What this course is, and what it is not
This is a course about a young field, written carefully. After this lesson you can say what AEO and GEO mean, and, more importantly, you can look at any piece of advice about them and decide how much weight it deserves.
It is not a list of tricks. Nobody outside the companies that run AI search products can promise that a page will be quoted or cited, and anyone who does is selling something. The course will never promise rankings, traffic, citations, leads or revenue. What it can do is teach sound habits, tell you what the vendors have written down, and keep guesswork clearly labelled.
Everything here is dated as of 2026-11. Check the linked vendor documentation again before you rely on any detail.
Two terms
AEO, answer engine optimisation, is the older idea. People ask a question, and the search page or a voice assistant gives an answer directly: a featured snippet, a People Also Ask box, a spoken reply. AEO means writing and structuring your page so it is a good source for that kind of short answer.
GEO, generative engine optimisation, is the newer one. Here a language model writes a fresh answer, often with links or citations to sources it used. GEO means making your content easy for such systems to find, understand, quote and cite, and making sure your brand is described correctly.
These are practitioner labels, not official categories. Google's own guidance as of 2026-11 treats them as part of ordinary search work rather than separate disciplines (the next lessons look at exactly what it says). You will still see both words everywhere, so you need to know them.
The three evidence labels
Every important claim in this course carries one of three labels. Learn them; you can use them on anything you read.
Documented. The company that runs the product says so in its own official documentation or announcements. Example: Google Search Central's page on AI features says a page must be indexed and eligible to show with a snippet to appear in AI Overviews and AI Mode. That is documented, as of 2026-11.
Established practice. Not written down as a promise by a vendor, but long-standing, widely observed and sensible even without AI: clear headings, one topic per page, accurate titles, fast pages. These carry low risk because they help human readers too.
Hypothesis. A plausible idea, a practitioner's opinion, or something observed in a few cases. Many popular claims live here: "put the answer in the first 40 words", "add a table of definitions", "mention your brand on many sites". They may even be true for some systems. But they are not established, and the systems change without notice.
A useful habit: when you read advice, ask "who says so, and could they know?" A vendor can know how its own crawler behaves. A consultant watching a handful of prompts cannot know how a model chooses sources.
An example of labelling
Take this sentence from an imaginary blog post about a bakery website: "Adding FAQ schema makes AI assistants cite you." Break it into parts.
- "FAQ markup exists and is described on schema.org": documented by schema.org.
- "Google shows FAQ rich results": this was documented once; Google's own documentation now says the feature no longer appears in Search (lesson FAQPage, HowTo and Q&A markup after the deprecations).
- "FAQ markup makes AI assistants cite you": nobody has documented this. At best a hypothesis, and Google's guide lists overfocus on structured data as unnecessary for its generative features.
One sentence, three different strengths of evidence. Splitting claims like this is the main skill of the course.
Claim: "Put the direct answer in the first sentence."
Label: established practice for snippets and for readers;
hypothesis that it changes AI citations.
Claim: "A page must be indexed and snippet-eligible for AI Overviews."
Label: documented by Google Search Central, as of 2026-11.What you will learn
The course moves in six steps: how retrieval and AI answers work conceptually; what Google and Microsoft document; how to write and structure answer-friendly pages and build trust; structured data, including what has been removed; technical access, including robots.txt and the llms.txt proposal; and finally measurement, risks and a project. The project uses this site as the real example.
If you have not read about basic indexing and snippets, glance at the previous lessons of any SEO course first. This course assumes you know what a title tag and a robots.txt file are.
Common mistakes
- Treating GEO as a secret second internet that needs a separate strategy from ordinary search quality.
- Copying a tactic because an agency's case study mentions it. Case studies rarely show method or a control group.
- Believing a tool dashboard measures "what the AI thinks of you". Lesson Measuring AI visibility honestly explains what such tools can and cannot see.
- Reading an old article as current. AI features have been renamed and redesigned repeatedly.
- Spending effort on speculative tweaks while the page itself is thin, unclear or unreachable.
Exercise
- Find two recent articles about "optimising for AI search". For each, write down five claims and label every claim documented, established practice or hypothesis. Note how many carry a source you could verify.
- Open Google Search Central's page called "AI features and your website". Write three things it documents in your own words, without copying.
- Pick one of your own pages and list what a reader needs from it. Do not change anything yet; you will use this list in the project lesson.
Show a sample answer
For step 1, a typical split: a claim that a page must be crawlable is established practice and, for Google, documented. A claim that a particular heading style raises citation rates is hypothesis, unless the article names a vendor document. If almost every claim lacks a verifiable source, treat the article as opinion.
For step 3, a good list is short: the question the page answers, who it is for, what proof you can show, and what the reader does next.
Key points
- AEO (answer engine optimisation) aims at direct answers such as featured snippets, People Also Ask and voice answers.
- GEO (generative engine optimisation) aims at being understood, quoted and cited by AI assistants and AI search features.
- The field is young, and much online advice is hypothesis presented as fact.
- This course labels each claim: documented, established practice or hypothesis.
- As of 2026-11, Google says there is no separate optimisation needed for its AI features beyond good Search practice.