Answer Engine Optimisation (AEO) is the work of making an AI assistant name your business when a customer asks it a buying question. You'll also see it called Generative Engine Optimisation (GEO), GenAI visibility or AI search optimisation: different names, same discipline, and everything on this page applies whichever term you arrived with. (Not to be confused with Authorised Economic Operator, the UK customs status that shares the acronym. Our own measurement shows AI engines genuinely mix the two up.) When someone asks ChatGPT "what's the best storage company in Leeds?", the assistant gives one answer, not a page of links. AEO is how you measure whether that answer names you, fix the reasons it doesn't, and prove the change.
How an AI answer actually forms
Three stages, and each one can lose you the customer:
- Rewrite. The AI doesn't search the customer's words. It silently rewrites the question into several hidden sub-queries ("cheap student storage London", "student storage London prices") and searches those instead. If your content doesn't cover the machine's phrasings, you can lose with a great page.
- Retrieve. It pulls candidate pages from live search indexes. If your pages aren't machine-readable, prices in images, comparisons as graphics, no dates, no structured data, the machine reads past you.
- Select. From what it retrieved, it names whoever the most independent sources corroborate. This is where quality signals, reviews it can read, and presence in the sources it trusts decide the answer.
| What the customer asks | What the AI actually searches |
|---|---|
| "best student storage London" | cheap student storage London student storage London prices student storage London comparison |
| "best cosmetic dentist Manchester" | Invisalign cost Manchester dental implants finance Manchester cosmetic dentist Manchester reviews |
The hidden sub-queries: one question becomes the price, finance and proof searches that decide the answer.
What AEO work actually is
- Measurement: running real customer questions across the engines, repeatedly, because answers vary run to run, and recording who gets named.
- On-page work: answer-first copy, real HTML tables, visible dates, accurate structured data, the things that make a page liftable into an answer.
- Evidence work: crawlable review content and original data the AI can quote, honestly assembled from what customers already say.
- Off-site work: getting listed and cited in the sources the AI actually pulls from in your trade, directories, comparison sites, ranked lists.
It overlaps with SEO and does not replace it, we wrote a separate page on exactly that: AEO vs SEO.
What the field's own textbook says
O'Reilly published the first major-publisher book on AEO in July 2026 (Answer Engine Optimization, by Rodrigo Stockebrand, a 20-year search practitioner). On the question that matters most for owner-led businesses, whether AI visibility comes from being famous in the model's training data or from being retrieved when the engine searches live, it says, verbatim:
"For large, well-covered brands, parametric responses can be surprisingly rich. But for most small and mid-sized brands? Ask ChatGPT to answer about you with browsing turned off, and you’ll see how thin parametric memory really is. The right posture for those brands is to get the live search layer working first while building the long-running off-site footprint that future training cycles will pick up." — ch. 6
The live search layer, what gets retrieved and cited when the engine looks right now, is exactly the layer we measure and fix. Our own client data agrees with him: in a 75-question baseline, the brand was named in 23% of answers where one of its pages was cited, and effectively never (0%) in answers with no citation.
The measured examples are public: the garden rooms reading, the London loft conversions reading and the EV charger installers reading, plus the instrument test behind all of them.
Common questions
Is AEO the same as GEO?
In practice, yes. Answer Engine Optimisation (AEO) and Generative Engine Optimisation (GEO) describe the same discipline: getting AI-generated answers to name and recommend your business. You'll also see 'GenAI visibility', 'AI search optimisation' and 'LLM SEO'. The industry hasn't settled on one name; the work, measurement, retrievable content, corroborating sources, is identical. We say AEO; everything we do applies whichever term you use.
How long does AEO take to work?
On-page fixes can change AI answers within weeks of a re-crawl, because assistants retrieve from live search indexes, the lag is crawler re-indexing, not model retraining. Authority work, becoming a source the AI trusts and cites, builds over months. Honest AEO separates the two and never promises the slow one on the fast one's timeline.
Can I do AEO myself?
Some of it, yes: date your pages, use real HTML tables instead of graphics, answer questions directly, keep structured data accurate. The hard parts are measurement (one screenshot proves nothing, answers vary run to run) and knowing which fix matters for which question, which is what a measurement engine is for.
Is AEO just hype?
The hype is real and so is the scepticism. Our position: the channel is young, volumes are growing, and positioning early is cheap. We refuse to promise booking volumes, and we measure everything, which is the difference between early and gullible.
Updated July 2026