You do not have an AEO problem. You have a proof map problem.

The AEO panic has arrived.

You can feel it in the questions clients are starting to ask. Who owns AI visibility? Is GEO different from SEO? Do we need a new tool? Are we showing up in ChatGPT? Why is a competitor getting mentioned and we are not?

Fair questions. Slightly manic framing.

The laziest answer is “it’s just SEO”. That one is too neat. It sounds reassuring, and it quietly dodges the parts that have actually changed.

The other lazy answer is “SEO is dead, buy my AEO system”. That one is worse. That is how you end up paying someone to generate 200 articles nobody asked for, then wondering why a language model still does not trust you.

The truth is less catchy and more useful.

AEO is not a separate magic channel. It is a harder proof test.

If an AI system has to recommend a plumber, a software platform, a law firm, a marketing company, a storage solution or a local builder, it needs evidence. Not vibes. Not a homepage full of adjectives. Evidence.

What do you do? Who is it for? Where do you operate? What makes you different in a way a buyer would actually care about? Who else says you are credible? Can your claims be checked somewhere other than your own sales page? Are your services clear? Are your reviews specific? Are your videos transcribed? Are your products structured? Do third-party sources describe you the same way you describe yourself?

That is the bit most AEO chatter skips.

A tracking dashboard can tell you that you are invisible. Useful diagnostic. It cannot create the proof.

This week’s Marketing School episode had a good line buried inside the usual industry noise. The AEO panic might be good for SEO budgets, but AEO and SEO are still tightly joined. The point was not “do nothing”. It was that clients are getting excited about GEO and AEO while still needing the boring foundations: technical SEO, crawlability, brand evidence, third-party mentions, and content that is actually worth citing.

That matches Google’s own guidance. In May, Google published a new resource for optimising for generative AI in Search. The short version: provide useful, unique, non-commodity content, keep SEO best practice in place, use the right local, shopping, image and video content, and stop believing the weirder AEO myths.

Not sexy. Correct.

Ahrefs gives the more interesting twist. In an updated study of AI Overview citations, they found only about 38% of cited URLs also appeared in the top 10 results for the same query. That is down from their earlier data, where AI Overview citations mirrored the standard results much more closely.

So yes, traditional SEO still matters. And no, ranking for the exact visible query is not the whole game.

Why? Query fan-out.

A person asks one question. The AI system quietly expands it into a set of related sub-queries, comparisons, background checks and implied follow-ups. It might look at the direct search result. It might also check reviews, YouTube, product data, comparison pages, directories, knowledge graphs and other source material.

One prompt becomes a little research project.

This is where a lot of businesses are going to get caught out. They think the job is “write an AEO article”. The job is much bigger than that. The job is to make the business legible across the places machines now check.

I would call that a proof map.

A proof map is the answer to a simple question. If an AI system had to decide whether to recommend us, what evidence would it find, and where would it find it? Not just on your website. Across the web.

Your proof map is made of ordinary things done properly. Service pages with actual detail instead of generic brochure copy. Case studies that say what changed and how. Customer reviews with concrete language, not just star ratings. YouTube videos, transcripts and descriptions that explain your expertise. Partner pages, directory listings and industry mentions. Comparison pages that honestly explain fit and non-fit. FAQs built from real sales questions. Schema and technical foundations that make the content easier to parse. Pricing, process and availability where it makes sense. Original data, field notes, demos and proof assets.

None of that is mystical. It is just work.

Which is exactly why the market will try to avoid it.

There will be tools. Some will be useful. There will be dashboards showing share of voice in AI answers. Also useful. There will be agencies selling AEO packages. Some will be good. A lot will be old SEO theatre with a new acronym stapled to the invoice.

The tell is simple. If the proposed strategy starts and ends with “publish more AI-generated content”, run.

More pages do not automatically create more proof. Usually they create more mush. Machines are already very good at finding mush. They were trained on half the internet. They know the smell.

The better offer is an audit.

Take one commercial topic. One service. One product category. One buyer problem. Then map the source layer around it. What would Google AI Mode, ChatGPT, Perplexity or Gemini need to know before recommending a provider? What hidden sub-queries might they run? Which competitors have stronger third-party evidence? Which videos are getting cited? Which review language keeps appearing? Which pages are technically crawlable but useless? Where does the business make claims that nobody else confirms?

That is the work.

For a marketing team, this changes the job spec. Content is not just content now. It is source material. Video is not just brand awareness. It is a citation surface, especially when the title, description and transcript carry useful language. Reviews are not just reputation. They are customer-language data. Case studies are not just sales collateral. They are evidence assets. Technical SEO is not just rankings. It is machine access to your proof. And brand is not just how you look. It is whether the same idea about you turns up in enough credible places for a machine to believe it.

This is why I think the AEO panic is useful, even though the acronym soup is already irritating.

It forces a better question. Not: can we trick AI search? Better: are we actually documented well enough to be recommended?

For most businesses, the answer is no.

Their site is vague. Their proof is buried. Their reviews are unmanaged. Their videos have no transcripts. Their service pages all sound the same. Their best customer language lives in sales calls and email threads. Their strongest examples are trapped in someone’s head. Their technical foundations are uneven. Their brand shows up differently across directories, social profiles and partner pages.

Then they ask why ChatGPT is not mentioning them.

Because there is not enough there.

That sounds harsh. It is also good news, because it is fixable.

You do not need to chase every AI search tool this week. You do not need to rename your SEO budget because LinkedIn got bored. You do not need a 90-page GEO strategy full of diagrams nobody will execute.

Start with one proof map. Pick the service you most want to be known for. Write down the questions a serious buyer would ask before choosing you. Then check what evidence actually exists across your site, search results, video, reviews, partner mentions, directories and third-party pages. Where the evidence is thin, create it. Where it is unclear, rewrite it. Where it is trapped in calls, extract it. Where it is scattered, structure it. Where it is missing from the wider web, build distribution and relationships instead of pretending your own blog can carry the whole thing.

That is not as glamorous as “dominate AI search”.

Good. The glamorous version is usually where the rubbish starts.

The serious version is better for clients, better for agencies and better for the web. Build proof. Make it readable. Make it checkable. Make it useful to humans first, then machines.

AEO is not replacing SEO. It is exposing whether your SEO, brand, content and proof were real in the first place.

And that is exactly the kind of uncomfortable audit most businesses need.

---

Jason Sibley is the founder of Cleo, a post-agency marketing and AI company. JasonVsTheNoise is where he writes about what is actually happening with AI, marketing, and how businesses should be thinking about both.