AI SEO — the two meanings, and what to do about each

The two meanings — which one are you after?

When someone types "AI SEO" into Google, they may mean two entirely different things. It is worth settling that up front, because the work, the tools and the outcome are not the same.

They are not mutually exclusive — most businesses end up needing both. But they solve different problems, and it helps to know which one is yours right now.

1. AI as a tool for your SEO work

This is the more common meaning: you use ChatGPT, Claude or an SEO tool with AI built in to get the work done faster. It works well for some tasks and badly for others, and the difference is worth knowing before you hand your content to a model.

Where AI genuinely saves time:

• Keyword and topic ideas. Getting 50 angles on a topic in a minute really is faster than doing it yourself.
• Drafts and structure. An outline or a first draft you then edit removes the worst of the starting friction.
• Meta titles and descriptions at volume. Mechanical work a model handles well.
• Untangling technical mess. Having an error in robots.txt or a piece of schema markup explained beats searching for it.
• Summarising research. Boiling ten competitor pages down to an overview.

Where it goes wrong — and what it cost us:

We write with AI ourselves. In July 2026 we went through our own marketing content and verified every statistical claim against its primary source. Of 26 claims, 4 held up. The rest were wrong — not invented from scratch, but wrong in ways that are hard to spot:

• Our own module name became an external source. An internal name for our own analysis engine appeared in the text as though it were a research project, and travelled on into eight separate claims.
• Real numbers, wrong technique. Genuine figures from a genuine research paper had been transplanted onto methods the paper never tested. The numbers checked out; the connection did not exist.
• A placeholder became a customer case. The example name from our own form fields was told as a real customer with real results.

None of it looked wrong. That is the whole problem: a language model produces plausible text, and plausible text is precisely what you fail to catch on a read-through. The only method that worked for us was demanding four things of every single number: who measured it, when, on what population, and what is the URL. If one is missing, the number leaves the text.

So this meaning of AI SEO is not a shortcut. AI writes faster than you do. It cannot check itself, and one wrong number in your marketing costs more trust than the saved quarter of an hour was worth.

From AI assistant to automatic SEO

There is a third answer sitting between the two meanings, and it is where most smaller businesses end up: you want neither to learn prompting nor to read up on a new discipline — you want the work done. That is what automatic SEO means in practice.

The difference between an AI assistant and automatic SEO is not how clever the model is. It is whether there is a routine around it. A chatbot answers when you ask, and has forgotten it tomorrow. A routine measures the same thing every week, proposes one concrete change at a time, and lets you approve it before anything is published. That is exactly the control missing from the section above — and it is what makes AI useful for SEO rather than risky.

Three things have to run on their own before it deserves the word automatic:

• Measurement that repeats identically. Same site, same dimensions, same cadence. Otherwise you cannot see whether something worked — only that something is different.
• One prioritised next step. Not a list of 40 errors, but the single fix that matters most right now.
• Execution with an approval in front of it. The change is written to your site through an integration — but only once you have said yes.

That is how the Geoa platform is built. It scores your site on 15 technical dimensions, tracks both your Google rankings (via Search Console and weekly rank tracking) and your mentions in ChatGPT, Claude, Gemini and Perplexity, puts the findings in a prioritised queue, and writes the approved fixes — llms.txt and structured data — straight to your site through the WordPress, Shopify, Cloudflare and other integrations. Which fixes can be written automatically depends on the platform. Every week you get a report on what changed and what moved.

And just as importantly, what it does not do. It does not publish text without your approval. It does not answer your reviews on its own — it drafts a reply that you send. And it does not promise that ChatGPT will mention you: that is a third-party model, and no vendor can guarantee an outcome inside it. We measure it. We do not control it.

2. Visible inside the AI answers

The second meaning is the new one, the part classic SEO does not cover: being part of the answer itself when a customer asks an AI instead of searching. Where SEO is about ranking in a list of links, this is about being the business or the source the AI brings up.

The formal disciplines are GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization). GEO is the academically established term for optimising towards generative engines — formalised in research from Princeton and Georgia Tech (arXiv:2311.09735, KDD '24) — while AEO is used more broadly for all answer surfaces. You will also meet the same field under the names AIO (AI Optimization) and LLM SEO.

The signals that matter most if you want to be citable:

• Answer one question directly and early. Put the short, concrete answer in the first couple of sentences — that is the block AI systems lift out.
• Structure with questions as headings. H2s phrased as the questions customers actually ask.
• Use lists, tables and short paragraphs. Structured content is extracted more easily than walls of text.
• Add structured data. FAQPage, HowTo and Organization make the answer machine-readable.
• Make sure AI can read the page at all. If your page needs JavaScript to show its content, AI crawlers see an empty shell — check it with the AI Readability Check.
• Publish an llms.txt. The file gives AI a curated overview of your business — create one free with the llms.txt generator.

What our own measurements show

Most of what gets written about AI SEO is advice with no measurement behind it. We have measured our own claims — including the ones that did not suit us.

Being cited and being recommended are two different outcomes. We put 576 questions about 465 businesses in Denmark to four AI engines (Geoa, 26 July 2026). The businesses AI cited as a source had a technical score of 77.9 against 72.7 for the rest — a strongly significant difference (p=0.000). But the businesses AI recommended scored 74.9 against 74.9. No difference at all (p=0.988).

In practice that means technical work decides whether your page gets used as a source. It does not decide whether your business gets recommended — there, sector, reputation and demand weigh heavier. Anyone promising you more customers through technical AI optimisation alone is promising more than our data can carry.

Which sources AI actually drew on. AI cited 4,034 sources across 494 domains. 61% were the businesses' own websites. Only six domains recurred across all three sectors, and every one of them was a directory or a register — national business directories, a company register, a review platform and LinkedIn. Your own site plus a handful of directory profiles is what gets read.

How little it takes to be invisible. In the same scan, 22 of the 465 sites had almost no readable text once JavaScript was not executed, and another 38 had under 150 words. Their content exists — but AI crawlers do not run JavaScript by default, so for them it does not. It is the cheapest error to fix and the most expensive to overlook.

AI SEO, GEO and AEO — how they connect

When "AI SEO" is used in the second sense, it is the umbrella; GEO and AEO are the two disciplines under it. They optimise towards different surfaces, but the work overlaps heavily:

And no — it does not replace your SEO. The foundation is shared: a technically sound, crawlable site with clear structure. But the goal shifts from ranking in a list to being part of the answer, and that rewards citability over raw domain authority. It is why a small, sharp business can win AI visibility against larger competitors — most have not started yet.

Getting started

1. Measure where you stand. Run the free GEO test (60 seconds, no login) and see which signals are keeping you out of AI answers.
2. Make sure AI can read your site. Check robots.txt and JavaScript dependency with the AI Readability Check. It is the error that hits hardest and costs least to fix.
3. Make the content answer-friendly. Answer concrete questions directly, with question-shaped headings and short paragraphs.
4. Add structured data and an llms.txt. Use the llms.txt generator and add FAQPage schema.
5. Tidy up your directory profiles. Directories and registers were among the few domains AI drew on again and again. A profile is an afternoon's work.
6. Measure again, and repeat. AI visibility is built over time — track your GEO Score and fix what drags it down.

If you would rather have it done automatically, the Geoa platform measures and fixes these signals for you through the WordPress or Shopify integration — with a 14-day free trial, no card required. See the full picture in the complete GEO guide.

Frequently asked questions about AI SEO

What does AI SEO actually mean?

It is used for two different things. One is using AI as a tool for your SEO work — letting ChatGPT find keywords, draft copy and explain technical errors. The other is becoming visible inside AI answers, when a customer asks ChatGPT or Perplexity instead of searching Google. They require very different work, and most guides conflate them.

Can I use ChatGPT to write my SEO content?

Yes, but not unchecked. We write with AI ourselves, and when we verified every statistical claim in our own marketing content against its primary source in July 2026, only 4 out of 26 held up. The errors were not invented from scratch — they were plausible, and plausible text is exactly what you fail to catch on a read-through. Demand four things of every number: who measured it, when, on what population, and what is the URL.

Is AI SEO the same as GEO?

Almost. AI SEO is the broad, everyday label; GEO (Generative Engine Optimization) is the precise term for being cited in generative AI answers. When people say "AI SEO" in that sense, they usually mean GEO plus AEO.

What is automatic SEO — is it just AI SEO under a new name?

Automatic SEO is not a tool you ask — it is a routine that runs. The difference between asking ChatGPT for SEO advice and having automatic SEO is not how sharp the model is, but whether the measurement repeats identically every week, whether you get one prioritised next step instead of a list of errors, and whether the fix is written to your site through an integration after your approval. That is the control freehand AI lacks.

Does AI SEO replace my regular SEO?

No. AI SEO builds on top of SEO. The foundation — a technically sound, crawlable site — is shared. But the goal shifts from ranking in a list to being part of the AI answer itself, and it rewards citability over raw domain authority.

Can I measure my AI SEO for free?

Yes. Run the free GEO test (60 seconds, no login). It scores the technical signals that decide whether AI can find, understand and cite you.

Which AI search engines does AI SEO apply to?

The main ones are ChatGPT, Perplexity, Claude and Gemini, plus Google's AI Overviews. The work is largely the same across them: make your content easy to understand, extract and cite.

Can a small business win at AI SEO?

Yes — often more easily than a large one. AI answers reward precise, well-structured content over sheer domain size, and most small and medium-sized businesses have not started yet. You can get far with the free GEO tools.

Is AIO (AI Optimization) the same as AI SEO?

In practice, yes — AIO (AI Optimization) is another name for making your business visible in AI search, like AI SEO, LLM SEO and the precise term GEO. Note, though: "AI optimization" is also used broadly for optimising the AI models themselves — so context decides the meaning.