AEO vs SEO: What Actually Changes, and What Does Not

Which is why "switch from SEO to AEO" is bad advice. Very little gets thrown away. What changes is where the last mile of effort goes.

What actually changes?

Six things, and none of them is the tooling. You stop competing for a slot and start competing to be named. The judge stops being an algorithm ordering pages and becomes a model summarising what it has read about you. And nothing tells you when you lose.

That last row is why this gets noticed late. A ranking drop shows up on a dashboard. Not being named shows up nowhere.

Does SEO still count?

Yes, and how much depends entirely on the engine. Inside Google's AI Overviews the link between ranking and citation holds tightly. Inside ChatGPT it mostly breaks. Treating AI search as one thing hides which half of your budget is still working.

In AI Overviews, a lot. Whitehat SEO found 97% of quoted pages were already ranking in Google's top 20. Off the first two pages, you are rarely quoted at all.

In ChatGPT, almost none. Chatoptic tested 15 brands and found rank barely predicted anything. A page at number 40 can get quoted while your number one page is ignored.

Meanwhile the click is going. Seer Interactive tracked 3,119 searches: click-through rate fell from 1.76% to 0.61% once an AI Overview appeared. On 1,000 searches, 18 visits down to 6.

So SEO has moved from finish line to entry requirement. You still need it to be eligible. It no longer decides the outcome on its own.

What AEO adds on top

Five things earn citations, and only four sit on your website: a short answer under each heading, that answer near the top, structured data, and a fresh date. The fifth is what other sites say about you. It is slow, and it decides whether a model names you at all.

AI Overviews is the AI-written answer Google puts above the blue links. Structured data is a small block of code that states plain facts about your page in a format machines read directly, instead of guessing from your sentences.

JSON-LD is the format Google asks for. Organization says who you are; FAQPage marks up questions so they can be quoted on their own.

What a machine can and cannot use. Most sites write this: "We are a leading provider of innovative solutions for the modern enterprise."

A model cannot do anything with that. There is no category, no buyer, no proof. So it skips you and quotes someone else.

Now this: "We build payroll software for construction firms with 50 to 500 staff. We handle multi-site timesheets, union rates and CIS deductions in the UK."

Same company. One sentence a model can lift, attribute and be judged on. That is the whole trick.

The first four are a writing and markup change on pages you already have. The fifth is a different job, and it is the one most plans leave out.

What to check on your own site

Ask the AI tools your buyers use, logged out of your own account, and write down who gets named. Then check your best pages for a short answer up top, schema, and a date. Where the gaps are tells you which half of the work you need.

1. Write down 20 to 30 questions a buyer would really type. Not "best payroll software", but "payroll software that handles CIS deductions for a 200-person builder". Questions, not keywords.
2. Ask them logged out, in a private window. Logged in, your own chat history gives you a false yes.
3. Open your five best sales pages. Does each part answer its own heading in the first fifty words?
4. Check the same pages for schema and a visible date. Paste each URL into Google's Rich Results Test, a free checker, and it answers in 30 seconds.
5. Search your company name on review sites and directories. Note where you have no profile at all.

Steps three and four you can usually fix in-house. Step five you usually cannot.

How long does it take?

Page fixes can show inside weeks, because engines refetch pages when a question is asked rather than waiting for a crawl. Anything resting on other sites runs on a two to six month cycle. Set a baseline before you change anything, or you will not be able to prove either way.

What speeds it up: pages that already rank, a category with few competitors, and an existing presence on review sites. What slows it down: a new domain, a crowded category, and no third-party profiles to build on.

Does this need its own budget?

For the page-level half, usually not. It is a change in how existing pages are written and marked up, and the team you have can do it. The off-site half is where a separate line starts to make sense, because it needs relationships rather than edits.

The more useful split is not SEO budget versus AEO budget. It is work you can do yourself versus work that depends on other people saying things about you. The first is cheap and fast. The second is neither, and it is what decides the result.

So the honest sequence: fix your own pages first, measure, and only then decide whether the outside half is worth paying someone for.

Check where you stand

Xtrusio runs that check for you: a fixed set of buyer questions across ChatGPT, Gemini, Claude and Google AI Overviews, showing where you are named, where a competitor is named instead, and which sources are shaping the answer.