AEO
What is answer engine optimization?
A working definition of AEO: what it covers, how it differs from SEO, what it is not, and how to tell whether it is working.
Answer engine optimization (AEO) is the practice of making a company legible and credible enough that AI answer engines include it when generating a response.
Where search engine optimization competes for a position in a list of links, answer engine optimization competes for inclusion in a synthesized answer. The work is different because the mechanism is different: an answer engine does not rank your page, it assembles a response from entities it can identify and evidence it can verify.
This piece is the definition we work from, written plainly enough to be useful to anyone evaluating the category.
The three things an answer engine needs
Every AEO decision reduces to one of these.
1. An entity it can resolve
Before a model can recommend you, it has to know what you are. That means an unambiguous answer to four questions—who you are, what you offer, who it is for, and how you differ—stated consistently everywhere it appears. Contradictions between your homepage, your LinkedIn profile, and a directory listing are not cosmetic. They are competing evidence.
2. Claims it can verify
Models repeat what they can corroborate. Assertions with no supporting source are the easiest thing to leave out of an answer, because including them carries risk. Original data, precise definitions, documented outcomes, and specific capabilities give the system something to stand on.
3. Corroboration it does not have to take from you
Your own website is one node in an evidence graph. Independent coverage, credible directories, customer discussions, and expert commentary decide whether your version of the story holds up. A company that is only ever described by itself is a company a model has no reason to trust.
AEO and SEO
The two overlap more than the framing usually suggests. Technical accessibility, crawlability, and useful content are prerequisites for both. What changes is the unit of value: SEO optimises pages and is measured in rankings and clicks; AEO optimises entities and evidence and is measured in mentions, citations, and accuracy.
One difference is routinely underestimated. Ask an answer engine the same question five times and you may get five different source sets. Anything measured in AEO has to be sampled and reported as a rate, never as a single observation.
For the full comparison, see AEO is not SEO with chatbots.
What AEO is not
It is not a new keyword game. There is no density target, and manufacturing more pages on the same thin claim makes the entity noisier, not clearer.
It is not a guarantee of placement. No one has a ranking dial for ChatGPT or Perplexity. Treat any vendor promising a specific position as a vendor telling you something they cannot know.
It is not a replacement for SEO. Answer engines still crawl the open web and still rely on it as a retrieval corpus. Removing the foundation does not help the layer above it.
It is not prompt manipulation. Text hidden for a model to find is a short-lived trick against systems that are actively adversarial to it—and it damages the corroboration you actually need.
How you know it is working
Rankings do not answer this question. Five measures do:
- Mention rate — how often the brand appears at all for a relevant question.
- Citation share — which sources the answer draws on, and how often those are yours.
- Answer accuracy — whether what is said about your product, pricing, and audience is correct.
- Competitive displacement — who gets recommended in your place, and on what basis.
- Evidence gaps — the questions where no source gives the model enough to work with.
The last one is usually the most actionable finding in a first audit. A gap that no competitor has filled is an opening; an accuracy problem is a fire.
Where to start
Start by asking. Take the five questions a good-fit buyer would ask before they know you exist, run them across the major answer engines, and read what comes back without flinching. Most companies discover two things at once: they are absent from more answers than they expected, and where they are present, something about them is wrong.
That baseline is the whole of the work in miniature. Fix what is inaccurate, clarify what is ambiguous, publish what is missing, and measure again.
Being indexed used to be the finish line. Now it is the entry fee.