Go search for an AEO agency right now. You’ll find a few hundred of them. Same for GEO. Most will tell you search has fundamentally changed, SEO is dead, and you need a new discipline, a new vendor, and a new line item to survive it.
Here’s the part almost none of them say out loud: there is no AEO schema.
No GEO schema either. No agent-ready markup. No special file, no new tag, no proprietary structured data format that unlocks the AI answer box. Everything sold as “AEO schema” is schema.org, the same vocabulary Google, Bing, Yahoo and Yandex agreed on back in 2011. Organization. Article. FAQPage. Product. sameAs.
If your agency or consultant is telling you there is, well, they are full of “bovine output”.
And this isn’t my opinion. Google published a generative AI optimization guide in May, updated in July, that says it directly: “There’s also no special schema.org structured data that you need to add.” The same guide has a mythbusting section naming llms.txt, chunking, and AEO/GEO tactics as things you can skip. Google’s Gary Illyes and Cherry Prommawin told Search Central Live attendees the same thing: GEO and AEO don’t require separate frameworks.
So no, I’m not going to sell you a new discipline. But I’m not going to tell you nothing changed either, because that’s not true. Let me split it honestly.
Four things, two buzzwords
SEO is being found: crawled, indexed, ranked, clicked.
AEO is being the answer, not one of ten links. The box at the top. The voice assistant reading one response aloud.
GEO is being a source: cited by name when ChatGPT, Claude, Perplexity or Gemini synthesizes an answer from a dozen places at once.
Agent readiness is the fourth, and it’s the one that actually keeps me up at night: whether software, not a person, can land on your site and complete a task for a customer.
Four different problems collapsed into two buzzwords and sold as a package. That collapse is your first clue.
What I’ll concede
The hype isn’t that AEO and GEO don’t matter. They matter, and they aren’t coming. They’re here.
The hype is that they’ve been packaged as a new discipline requiring new infrastructure. Open any credible AEO or GEO checklist and go line by line: roughly nine of ten items were SEO best practice a decade ago. Can crawlers reach you. Does the page render server-side or only after JavaScript runs. Is the architecture logical. Do headings describe what’s under them. Is your schema correct and does it match the visible page. Are your business details consistent everywhere they appear. Is the content worth citing at all.
None of that is new. All of it is load-bearing. That’s why the foundation is still SEO. Not nostalgia, engineering. You can’t bolt an addition onto a house with no footing.
The llms.txt example
If anything were going to be the “agent-ready file,” it’s llms.txt, a plain text file at your domain root handing AI systems a curated map of your site. Closest thing the industry has produced to a new standard, discussed for nearly two years.
The data isn’t kind. An SE Ranking study of 300,000 domains put adoption around 10%, one site in ten. When SE Ranking modeled whether having the file correlated with getting cited more often by AI, removing llms.txt from the model improved its prediction accuracy. It was adding noise, not signal. Google has said on the record that Search doesn’t use it, and no major provider has committed to it as a production citation signal.
The fair counterweight, because I’m not interested in dunking: llms.txt is useful for a specific audience. Coding agents (Cursor, Copilot, Windsurf) do read it for documentation. If you run an API product, a developer tool or a large documentation site, ship one. It costs almost nothing, and Google’s own updated wording says maintaining one is fine for other systems that use it. There’s even a split inside Google: Chrome’s Lighthouse audits llms.txt under agentic browsing while Search Central tells you to skip it. Two teams, two answers, both right for what they’re each optimizing.
It isn’t useless. It’s useless for the thing it’s being sold for. That distinction is most of this article.
The 10% that actually is new
I’d be doing you a disservice if I stopped at “it’s all just SEO.” Some things did change:
You’re synthesized, not ranked. A model builds its answer about you from your site, directories, reviews, forums, press, LinkedIn. All at once. Inconsistency across those used to be untidy. Now it degrades what the model believes about you.
You can be recommended without being visited. A model can cite you in an answer that never touches your homepage. Off-site mentions feed visibility directly now. That’s a PR problem wearing an SEO costume.
Extraction beats persuasion. Answer in the first paragraph. Write paragraphs that survive being lifted out of context. Old advice, higher stakes.
Rendering matters more. Googlebot executes JavaScript. Most AI crawlers are far less generous. A site that only assembles content in the browser is a site some models can’t read.
And measurement is the real gap. This is where “add-on” undersells it. Search Console won’t tell you whether ChatGPT recommends you by name. Rankings aren’t citations. Clicks aren’t mentions. Same engine, new instrument panel. And most businesses don’t have one, aren’t looking at one, and have no baseline to compare against.
So: hype or real?
Both, and the distinction is precise: it’s hype as a product category. It’s real as a checklist item.
You don’t need an AEO agency. You need your existing site work done correctly, a handful of additions that take a fraction of the effort the packaging implies, and a way to measure something you currently can’t see.
What I do about it
I built a site audit that scores technical SEO, on-page, content, security hygiene, structured data, AEO and GEO as separate dimensions. Separately on purpose, because I wanted to see whether the AEO and GEO sections would surface findings the SEO sections missed.
They mostly don’t. Run it against a real site and the AEO section and the technical section keep flagging the same fixes: render-blocking content, thin or missing schema, headings that don’t match the page, entity details that disagree with each other across the web. The AEO-specific findings are real, but they sit on top of the foundation. Fix the foundation and most of them resolve themselves.
That’s not a theory I’m arguing. It’s what the reports keep saying.
So when someone quotes you for an AEO engagement, ask one question: what are you doing that isn’t already SEO? A good answer exists. It’s short, it’s specific, and it should sound a lot like that 10%. If they can’t give it to you in plain English, you’re paying a premium for a rename.
Bottom line
A well-built, structurally sound, technically clean site (built for SEO, AEO and GEO together) is the best foundation you can have. Full stop.
But a foundation isn’t a house. The house is content, and content is still king.
Next week, let’s talk about that: what content actually is, why it matters more than any of this, and how to rank for the words that matter to your business. Watch this space.
Want to know where your site actually stands, with a number attached instead of an opinion? That’s a 30-minute conversation: request a call. I’ve shared this on LinkedIn too, arguments welcome there.
Previously: You’re Not Stuck. You’re Just Waiting to Feel Ready.
