Generative engine optimization
Generative engine optimization (GEO) is the work of making your content the source an AI engine retrieves, quotes and cites when it writes an answer. It sits on top of SEO, changes how pages are written, and is measured by how often the answer names you.
Optimizing content and the sources around it so that generative AI engines (ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini, Copilot) retrieve it, use it in their answer and cite it. Also called AEO, LLMO or AI search optimization.
Where the term comes from
The term comes from GEO: Generative Engine Optimization, a 2023 paper by researchers at Princeton and partner institutions, later published at KDD 2024. They built a benchmark of queries, rewrote source pages in different ways, and measured how much more visible each version became in generative engine answers.
The headline finding: the right edits raised a source’s visibility in the generated answer by up to 40%. The edits that helped most were the ones that make a page more credible to quote: citing sources, adding quotations from relevant authorities, and adding statistics. Keyword stuffing, the reflex of old SEO, did little or nothing.
GEO, AEO and SEO
| Dimension | SEO | GEO | AEO |
|---|---|---|---|
| The goal | Rank for a typed query | Be retrieved and quoted in a generated answer | Be cited and recommended in the answer |
| The unit | Keywords | Passages and claims | Questions and personas |
| What wins | Relevance, links, speed | Statistics, sources, quotable answers | Proof on your site and across the web |
| Measured by | Rankings and clicks | Citations in answers | Share of answer |
| Who says it | Everyone | Researchers and SEO press | Agencies and buyers |
Same work, different emphasis. Every AEO.ca plan does all three, because an engine can only quote a page it found, and only recommends a company the rest of the web agrees on. More on the names.
What GEO work looks like
The answer in the first two sentences, then the proof. Engines lift passages, so each one has to stand on its own.
Dated numbers with a source on every claim worth quoting. The study found this among the strongest levers.
Named experts, real quotations and a clear point of view an engine can attribute to you.
Organization, Service, FAQ and Article markup that says exactly what the visible page says.
Press, reviews, forums and video that repeat the same facts, so the engine trusts them.
Pages AI crawlers can reach and render. llms.txt as hygiene, never as a trick.
Proof
Keywords where Google’s AI Overview cites the client. Semrush, Canadian database. All five cases.
GEO in Canada
Canadian answers draw on Canadian press, review sites and directories, and in Quebec and Ottawa the same question gets asked in French against a different set of sources. French-speaking members of our team run French GEO work, priced the same as English.
How we start
The questions your buyers put to AI, by persona.
How often each engine names you today, and who it names instead.
The pages and sites the engines quote for your category.
Agreed together, reported monthly on two instruments.
Plans
Be the answer in your own market
One market. The SEO foundation, persona-planned content, structured data, light PR and links.
Reach the parts of the web your site cannot
AEO and SEO plus PR and social media on one media plan, for more markets or a harder category.
Your name everywhere the agents read
Everything in Growth plus branded content, press, events and the full awards program.
Same job, different label. GEO is the research term for optimizing content so generative engines retrieve and cite it. AEO is the business term for being cited and recommended in the answer. We use AEO as the umbrella and do GEO work inside every plan.
No. Generative engines retrieve from the web, and the pages they retrieve are mostly the pages that already rank. In our experience about 70% of the work overlaps. GEO changes how the content is written and where else it has to exist.
Google AI Overviews and AI Mode, ChatGPT, Perplexity, Gemini and Copilot. Each retrieves differently, so we track each one separately against the same question panel.
No, and nobody honestly can. Nobody controls what an engine says. We take a baseline, do the work, and report how often you are named against that baseline every month, on two instruments.
No. Staff writers write it, and it reads at least 80% human by Grammarly’s measure. An engine gets nothing from text it could have produced itself.
At AEO.ca it is part of every plan: Local from $5,000 a month, Growth $7,500 to $15,000, Enterprise $20,000 to $30,000. Month one is setup, with no setup fee.