AEO vs SEO: What's Actually Different, and What Isn't

AEO is on every agency's services page this year, and clients are asking about it. That's a fair question. Search is changing, and nobody wants to find out two years from now that they missed the shift.

I don't sell AEO as a standalone service. I plan for it inside the SEO, digital and design work I already do with clients. That's not because I think AI search doesn't matter. It's because, when you look at how it actually works, most of what earns visibility in an AI answer is the same work that earns visibility anywhere else.

In my last post I wrote about experience being the real AI skill. Part of that experience is knowing which new labels describe something genuinely new, and which are a fresh coat of paint on something we've been doing for years. AEO is a bit of both. So here's an honest breakdown of what's different, what isn't, and what to ask before you spend money on it.

What's genuinely different

Some of this really is new, and it's worth being precise about which parts.

The answer replaces the list

A traditional results page gives someone ten blue links and lets them choose. An AI answer reads a set of pages and writes a summary, with a handful of sources attached. Your page is no longer competing for a click so much as competing to be one of the sources the answer is built from.

One question becomes many searches

Both Google and ChatGPT break a question into several related searches behind the scenes. Google calls this query fan-out, and OpenAI says ChatGPT rewrites a question into targeted queries and sends them to search providers. In practice, depth across a whole topic matters more than one page targeting one phrase.

Fewer clicks

This is the part clients feel. Pew Research found Google users clicked a traditional result in 8% of visits when an AI summary appeared, against 15% when it didn't. Being visible and getting a visit are no longer the same thing.

Google isn't the only engine

ChatGPT runs its own search. To be eligible, a site needs to let OpenAI's search crawler, OAI-SearchBot, in, and OpenAI is clear that placement isn't guaranteed. Most sites never think about this, and some block it without realising.

What's said about you elsewhere shows up

AI answers draw on reviews, forums, videos and articles about you, not just your own site. Reputation has always mattered for search. It's now more visible, because it gets summarised straight back to the person asking.

Measurement is harder

Google now has a Generative AI performance report in Search Console, and GA4 can show some AI referral traffic. Beyond that, most tracking is sampled prompts run through third-party tools. That's useful as a direction of travel, but it's a long way from the certainty of rank tracking. I covered the tools I'd use in Your 2026 reporting stack.

What hasn't changed

Here's the part that tends to get lost in the sales pitch. AI answers mostly sit on top of search. Google says its AI Overviews and AI Mode retrieve pages from its own Search index and are rooted in its core ranking systems. In its own guide, published this year, it describes optimising for generative AI search as still SEO.

So the work that earns a place in an AI answer looks very familiar:

• Pages that can be found and read. If a crawler can't reach or render your content, no answer engine can use it. I went into detail on this in AI Crawlers Don't Render JavaScript.
• Clear structure. Sensible headings, readable pages and internal links that make sense to a person.
• Accurate business information. A current Google Business Profile, consistent details across the web and, for ecommerce, a maintained Merchant Center feed. Google names these directly as ways to appear in AI responses.
• Content with something to say. Google's guide puts more weight on original, experience-led content than anything else in it. Recycled summaries of what everyone else has written are exactly what an AI can already produce without you.
• A reputation worth summarising. Real reviews and genuine mentions, earned over time.

Google's guide also lists things you can ignore for Google Search: llms.txt files, special AI markup, chopping content into small chunks, and rewriting pages purely for AI. Other platforms may treat some of these differently, but they're not the shortcut they're often sold as.

None of this is a new discipline. It's the fundamentals, done properly, with a sharper reason to do them. I made a similar case for B2B in AEO for B2B.

Questions worth asking before you buy

Wanting to get ahead of this is sensible, and there are good people doing good work under the AEO label. The label itself just doesn't tell you much. These questions will tell you more:

1. What will actually change on my site, and why? A good answer names specific pages, fixes and content. A vague answer about "AI signals" is a warning sign.
2. How is this different from the SEO we already pay for? If the honest answer is "it mostly isn't", that's fine, as long as you're not paying twice.
3. How will you measure it, and what can't you measure? Anyone confident they can track AI visibility precisely is overselling. Google itself says no third-party tool has access to its internal ranking or AI systems.
4. What are you promising? Guaranteed citations or a guaranteed spot in ChatGPT can't be delivered. Neither Google nor OpenAI guarantees inclusion.
5. Does it rely on tactics the platforms say don't work? Bulk-produced pages for every possible question, or manufactured mentions, are the ones to watch. Google treats both as things that won't help, and mass-produced pages can breach its spam policies.

How I approach it

What I sell is good digital practice: SEO, design, and sites built properly from the ground up. AI visibility is planned into that work with every client, not sold as a separate layer on top.

That means checking that AI crawlers can reach the site, that the business information is right everywhere it appears, that the content says something only that business can say, and that reporting includes AI referrals alongside everything else. It's less exciting than a new acronym. It's also the part that keeps working whichever platform wins.

If you'd like to know where your own site stands, that's the kind of thing my digital consultancy work starts with.

Taking the question to DudaCon

On 28 October I'm hosting the Day 3 sessions at DudaCon '26, Duda's free online conference. The first is The Value of Visibility: How to Sell Local AEO, with Naama Mizrahi from Duda and Ruan Marinho of Develomark and SplashDash.

I'll be bringing the questions from this post with me. Local businesses already understand being found in "near me" searches, so how do agencies explain AI visibility honestly, measure it fairly, and price it without overpromising? I'm genuinely interested in how the panel answers that.

It's followed by Stop Selling Leads. Start Selling Customers., with Jason Swenk and Liron Hilb Horev, and Design That Converts: Live, with Stefán Andrésson and Anath Abenzour. Sessions run 10am to 1pm PT, which is 5pm to 8pm in the UK that week, and registering gets you the recordings if you can't watch live.