Answer Engine Optimisation, explained without the hype

What actually happens when someone asks an assistant a question

The assistant retrieves a set of candidate sources, extracts passages it judges relevant, synthesises them into one answer, and then decides which sources to name. Ranking influences the first step. Everything after it is decided by how quotable and how verifiable your content is.

That is the whole reason AEO is a separate discipline. You can win the retrieval step and still lose the citation step, which is the one that sends you a customer.

Ranking gets you into the candidate set. Being quotable is what gets you named in the answer.

The six things that decide whether you get cited

• Structured data. Typed entities with stable identifiers are far easier to quote confidently than prose an engine has to interpret.
• Self-contained passages. A sentence that needs the previous paragraph to make sense cannot be lifted.
• Factual consistency. If your service area says one thing on your site and another on your Business Profile, an engine has a reason to hesitate.
• Depth on a narrow topic. Three articles that genuinely cover one subject beat thirty that touch everything.
• Sourced statistics. An unsourced number is a liability, because attributing it makes the engine responsible for it.
• Freshness signals. Dated and updated content is preferred where the answer could plausibly have changed.

What to actually do first

Start with schema coverage, because it is the cheapest and most mechanical win. Organisation, LocalBusiness, Service, FAQ and Breadcrumb markup, connected through consistent identifiers so the graph resolves to one entity rather than several.

Then restructure your highest-value pages so each section answers one question in its first sentence. Not building to a conclusion, leading with it. This is the single biggest change most sites need and it costs nothing but editing.

Then add an llms.txt file. It is a plain-text brief crawlers read instead of parsing your markup, and adoption is early enough that having one is still a differentiator.

How to measure it

No tool reports this reliably yet, so measure it by hand. Write down the twenty prompts a buyer would actually type, run them across ChatGPT, Perplexity, Gemini and Claude, and record which businesses get named. Re-run monthly.

It takes about an hour and it is more honest than any dashboard currently on the market. It also tells you exactly which competitors the assistants consider default answers in your category, which is useful well beyond SEO.

What AEO is not

It is not stuffing your page with question headings. It is not a plugin. And it is not a reason to stop doing SEO, since the retrieval layer that feeds most assistants still leans heavily on conventional search infrastructure.

The honest summary: do the SEO fundamentals properly, then add structure and quotability on top. Anyone selling AEO as a replacement rather than a layer is selling you a rebrand.