Answer Engine Optimisation is just SEO with better manners

A new acronym is a commercial event before it is a technical one. Answer Engine Optimisation arrived with a full complement of vendors, certifications, scoring dashboards and conference tracks, and it arrived remarkably fast — faster than anyone could plausibly have accumulated the evidence base to justify calling it a separate discipline.

We work in this area and we sell services in it, so treat what follows accordingly. But we would rather be trusted than fashionable, and the honest position is this: the overwhelming majority of AEO is search engine optimisation, competently executed, on a surface where the output happens to be a paragraph instead of a list of links.

The judgement has not changed

Consider what a search engine has always had to do. Given a question, it must work out which sources are relevant, which are credible, which are current, and which can be presented to a user without embarrassing the company presenting them.

Now consider what an assistant does. Given a question, it must work out which sources are relevant, which are credible, which are current, and which can be summarised without embarrassing the company doing the summarising.

The output format changed. The underlying judgement did not.

This is not a rhetorical flourish; it has direct practical consequences. It means the levers are largely the same levers. Be findable. Be clear. Be corroborated. Be specific. Be current. Do not make claims you cannot support. These were the correct answers before assistants existed and they remain the correct answers now.

It also means the failure modes are the same. A site that cannot be crawled cannot be cited. A page that buries its answer will not be extracted from. A brand described three different ways across four properties will be understood as a blur. None of this required a new acronym to become true.

What genuinely is different

Dismissing the whole category would be as lazy as inflating it. Four things really have changed, and they matter.

The query is longer and more constrained. People describe situations to assistants rather than typing keywords at them. That shifts which content gets surfaced — away from head terms and towards pages that address specific circumstances. If your content strategy is a keyword list, it is aimed at a behaviour that is declining.

There is no results page to compete on. In search you could be fourth and still get clicked. In an assistant's answer there are typically two or three sources, and being fourth is being absent. The distribution of attention is far more concentrated, which raises the value of being in the set and reduces the value of incremental improvement once you are not.

You are being restated rather than displayed. The user does not see your page; they see the model's summary of it. That makes ambiguity actively dangerous. A page that a human reader would interpret charitably can be summarised in a way you would never have approved, and you have no editorial control over the result.

The measurement model is broken. Rank tracking assumes a stable, checkable position. Assistant citation is probabilistic and varies between runs of the same prompt. You cannot check it once; you have to sample it repeatedly and reason about frequency. Most of the genuinely new work in this field is measurement work, not optimisation work.

The better manners part

The one stylistic shift we would actually insist on is in how content is written, and it is the reason for this piece's title.

A decade of SEO content production taught the industry a set of habits that assistants punish. The long preamble before the answer. The hedge that avoids committing to a recommendation. The paragraph of keyword-bearing filler. The claim of being the leading provider of anything, stated without evidence. These habits survived because they were, at various points, rewarded — or at least not penalised — by search.

A model summarising your page is a fast, literal, slightly sceptical reader. It rewards content that answers the question, states the caveats, and stops.

Content written with better manners — direct, specific, honest about limitations, structured so a reader can find the answer without hunting — performs better. It also happens to be better content by any reasonable human standard, which is a pleasant coincidence and probably not a coincidence at all.

How to tell a real practitioner from a rebrand

If you are evaluating help in this area, a few questions separate substance from packaging quickly.

• Ask how they measure citation. If the answer does not involve repeated sampling across time and across assistants, they are checking once and calling variance a result.
• Ask what they would fix first. If the answer is not something unglamorous about crawlability, rendering or indexation, they have skipped the constraint that determines everything else.
• Ask what proportion of the work is off-site. If the answer is none, they have missed that assistants disproportionately cite independent sources.
• Ask what they cannot influence. A practitioner who claims control over model behaviour, provider licensing deals or update cycles is describing something that does not exist.
• Ask for the mechanism, not the score. A proprietary visibility score with an unexplained methodology is a dashboard, not a diagnosis.

Why this matters commercially

The reason to be sceptical here is not intellectual tidiness. It is that the framing determines the budget line, and the budget line determines who does the work.

If AEO is a new discipline, it needs new tools, a new vendor and incremental budget — usually taken from the search budget, on the argument that search is the declining channel. If it is search fundamentals applied to a new surface, the answer is to fund the fundamentals properly, which most organisations have quietly been under-doing for years.

We have yet to see a company with a genuinely well-executed technical foundation, clean structured data, consistent entity information and strong independent coverage that had an AI visibility problem requiring exotic intervention. We have seen a great many companies with none of those things buying an AEO product to solve a problem those things would have solved.

Do the boring parts. The new surface rewards them more, not less, than the old one did.