▲ Now tracking visibility across Google, ChatGPT, Perplexity, Claude, Gemini & AI Overviews
Answer Engine Optimisation is what SEO becomes when search engines stop sending traffic and start writing answers. This page explains what AEO actually is, how the surfaces work, and what carries over from the SEO program you already run.
Answer Engine Optimisation (AEO) is the practice of structuring your web content, schema and entity signals so that AI-driven search interfaces, ChatGPT, Perplexity, Claude, Gemini, Google’s AI Overviews and Microsoft Copilot, cite and quote you when they generate answers.
Where classic SEO competes for a click out of a list of blue links, AEO competes for inclusion in a written answer. The user sees a paragraph, not a SERP. Your brand is either in the paragraph (cited, paraphrased, recommended) or it isn’t.
The mechanics are familiar. The same things that make Google understand your site, crawlability, structured data, topical authority, clean entities, are what make ChatGPT and Perplexity reach for it. The difference is what ‘winning’ looks like. You measure citation share, not ranking position. You track quotes, not impressions.
Practitioners, c. 2023, as ChatGPT plugins and Perplexity emerged
ChatGPT, Perplexity, Claude, Gemini, AI Overviews, Copilot
Entity-level authority + structured, citable content patterns
Citation share across engines, not ranking position
Weekly tracking across 6+ engines via prompt panels
01
01
03
Anthropic
04
05
Google Search
06
10 blue links on a SERP
A written answer with inline citations
A page that ranks for a keyword cluster
A passage that gets pulled into a generated answer
Links + content + technical health
Entity authority + schema + content structure + brand mentions
Ranking position, impressions, CTR, sessions
Citation share, quote frequency, sentiment, prompt coverage
Daily/weekly ranks, monthly traffic
Weekly prompt-panel scans across 6+ engines
3 to 9 months
4 to 12 weeks for citation lift, 6+ months for share dominance
Algorithm updates, link rot
Model refreshes, training-data cutoffs, prompt drift
Googlebot, Bingbot
GPTBot, PerplexityBot, ClaudeBot, Google-Extended, plus humans copy-pasting URLs into prompts
Stage 01
Build a clean, machine-readable identity for your brand: Organization schema, Wikipedia/Wikidata presence, consistent NAP across the web, sameAs relationships. Engines need to know who you are before they can decide whether to cite you.
A clinic registers as a Wikidata entity, links sameAs from its site, and ships full Organization + Service schema. Citation share in Perplexity rises within 8 weeks.
JSON-LD for Article, FAQ, HowTo, Product, Service, Organization, Person. Adds machine-readable scaffolding the engines reward when reranking.
Specific statistics, dated quotes, original research, named experts. Engines prefer to cite sources that contain concrete, attributable claims over generic prose.
Method 05
Decide which AI crawlers can access your site (GPTBot, PerplexityBot, ClaudeBot, Google-Extended) and which cannot. Ship an llms.txt that tells LLMs how to use your content.
Method 06
Define 50-200 prompts your customers ask. Track citation share across 6 engines weekly. Same discipline as rank-tracking, applied to a new surface.
A site engines cannot crawl cannot be cited. Core Web Vitals, render budget, schema, sitemaps, internal linking. All still essential.
If you didn't earn authority on a topic in Google, you won't earn citation share in ChatGPT either. The bar is the same; the surface is different.
Author bios, real expertise, citations to primary sources, brand consistency. Engines weigh these heavily when reranking candidates.
JSON-LD work that helped you win rich results in Google directly helps engines parse your content for AI synthesis.
Being #1 in Google for a query no longer guarantees being cited. AI Overviews and ChatGPT often pull from sources ranked #4-12 if those sources structure their answer better.
Engines don't think in keywords. Optimise for the underlying intent and the 20+ ways a user might phrase it conversationally.
Engines pull passages, not pages. A 5,000-word guide that hides the answer in section 7 loses to a focused, structured 800-word answer.
Links still matter as an authority signal, but unlinked brand mentions and entity associations now matter at least as much. Earn the mention.
Answer Engine Optimisation. The practice of optimising for citation and inclusion in AI-generated answers.
Retrieval-Augmented Generation. The technique answer engines use: retrieve sources first, then generate an answer using them.
A defined set of 50 to 200 prompts you track citation share against, weekly, across multiple engines.
Google’s separate crawler for Gemini training. Can be blocked without affecting Google Search indexing.
In AEO, the proportion of generated answers that mention your brand, by surface and prompt category.
The generated answer block Google shows above classic search results for many queries. Separate ranking layer from organic.
When an engine generates a factual claim that no source actually supports. Often happens to brands with weak entity authority.
Run your top 50 prompts across 6 engines. Benchmark citation share, sentiment, accuracy. Compared to 3 named competitors.
Audit Organization, Service, Article and Person schema. Fix entity inconsistencies. Open or block AI bot crawlers per policy.
Prompt panel rerun every week. Monthly executive report on share-of-voice trend, by engine, by prompt category.
Almost never as a default. Blocking GPTBot means giving up your share-of-voice in ChatGPT to your competitor. There are legitimate reasons to block (legal, copyright, paywall protection) but they should be a deliberate decision, not an accident in robots.txt.
Free 30-minute review across ChatGPT, Perplexity, Claude, Gemini, AI Overviews and Copilot. You will leave with a ranked list of where you are cited, where you are not, and what it would take to change that.
Senior strategist · No lock-in · Plain-English report
No spam. Unsubscribe in one click. We’ll never share your email.
