that could rank in AI search with structural changes. Fixing existing pages is faster and higher-ROI than publishing new ones.
Audit Scope
What to audit
Prioritize pages by AI search potential, not by organic traffic.
| Priority | Page type | Why |
|---|---|---|
| 1 | Comparison pages (vs, alternatives) | Highest volume AI queries for SaaS |
| 2 | Category / definition pages | AI engines need clean category definitions |
| 3 | How-to guides and tutorials | Step-by-step content is highly extractable |
| 4 | Pricing pages | "How much does X cost?" is a top AI query |
| 5 | Feature / product pages | Build entity authority |
| 6 | Blog posts (top 20 by traffic) | May already rank — optimize for extraction |
Skip: Press releases, event pages, company news, career pages. AI engines rarely cite these.
Audit batch size
- First audit: top 20 pages by priority
- Ongoing: audit 10 pages/month on a rolling schedule
- Trigger re-audit: when AI search tools show citation drops
The AEO Audit Scorecard
Score each page on 10 criteria. Each criterion is 0-2 points. Maximum score: 20.
| # | Criterion | 0 (Fail) | 1 (Partial) | 2 (Pass) |
|---|---|---|---|---|
| 1 | Answer-first | Answer buried below fold or absent | Answer present but after 100+ words | Direct answer in first 50 words |
| 2 | Question-shaped H2s | Generic or clever H2s | Some H2s match queries | All H2s match real buyer queries |
| 3 | Declarative language | Hedged, passive, vague | Mixed hedged and declarative | Confident, specific, extractable |
| 4 | Tables | No tables | 1 table but comparison data in prose elsewhere | All comparison data in tables |
| 5 | Structured data | No schema markup | Basic Article only |
Page-type-specific schema (FAQPage, HowTo, Product) |
| 6 | Author + date | No author, no date | Author OR date present | Author + datePublished + dateModified all present |
| 7 | Recency | Last updated 12+ months ago | Updated 6-12 months ago | Updated within 6 months |
| 8 | Entity consistency | Brand name varies across page | Mostly consistent | Perfectly consistent brand name |
| 9 | Content depth | Thin or generic (< 500 words, no original insight) | Adequate coverage | Comprehensive with proprietary data or unique POV |
| 10 | Accessibility | Content gated, JS-rendered only, or image-based | Partially accessible | Fully crawlable HTML, no gates, no JS dependency |
Score interpretation
| Score | Rating | Action |
|---|---|---|
| 16-20 | AEO-ready | Monitor citations. Minor tweaks only |
| 11-15 | Fixable | Prioritize structural fixes. 2-4 hours of work per page |
| 6-10 | Major rework | Likely needs a rewrite with AEO-first structure |
| 0-5 | Not AEO-viable | Rebuild from scratch or deprioritize |
Audit Process
Step 1: Build the target query list
Before auditing pages, define what AI queries each page should answer.
Per page, identify:
- Primary query: the single most important question this page answers
- Secondary queries: 2-3 related questions the page should also address
- AI phrasing: how a user would ask this in ChatGPT vs Google (usually longer, more conversational)
| Page | Primary query | AI phrasing |
|---|---|---|
| /vs/notion-vs-confluence | "Notion vs Confluence" | "What's the difference between Notion and Confluence? Which is better for a startup?" |
| /what-is/revenue-intelligence | "What is revenue intelligence?" | "What is revenue intelligence and how does it work?" |
| /pricing | "[Product] pricing" | "How much does [Product] cost per month?" |
Step 2: Test in AI engines
For each page's primary query, test in all three major AI engines:
| Engine | URL | What to record |
|---|---|---|
| ChatGPT | chat.openai.com | Cited? Accurate? Who else cited? |
| Perplexity | perplexity.ai | Cited? Accurate? Source ranking? |
| Gemini | gemini.google.com | Cited? Accurate? How answer differs? |
Record per query:
- Were you cited? (Yes / No / Partially)
- Was the answer accurate? (Yes / No / Inaccurate details)
- Which competitors were cited instead?
- What did the cited source do that yours doesn't?
Step 3: Score each page
Apply the scorecard criteria. Be honest — a 7 is a 7, not "almost a 10 with some fixes."
Step 4: Diagnose patterns
After scoring all pages, look for systemic issues:
| Common pattern | Typical cause | Fix |
|---|---|---|
| Low scores on "answer-first" across all pages | Editorial style prioritizes narrative over directness | Create an AEO writing guide and retrain writers |
| No structured data on any page | Schema was never part of the publishing workflow | Add schema to publishing checklist, batch-add to existing pages |
| All pages fail recency | No content refresh process | Build quarterly refresh cycle |
| Tables missing everywhere | Writers default to prose | Add "use tables for comparisons" to content brief template |
| No authors on pages | Company-branded content only | Add real author bylines to all content |
Step 5: Prioritize fixes
Use the impact-effort matrix:
| Fix type | Impact | Effort | Do when? |
|---|---|---|---|
| Add answer to first 50 words | High | Low (15 min/page) | Immediately — batch all pages in a day |
| Add structured data | High | Low (30 min/page) | Week 1 |
| Rewrite H2s to question-shape | Medium | Low (20 min/page) | Week 1-2 |
| Add tables for comparisons | High | Medium (1 hr/page) | Week 2-3 |
| Add author + dates | Medium | Low (10 min/page) | Week 1 |
| Full content rewrite | High | High (4-8 hrs/page) | Prioritize top 5 pages only |
| Add original data / unique POV | Very high | High (varies) | Ongoing — build into editorial process |
Audit Output Template
Deliver the audit as a spreadsheet or table with these columns:
| Page URL | Primary query | Current citation? | Score (/20) | Top issue | Fix | Priority | Est. time |
|---|---|---|---|---|---|---|---|
| /vs/x-vs-y | "X vs Y" | Not cited | 8 | No answer in first 50 words, no tables | Rewrite intro, add comparison table | P1 | 2 hrs |
| /pricing | "X pricing" | Cited (inaccurate) | 12 | Outdated pricing, no schema | Update prices, add Product schema | P1 | 1 hr |
| /blog/how-to-z | "How to Z" | Cited | 16 | Missing dateModified | Add dateModified schema | P3 | 15 min |
Ongoing Monitoring
An AEO audit is not a one-time event. Build a monitoring cycle.
| Frequency | Activity |
|---|---|
| Weekly | Spot-check 3-5 priority queries in AI engines |
| Monthly | Re-test full query list (20-50 queries) across all engines |
| Quarterly | Re-score top 20 pages with full scorecard |
| On content publish | Run AEO checklist before every new page goes live |
| On product change | Update any page referencing changed features, pricing, or integrations |
Pre-Audit Checklist
Before starting an AEO audit:
- [ ] Target query list built (20-50 queries minimum)
- [ ] AI search testing accounts set up (ChatGPT, Perplexity, Gemini)
- [ ] Page priority list defined by page type
- [ ] AEO scorecard template ready
- [ ] Access to CMS for implementing fixes
- [ ] Access to schema markup tools or templates
- [ ] AI search monitoring tool set up (Profound, Otterly, or manual tracking sheet)
- [ ] Stakeholder buy-in for content updates (some pages may need rewriting)
- [ ] Baseline citations recorded before making changes
Anti-Pattern Check
- Auditing only blog posts → Blog posts are rarely the highest-value AEO pages. Start with comparison pages, definitions, and pricing. These get cited 3-5x more often
- Scoring pages without testing in actual AI engines → The scorecard predicts AEO readiness. The real test is asking the query in ChatGPT and seeing if you're cited. Always do both
- Fixing everything at once → Batch the quick wins (answer-first rewrites, schema markup, date additions) in week 1. Save full rewrites for the top 5 highest-impact pages
- Auditing once and never again → AI engine behavior changes. Sources get displaced. Build a monthly monitoring cycle or the audit becomes stale
- Only checking ChatGPT → Perplexity, Gemini, and Claude cite different sources. A page cited in ChatGPT may not be cited in Perplexity. Test all three
- Ignoring what competitors are doing right → When a competitor is cited and you're not, read their page. Note what they did structurally that you didn't. Copy the structure, not the content
Want agents that use skill files like this?
We customize skill files for your brand voice and methodology, then run content agents against them.
Book a call