AEO & GEO skills
AEO & GEO Skills-md for SEOs, Marketers, Claude & Codex
AEO and GEO skills are SKILL.md files that turn Claude, Codex and other AI agents into careful AI search strategists. They audit how a brand appears in Google AI Overviews and AI Mode, ChatGPT, Perplexity, Copilot, Claude and Gemini, fix access, content, entity and citation gaps, track results with first-party data, and prepare sites for agents that shop and book. Every claim is labelled Confirmed, Consensus or Speculation. All 13 are below: copy one, download it, or take the set.
13 skills, 17,000 words, version 1.0.0, researched October 2026.
AI Search Visibility Audit
Audit how a brand, product or person shows up across AI answer engines (Google AI Overviews and AI Mode, ChatGPT search, Perplexity, Microsoft Copilot, Claude and Gemini): crawler and index access, retrieval eligibility, answer-ready content, entity clarity, third-party citations, brand accuracy and measurement, scored across seven layers with every claim labelled Confirmed, Consensus or Speculation and a 90-day plan.
AI Visibility Tracking and Reporting
Measure visibility in AI answers you can trust: fixed prompt sets across ChatGPT, Perplexity, Copilot, Claude, Gemini and Google AI Overviews and AI Mode, repeated runs, mention, citation, accuracy and share-of-voice rates, Google Search Console's Generative AI performance report, Bing Webmaster Tools AI Performance, GA4 AI referral channels and bot logs, turned into a monthly report.
Query Fan-Out and Prompt Research
Research the long, conversational prompts buyers ask AI assistants, model the sub-queries AI search runs behind them (query fan-out, which Google documents for AI Overviews and AI Mode), check who ranks and gets cited for each, and map every sub-query to an existing page or section without creating a page per variation.
Answer-Ready Content for AI Search
Rewrite pages so AI answer engines can use and credit them: non-commodity content with first-hand evidence, direct answer passages under question headings, sourced statistics and expert quotes, comparison tables, original data, explicit brand and author attribution and current facts, while ignoring myths such as chunking, special markup and writing for robots.
Google AI Overviews and AI Mode
Diagnose and improve visibility in Google AI Overviews and AI Mode: eligibility, retrieval and query fan-out, what gets cited and why, snippet controls and the Search Console AI opt-out, the Generative AI performance report, click impact from Pew, Ahrefs and Seer studies, images, video, local and shopping, and the post-click experience for visitors who still arrive.
ChatGPT Search Visibility
Get a brand found, cited and described correctly in ChatGPT: OAI-SearchBot, ChatGPT-User and GPTBot access verified by IP, search-grounded answers versus model knowledge, Bing indexing as a baseline, prompt testing with search on and off, utm_source=chatgpt.com referral tracking and fixes for wrong facts and missing category mentions.
Perplexity, Copilot, Claude and Gemini Visibility
Improve visibility engine by engine in Perplexity, Microsoft Copilot and Bing AI summaries, Claude and the Gemini app: each engine's retrieval source and documented crawlers, Bing Webmaster Tools AI Performance citation data, IndexNow, Google-Extended and Claude crawler decisions, per-engine prompt testing and diagnosis.
Entity Optimisation for AI Search
Make a brand, product or person unambiguous to search engines and language models: a signed-off fact sheet and canonical descriptor, an entity home page, connected Organization and Person markup with stable @id and sameAs, consistent official profiles, disambiguation for shared names, Wikidata and Wikipedia rules, knowledge panel verification and independent confirmation.
Earning Citations and Mentions for AI Answers
Get named in AI answers by earning authentic, accurate mentions on the sources engines cite: build a citation map from real prompts, find gaps on review platforms, editorial comparisons, media, communities and video, create citable assets such as original data, run honest outreach and community participation, and track mentions and cited-source presence with a risk register for tactics to avoid.
AI Crawler Access and Technical Readiness
Set and enforce a deliberate policy for AI crawlers: search and index bots (OAI-SearchBot, Claude-SearchBot, PerplexityBot, Bingbot), user-triggered fetchers (ChatGPT-User, Claude-User, Perplexity-User) and training crawlers and tokens (GPTBot, ClaudeBot, CCBot, Google-Extended), aligned across robots.txt, CDN and WAF rules and logs, verified by IP, with server-rendered content, IndexNow and an honest view of llms.txt.
Brand Accuracy in AI Answers
Find what ChatGPT, Perplexity, Copilot, Claude, Gemini and Google AI features get wrong about a brand (prices, products, leadership, safety, regulation, comparisons), grade each error by severity, trace it to outdated own pages, stale third-party profiles, old coverage or entity confusion, fix it at the source and re-test monthly.
Agent Readiness for AI Browsers and Assistants
Prepare a site for AI agents that browse and act for people: audit key tasks with real browser agents, fix the accessibility tree, labels, semantic HTML and flows agents rely on, keep prices and policies machine-clear, tune bot protection for user-directed agents, and assess agent commerce protocols such as Google's Universal Commerce Protocol, OpenAI and Stripe's Agentic Commerce Protocol and MCP.
Product Visibility in AI Shopping and Commerce
Get products shortlisted, priced correctly and purchasable in AI shopping answers across Google AI Mode, ChatGPT, Copilot and Perplexity: shopping prompt tests, Merchant Center and feed quality, shopper-intent attributes, page, markup and feed alignment, reviews and editorial comparisons, and agent checkout through the Universal Commerce Protocol, the Agentic Commerce Protocol and platform integrations.
No skill matches that search. Clear it or choose All.
On this page
What AEO and GEO skills do for an AI agent
AI search is the part of marketing with the most confident advice and the least evidence. Ask a general model how to “rank in ChatGPT” and you will often hear about llms.txt files, chunked content, special schema and planted Reddit posts. Google has said in writing that Search ignores llms.txt and doesn’t need chunking or special markup, and that inauthentic mentions don’t help. Server logs show almost no AI crawler reads llms.txt. Much of the popular playbook is noise.
These skills start from what platforms document and what independent studies measure. They check whether AI crawlers can reach you and read your content, whether your pages rank for the searches AI engines run behind each question, whether the web describes your brand consistently and independently, and what each engine actually says about you. They report with Google’s Generative AI performance report, Bing’s AI Performance citations, GA4 referrals and fixed prompt sets, not screenshots.
AEO, GEO and SEO: what the platforms say
AEO usually means answer engine optimisation: being the source an AI or search feature quotes as the answer. Microsoft also uses AEO for agentic engine optimisation. GEO means generative engine optimisation: being included and represented well in generated answers. Google’s position is that optimising for its generative AI features is still SEO. Microsoft treats AEO and GEO as a discipline in its own right and gives site owners citation data. The skills take a practical line: SEO is the foundation, and AI search adds engine-specific access, measurement, entity and citation work on top.
Where to start: baseline every engine
Start with the ai-search-visibility-audit skill. It runs a fixed set of buyer prompts across every engine that matters, then audits seven layers: access, eligibility, retrievability, answer-readiness, entity clarity, third-party presence and measurement. Each finding goes to a specialist: blocked crawlers to ai-crawler-access, weak pages to answer-ready-content, wrong facts to brand-accuracy-ai-answers and missing recommendations to ai-citation-earning.
Once the baseline exists, ai-visibility-tracking turns it into a monthly report. Retailers and booking businesses should add agent-readiness and ai-commerce-product-visibility early, because assistants already compare and buy.
Which skill fixes which problem, and the evidence
Use this table to go straight to the right skill when you already know the symptom.
| Symptom or task | Likely cause | Skill to use |
|---|---|---|
| You don’t know if AI engines mention you, or where to start | No baseline across engines | ai-search-visibility-audit |
| Leadership wants AEO and GEO results reported | No fixed prompt set or first-party data | ai-visibility-tracking |
| Buyers ask long questions your keyword list misses | Content planned around short keywords | query-fan-out-mapping |
| Pages rank but are never cited | Commodity content, no clear answers or evidence | answer-ready-content |
| Clicks fell while Google rankings held | AI Overviews absorbing clicks, or you aren’t cited | ai-overviews-ai-mode |
| ChatGPT recommends competitors, not you | Blocked OAI-SearchBot, weak results or few mentions | chatgpt-search-visibility |
| One engine mentions you and another doesn’t | Engine-specific access or indexing gaps (often Bing) | answer-engine-platforms |
| AI describes you vaguely or confuses you with others | Unclear entity, inconsistent profiles | entity-optimisation |
| You’re missing from “best X for Y” answers | Few independent mentions on cited sources | ai-citation-earning |
| AI crawlers never visit, or get blocked | robots.txt, CDN or WAF rules, client-side rendering | ai-crawler-access |
| AI quotes old prices or wrong facts | Outdated pages and third-party profiles | brand-accuracy-ai-answers |
| Assistants can’t complete tasks on your site | Hidden facts, unlabeled forms, blanket bot blocks | agent-readiness |
| Products missing from AI shopping answers | Weak feeds, missing attributes, inconsistent prices | ai-commerce-product-visibility |
The 2026 evidence base
Every skill in this set separates what platforms have documented from what studies and practitioners report. These are the facts the skills build on, checked on 10 October 2026:
| Finding | Source | Label |
|---|---|---|
| AI Overviews and AI Mode use retrieval-augmented generation and query fan-out over Google’s index, rooted in core ranking systems | Google Search Central guide to generative AI features (updated July 2026) | Confirmed |
| To appear, a page must be indexed and snippet-eligible, and the site included in generative AI features in Search Console | Same guide | Confirmed |
| Google Search ignores llms.txt; chunking content and special markup aren’t required | Same guide | Confirmed |
| Search Console’s Generative AI performance report shows impressions in AI Overviews and AI Mode (no AI-only clicks) | Google, announced 3 June 2026; all sites by 31 August 2026 | Confirmed |
| Bing Webmaster Tools AI Performance shows Copilot citations by page and grounding query, with citation share | Microsoft, 2026 | Confirmed |
| Sites that block OAI-SearchBot won’t be shown in ChatGPT search answers | OpenAI crawler documentation | Confirmed |
| Searchers clicked a result on 8% of visits with an AI summary versus 15% without | Pew Research Center, July 2025 (900 US adults, 68,879 searches) | Study |
| Position-one CTR about 34.5% lower when an AI Overview appears | Ahrefs, April 2025 (300,000 keywords) | Study |
| 97% of llms.txt files received no requests in May 2026 | Ahrefs, June 2026 (137,210 domains) | Study |
| Adding citations, quotes and statistics raised generative visibility up to about 40% on a benchmark | Aggarwal et al., GEO, KDD 2024 | Research, not live engines |
| Google launched the Universal Commerce Protocol for agent checkout with Shopify, Etsy, Wayfair, Target and Walmart | Google, 11 January 2026 | Confirmed |
Example prompts for each skill
Each skill triggers on everyday wording. These prompts are good starting points:
- Audit: “Audit how our brand shows up in ChatGPT, Perplexity, Copilot and AI Overviews. Here are 30 buyer questions.”
- Tracking: “Set up monthly AI visibility reporting with Search Console, Bing AI Performance and GA4.”
- Fan-out: “Map the questions people ask AI about switching accounting software, and which of our pages should answer each sub-query.”
- Content: “This guide ranks fifth but is never cited. Make it citable.”
- AI Overviews: “Our clicks fell 40% but rankings held. What is happening in AI Overviews and AI Mode?”
- ChatGPT: “Why does ChatGPT recommend three competitors and never us?”
- Other engines: “We appear in Perplexity but never in Copilot. Diagnose it.”
- Entity: “AI tools describe us as a design agency. Fix our entity signals.”
- Citations: “Build a plan to get into AI answers for best payroll software for startups.”
- Crawlers: “We want AI search visibility but no training use. Write the robots.txt and CDN rules.”
- Accuracy: “ChatGPT quotes our 2024 prices. Find the sources and fix them.”
- Agents: “Can a browser agent find our SSO plan and book a demo? Test it.”
- Commerce: “Why are our jackets missing from AI shopping answers?”
What the agent needs from you
AI search work depends on first-party data. The most useful inputs are:
- Buyer questions: 20 to 50 real questions from sales, support and search data.
- Google Search Console: the Performance and Generative AI performance reports.
- Bing Webmaster Tools: indexing and AI Performance citations.
- GA4: referrals from chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com and claude.ai.
- Access data: robots.txt, CDN and WAF bot settings, and 30 days of logs.
- Brand facts: a signed-off fact sheet with prices, products, people and policies.
- Commerce data: feeds and Merchant Center diagnostics, if you sell products.
If something is missing, the skill works with what is public and says which findings would change with more data.
What an agent cannot do, and myths to ignore
No agent, tool or agency can guarantee a mention in ChatGPT or a citation in AI Overviews. Answers vary by session, location and model; the skills measure rates over many prompts and say so. An agent can’t see your Search Console, Bing or GA4 data unless you export or connect it, and it can’t create independent coverage for you: earning mentions takes real data, expertise and outreach from your team.
| Myth | What the evidence says |
|---|---|
| An llms.txt file gets you into AI answers | Google Search ignores it; 97% of files got no requests in a 137,210-domain study |
| You must chunk content into tiny blocks | Google says it isn’t required; write for readers |
| Special schema gets you into AI features | Google says no special markup is needed |
| Blocking GPTBot removes you from ChatGPT search | OAI-SearchBot controls ChatGPT search; GPTBot is training |
| Planted reviews and Reddit posts work | Against spam policies and platform rules, and easy to expose |
Frequently asked questions
What is the difference between AEO and GEO?
AEO usually means answer engine optimisation: being the source an AI or search feature uses as the answer. GEO means generative engine optimisation: being included and described well in AI-generated answers. In practice the work overlaps, and both rest on SEO.
Is AEO or GEO different from SEO?
Google says optimising for its AI features is still SEO. Microsoft treats AEO and GEO as their own discipline. The skills treat SEO as the foundation and add AI-specific work: crawler policy, entity clarity, citations, brand accuracy and AI measurement.
Can these skills get my brand into ChatGPT?
They find and fix the reasons you’re missing: blocked crawlers, weak pages for the searches ChatGPT runs, unclear entity signals and few independent mentions. No one can guarantee placement, and the skills say so.
How do I measure AI search visibility?
Use Google Search Console’s Generative AI performance report, Bing Webmaster Tools AI Performance, GA4 referrals from AI domains, and a fixed prompt set run monthly across engines. The tracking skill sets all of this up.
Should I block AI crawlers?
Decide by type. Blocking search crawlers such as OAI-SearchBot removes you from that engine’s answers. Training crawlers such as GPTBot are a business decision. User-triggered fetchers represent real people asking about your page.
Do I need an llms.txt file?
Not for visibility. Google Search ignores it and studies show very few AI crawler requests. It’s fine to publish one for developer tools, but don’t treat it as a ranking lever.
Does structured data help AI search?
Google says it isn’t required for AI features. It still helps rich results and entity understanding, and the skills use it for that, not as a shortcut.
What is agentic commerce?
AI agents completing purchases or bookings for people. Google’s Universal Commerce Protocol and OpenAI and Stripe’s Agentic Commerce Protocol let assistants check out with merchants. The agent and commerce skills cover readiness.
Sources and more marketing skills
- Guide to generative AI features on Google Search, Google Search Central
- Search Generative AI performance reports, Google Search Central Blog, June 2026
- OpenAI crawlers and Perplexity crawlers
- Pew Research Center on AI summaries and clicks, July 2025
- Ahrefs llms.txt server-log study, June 2026
- GEO research paper, Aggarwal et al., KDD 2024
- Universal Commerce Protocol
Checked on 10 October 2026. AI search depends on crawling and content basics, so pair this set with the Technical SEO skills and the Content and On-Page SEO skills. See every subcategory on the Marketing Skills-md page, or read how I approach this work on my AEO and GEO page.
