What Is AEO? Answer Engine Optimisation Explained for UK Businesses

AEO stands for answer engine optimisation: the work of earning recommendations and citations inside AI assistants such as ChatGPT, Claude, Gemini and Google AI Overviews. Generative engine optimisation (GEO) and AI SEO name the same job with different labels.

It covers technical access for AI crawlers, answer-first pages, entity consistency, and earned mentions in the sources those engines already trust. Classic SEO still matters. AEO changes what counts as a result from ranking position to whether an assistant names and cites you on the questions your buyers type.

We run this on code23.com and sell it as a productised programme under AEO & AI Search Visibility. The longer playbook for how assistants pick brands sits in How do you get recommended by ChatGPT and Claude?.

How ChatGPT, Claude, Gemini and Google AI Overviews assemble an answer

Treat the four engines as different retrieval machines, not as one “AI search” blob. Some answers come from parametric memory. Some are assembled live from web sources. The mix varies by product, prompt and day.

Research summarised from Searchable’s Claude study (analysis across 90,000+ Claude sources) is directionally clear: Claude answers from memory roughly 52% of the time and cites a brand’s own site only about 32% of the time. ChatGPT cites owned sites more often (around 61% in comparable framing) and searches the web on most prompts. Google AI Overviews and Gemini sit closer to classic search retrieval, still lifting passages rather than whole pages.

When Claude does search, it follows Brave’s top results over Google’s at roughly 6:1 in the same briefed research. Ranking only in Google while ignoring Brave leaves a blind spot. Within Brave’s top 10, position barely matters; Claude re-ranks by prompt relevance.

Correlations reported in the wider Ahrefs 2025 research stream (treat as hypotheses for your category, not guarantees): YouTube mentions ~0.737 with AI Overview-style visibility, branded web mentions ~0.664, classic backlinks weaker ~0.218. Being written about and shown in video beats another thin service page.

None of this replaces ordinary search hygiene. Entity consistency and earned mentions move up; keyword-stuffed owned pages move down. Technical SEO still gates whether crawlers can fetch you - see our website checklist for a technical SEO audit.

Where AI citations come from by industry: why two thirds of the work is off your site

Searchable’s analysis of AI citations (September 2026) measured the share of citations that come from a brand’s own domain, by industry. The numbers are medians across brands with enough sample, not a promise for your niche.

Industry	Share of AI citations from the brand’s own domain
Professional services	26.9%
Retail and ecommerce	28.6%
Finance	29.6%
Hospitality	33.7%
B2B tech and media	41.4%
Transport and logistics	43.5%

Even in the strongest brand-owned industries in that cut, most citations still come from somewhere else. Searchable’s rule of thumb in the same webinar puts AEO at roughly two thirds off-site sources. Earned media accounts for 82-89% of AI citations in the framing we work from across the playbook research.

What that means for a UK firm:

• Owned pillars still matter for ChatGPT-style retrieval and for humans who click through
• Third-party listicles, review sites, YouTube, trade press and directories often decide the shortlist
• Scope the programme to which third-party sites AI already cites in your category, not to a generic “publish more blogs” plan

Professional services sit at the bottom of that owned-share table. If you sell advice, audits or agency work, off-site mentions are not a nice-to-have. They are most of the citation surface.

What is an example of AEO?

A practical example of AEO: you discover ChatGPT recommends three competitors on “best [your category] agency UK” while your site is absent. You open blocked crawlers, publish an answer-first page that states a citable method, earn a mention on a listicle those answers already use, then re-measure the same question thirty days later against where you started.

That sequence is the whole job in miniature: prove the gap, fix what you control on your site, earn presence where assistants already look, then measure the same prompts again. Guaranteed placements and paid mentions are not part of it.

Worked example: AEO for a UK professional services firm (hypothetical)

This section is a labelled hypothetical. It is not a client case study and it does not claim Code23 results.

Imagine a 12-person Reading consultancy selling B2B compliance reviews. Buyers type questions such as “best compliance consultancy UK”, “how much does a GDPR audit cost UK” and “compliance consultant near Reading”. Classic Google rankings for brand terms look healthy. Unbranded assistant answers name two London rivals and one directory listicle the firm has never pitched.

Where they stand at the start (hypothetical numbers for illustration only):

• Cited on 4 of 50 tracked questions across ChatGPT, Claude, Gemini and Google AI Overviews
• Own-domain share of citations in the mid-20s, consistent with Searchable’s professional-services median (26.9%, September 2026)
• robots.txt blocking ClaudeBot; FAQ pages buried behind client-side tabs
• Three different one-line descriptions across the site, LinkedIn and Clutch

A 90-day AEO programme for that firm would typically:

1. Unlock crawler access and ship answer-first versions of the money pages
2. Lock one canonical entity description everywhere
3. Publish cost and method pillars that state citable answers in the first screen
4. Pitch the three listicles and two trade sites already feeding the answers
5. Re-run the same 50 questions and decide at month three whether to continue, change course or expand

Success in this hypothetical is a measurable lift in citation presence and sentiment on the commercial question list, with the three numbers that show it is working in the firm’s own analytics. Failure is activity without a starting picture, or a retainer that only ships owned blog posts while ignoring which sites AI already cites.

What an AEO audit checks

An AEO audit measures where you stand before anyone changes the site. Ours is the fixed-price AI Visibility & Search Audit at £1,500 + VAT: fifty buyer questions, four engines, ten working days, credited against the first retainer month if you continue.

A serious audit checks:

Check	What you should see
Question list	Commercial questions across learn, consider and purchase, overwhelmingly unbranded, agreed with you before measurement
Starting picture	Visibility, citations and sentiment across ChatGPT, Claude, Gemini and Google AI Overviews before any change
Gap	Which questions you are absent from, and which competitors appear instead
Where AI looks	Which third-party sites those answers already pull from
Crawler access	OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, PerplexityBot, Googlebot and friends can fetch clean HTML
Schema and entity	Organization, Service and FAQ markup that match one canonical description
Answer-first structure	Question-shaped pages with a direct 2-3 sentence answer in the first screen
Analytics	Connected GA4 and Search Console so later reporting of the three working numbers has a home

Without that starting picture, nobody can show what changed after three months of AI SEO or generative engine optimisation work. Measuring first is buyer protection, not an agency formality.

Ongoing monitoring and delivery are quoted from the audit roadmap: measurement only, closing the gaps in the codebase and content, or winning mentions away from competitors on the sites AI already quotes. Some clients keep related care under Support & Growth when the stack already sits with us.

AEO vs SEO: is AEO similar to SEO?

Yes in the foundations: a clearable site, honest measurement and content that answers real questions. It differs in what counts as a result. Classic SEO tracks rankings and clicks. AEO tracks whether assistants name and cite you on the questions your buyers type. For the full comparison, including GEO labelling, a worked week and whether you need both, see AEO vs SEO.

Technical SEO still gates retrieval. If the page cannot be fetched, indexed or rendered as clean HTML, answer engines have nothing honest to quote. Schema still reduces ambiguity. Internal links and crawl depth still matter. What changes is priority and reporting: citation presence, which sites AI quotes, and sentiment sit beside ordinary search metrics, not instead of them after you abandon Google.

What is genuinely new vs ordinary SEO

Most agencies already have technical SEO, content and digital PR skills. What is new in answer engine optimisation is the operating layer around assistants.

Buyer question lists. You size and weight a commercial list of questions your buyers ask AI assistants, across learn, consider and purchase. A twenty-question toy list is not a programme. Coverage and commercial intent matter more than vanity volume.

Citation tracking. What counts is whether you are named and linked inside the answer, not only whether you rank for a head term.

Earning mentions where AI already looks. You map the listicles, review sites, YouTube channels and press that AI already cites in your category, then work those sources. Owned posts alone fight the industry numbers above.

Tying visibility to outcomes. Crawler activity, AI referral quality and branded search are the three numbers that show it is working. None of them alone proves revenue. Monthly reports must say what changed, what you shipped, what it means and what happens next. Month three is the review against where you started.

Conductor CMO study, 2026: 94% of enterprises plan to increase AEO or GEO investment in 2026. Demand is rising. Undifferentiated “we do AI search” claims are rising with it. Packaging, a clear starting picture and honest reporting are what separate a useful retainer from a label.

How do you optimise for answer engines?

Start with an agreed list of buyer questions and a clear picture of where you stand across the engines your buyers use. Fix crawler access and entity consistency first, publish answer-first pages for the biggest question gaps, then earn mentions in the third-party sources those answers already cite. Re-measure the same questions on a fixed cadence and decide at month three whether to continue, change course or expand.

That is the optimisation loop in plain language. Tools help with sampling and reporting. People still own diagnosis, codebase fixes and the quarterly call. Anyone selling a plugin that “submits” your brand to ChatGPT is selling a fairy tale.

What is the difference between AEO, GEO and AI SEO?

Answer engine optimisation (AEO), generative engine optimisation (GEO) and AI SEO / AI search optimisation are overlapping labels for the same job: becoming the brand assistants retrieve and cite. We use AEO on our service page, keep GEO and AI SEO in the copy so searchers find us, and run one programme rather than three retainers.

Common mistakes

Twenty-question lists. Too thin to represent the funnel, too easy to game in a pitch deck, too weak for month-three decisions.

Paid mentions dressed as earned media. Assistants and buyers both notice. Paid directory spam is not the same as an editorial listicle citation.

No starting picture. Activity without a clear measure of where you started is storytelling. You cannot prove contribution later.

Blocking AI crawlers. WAF rules, robots blocks and JavaScript-only rendering silently opt you out of retrieval. “Stop scrapers” is not the same as “stop every assistant”.

Owned content only. Publishing more on your domain while ignoring which sites AI already cites fights the industry numbers above, especially in professional services.

Guaranteed placements. Anyone promising a fixed ChatGPT listing is promising something they cannot keep.

FAQ

What is an example of AEO?

A practical example: you find ChatGPT names competitors on a high-intent question where you are absent, unlock crawlers, publish an answer-first page, earn a mention on a source those answers already use, then re-measure the same question against an agreed starting picture.

AEO vs SEO: is AEO similar to SEO?

Yes on foundations (clearable site, honest measurement, real answers). It differs on what counts as a result: citations and recommendations inside assistants, not only rankings and clicks.

How do you optimise for answer engines??

Measure a commercial list of buyer questions across engines, fix crawler and entity issues, ship answer-first pages for real gaps, earn mentions on the sites AI already quotes, and report the same questions monthly with a month-three review against where you started.

What is the difference between AEO, GEO and AI SEO?

They are overlapping names for one programme. AEO emphasises answer engines, GEO emphasises generative engines, AI SEO is the search-shaped label. Buy the work, not three retainers.

Does AI SEO really work?

It works as a contribution, not a guarantee. Conductor CMO study, 2026, shows enterprise investment rising. Searchable’s September 2026 data on which sites AI cites shows most citations in many industries still come from third parties. Measure in your own analytics. Do not buy a placement promise.

Assistants will keep changing. The durable advantages stay dull: identical facts everywhere, people writing about you where models already trust, pages that answer in the first screen, and crawlers that can fetch them. That is how you earn recommendations in a way that survives a model update.

Start with the free snapshot or the fixed audit on AEO & AI Search Visibility, or read How do you get recommended by ChatGPT and Claude? for the recommendation mechanics. For UK prices and how ongoing work is quoted after the audit, see How much does AEO cost in the UK?.