AEO vs SEO, Answered With Measurements Instead of Definitions

Search “AEO vs SEO” and the first twenty results agree with each other. SEO is about ranking, AEO is about being in the answer. That sentence is true, it is on every page, and it does not tell you what to do on Monday. This piece answers the same question with three things we measured this week instead: whether an answer exists for your question at all, who holds its citations when it does, and which of your pages wins which kind of question.

The tell that the definition is not landing is on the results page itself. For a query with 1,900 monthly searches and a ten dollar cost per click, the highest organic result is a Reddit thread asking what the difference is, ranked above Semrush, HubSpot and Optimizely, all of whom have published the definition. When a forum question outranks four vendors who answered it, the answer is not the part people are missing.

Disclosure: EchoWi sells AI visibility measurement, so an article arguing you should measure before choosing between two disciplines is an article with an interest. Every figure below comes from our measurement register, with the prompt, the surface, the market, the date and the run count on each row, so all of it can be repeated against us.

The short version

1. Sometimes there is no answer to be in. For eight local buying questions across three markets, Google’s AI Overview did not render once in sixteen runs. Optimising to appear in it is optimising for a box that is not on the page.
2. When there is one, it usually does not cite the seller. Across eleven categories, 84 domains were cited in every run of their question. Twenty-four of them belong to a company that sells the product. In four categories the number is zero.
3. The same site wins one kind of question and loses the other. Asked which ecommerce platform is best, Shopify holds no durable citation, twice. Asked what Shopify costs, www.shopify.com is cited in every run, same surface, same day.
4. So the useful question is not which discipline to fund. It is which of your questions has an answer, whether a company like yours can hold a slot in it, and which of your pages is the one that can.
5. And the answer is different per market. The same five categories measured in Spain and Germany keep their ordering and lose their zeroes, so a benchmark from one country gives you the shape of another and none of its names.

What everyone says, and what it leaves out

The standard formulation is that SEO earns a ranking and AEO earns a mention, that SEO wants a deep page and AEO wants an extractable block, and that you need both. All of that is correct.

What it assumes is that an answer exists. Every version of the advice starts from a page with an AI answer on it and asks how to get into that answer. That assumption is doing an enormous amount of unexamined work, and it is wrong for a whole class of commercially important questions.

It also assumes that the slot is winnable by you, and that assumption fails in about six categories out of ten in the sample we measured.

Measurement one: sometimes the answer is not there

We asked eight versions of “which local business is best” across Spain, the United States and Germany, in five verticals: employment law, dental, accountancy, restaurants and trades. Two runs each, on Google’s AI Overview and AI Mode at the same moment.

AI Overview did not appear once, in sixteen runs. AI Mode answered all sixteen and cited between four and fourteen domains each time.

Four control questions rule out the instrument. They are local in exactly the same way, name a city, were asked in the same languages and markets on the same day, and differ only in asking for information rather than a recommendation. All four returned an AI Overview on both runs, eight for eight. And the live results page for one commercial query carries six local pack blocks and no AI Overview element at all, while the control’s carries one in first position. The local-intent study sets out all eight questions and the four controls.

The practical consequence for a local business is blunt. If your most valuable question is “which is the best one near me”, then on that surface there is no answer to be cited in, and a report telling you that you appear in zero AI Overviews is describing Google’s layout rather than your website. No amount of extractable answer blocks changes it.

That is not an argument against AEO. It is an argument for checking, once, whether the surface runs for your question, before you buy anything aimed at it.

Measurement two: the slot usually is not the seller’s

For eleven software categories in the United States we asked the canonical buying question three times each and kept only the domains cited in every run.

Not Mailchimp in email marketing. Not Shopify in ecommerce platforms. Not NordVPN in VPNs. These are category leaders, absent from the stable core of the answer to their own category’s buying question, while agency blogs, comparison sites and a Madrid marketing school hold slots. The citation-slot study lists every durable domain by name.

The share varies enormously and not along the lines you would guess. Help desk software is 83 per cent vendor-held. Web hosting is zero, in three markets and three languages. Both are business software bought by the same person.

So “do AEO” is not a decision you can make from the outside. In accounting or help desk, your own pages are a live route. In hosting, they are not, and the work that pays is being the subject of the pages that do hold the slots.

Measurement three: it is the question, not the site

The obvious objection to measurement two is that those vendors might simply be weak, or blocked, or badly structured. So we asked one of them a different kind of question, on the same surface, in the same market, on the same day.

Shopify is reachable, indexed and citable. When the question is about Shopify, Shopify is the source. When the question compares its category, Shopify is not in the answer.

This is the sharpest thing we know about the difference between the two disciplines, and it is not a definition. A comparative question selects against the pages of the things being compared. Not because anyone penalises them, but because a page saying “we are the best” cannot be the source for “which is best”, while a page that compares can.

Which means the split inside your own site is not AEO pages versus SEO pages. It is questions about you, where your own pages can win, against questions about your category, where usually they cannot.

What to actually do with this

Three checks, in this order, before either budget line.

One: does the surface run for your question? Ask your single most valuable buying question and look at whether an AI answer appears at all. If it does not, that is settled and cheap, and the surface to work is whichever one does answer.

Two: does anyone like you hold a slot? Ask it three times and keep the domains that appear every time. If none of them sells anything, your own pages are not the asset that wins that question, no matter how well structured they are.

Three: sort your questions by who they are about. Questions about your product, your pricing, your integrations and your alternatives are ones your own pages can win, and that is where the extractable-answer work pays. Questions comparing your category are ones intermediaries win, and the work there is being the thing they write about.

None of that replaces SEO. Every source that holds a durable slot got there by ranking somewhere first. What changes is what you are ranking for and who you expect to be cited.

Limits

One surface family. The citation measurements are Google’s AI Overview, and the local study adds AI Mode. ChatGPT, Gemini and Perplexity retrieve differently and are not covered here.

Small samples. Eleven categories, eight local questions, five verticals, three markets. Enough to show that the share varies and that the zeroes are real; not a census of anything.

Three runs, or two. Enough to separate a stable source from a coincidence, not enough to estimate a rate precisely. A domain at two of three is treated as absent, which undercounts on purpose.

One or two days. Google moves these blocks. Everything here describes 10 and 11 August 2026, and the layout finding in particular is the kind that could change next month.

Nothing is causal. No intervention was run. This is what the answers contained, not evidence that changing a page changes a slot.

Common Questions About AEO and SEO

What is the actual difference between AEO and SEO?

SEO earns a position in a list of links; AEO earns inclusion in a generated answer. That much is agreed everywhere. What the definition omits is that for some questions no generated answer exists, and that for many categories the answer’s citations do not go to the companies being asked about, so whether AEO is available to you is a property of your question and your category rather than of your effort.

Should I stop doing SEO and do AEO instead?

No, and the measurements argue against it more strongly than the slogans do. Every domain we found holding a durable citation slot is a page that ranks somewhere. The change is in which questions you target and whether you expect to be the cited source or the subject of one.

How do I know if AI Overview even answers my question?

Ask it and look. If a report shows zero citations, ask how many answers it saw for that query. Zero out of zero is a fact about the layout of the results page; zero out of twenty is a fact about your site. Most tools report one number for both.

Is AEO the same as GEO?

They are used interchangeably by most people writing about them, and the distinction is not settled enough to be worth arguing about. What matters is measurable either way: does an answer exist, who is cited in it, and can a page like yours be one of them.

My category has no vendor in the stable core. What now?

Then your own pages will not win that question soon, and the leverage is elsewhere: being covered by the comparison sites and communities that do hold the slots, and owning the questions that are about you rather than about your category. Those you can win with your own pages, as the pricing control above shows.

Does this transfer to my market?

Partly. Measuring five categories in Spain and Germany kept the ordering between categories and did not keep the zeroes, and once global platforms are removed, not one market-specific source appeared in more than one country. So the shape transfers and the names do not.

Where this leaves you

The reason the definition of AEO versus SEO keeps getting rewritten is that it is not the part that is hard. Everyone agrees what the two words mean and nobody can tell you, from the definition, whether either is available for the question you care about.

That part is measurable, it takes an afternoon, and it decides where the money goes. Ask your question, count what comes back, and read the answer you get rather than the one everyone published.