The short answer
Most of what's being sold as AEO and GEO is SEO with a new name. Google says so itself: optimising for its AI search features is "still SEO". If your site was set up properly and someone's kept it up, you're not starting from scratch. The genuinely new bit happens off your website: whether other people mention you, and whether AI can explain what you do in one sentence.
Every time I share a client's AI search results on my Instagram stories, someone slides into my DMs with the same question.
"How's AI search different to SEO?"
There's usually a follow-up too. Do I need to be doing it? Have I missed the boat?
When your website went live, someone told you it was optimised for SEO. You nodded like you knew exactly what that meant (we've all done it), paid the invoice and cracked on with running your business.
Then the acronyms turned up. AEO. GEO. AI SEO. You keep seeing people (hi) posting screenshots of clients getting found through ChatGPT, and a little voice goes: I want a bit of that.
Then the wobble sets in. Is my website already set up for this? Or did it quietly go out of date?
So here's the plain-English version. What's changed, what hasn't, and a 10-minute check you can do yourself to see where you stand. Kettle on.
SEO, AEO, GEO and AI SEO: the jargon in 60 seconds
Brace yourself, because this is far less dramatic than LinkedIn would have you believe. They're mostly four names for the same job: making sure that when someone goes looking for what you do, they find you.
| Term | Stands for | What people mean by it | Do you need to care? |
|---|---|---|---|
| SEO | Search engine optimisation | Getting your website found in search results, mainly Google | Yes. Everything else sits on top of it |
| AEO | Answer engine optimisation | Getting your business into the written answer at the top of Google, or into a chatbot's reply | Only as part of your SEO |
| GEO | Generative engine optimisation | Same idea, aimed at AI tools like ChatGPT, Perplexity and Gemini | Only as part of your SEO |
| AI SEO or LLM SEO | Exactly what it sounds like | Whatever the person selling it needs it to mean | Ask them to define it |
Even Google lumps AEO and GEO together. Its own guidance says that, from where Google's sitting, optimising for AI search is optimising for search. Still SEO. Same job, new lanyard.
So why all the new names?
Partly because AI search does work a bit differently (more on that in a sec). And partly because confusion sells. If you don't know what GEO is, you can't tell whether you need it. That's a very easy gap to slide an invoice into. (Hence this post.)
What's actually different about AI search
The big difference is what you get back when you search.
Traditional search hands you a list of links and leaves you to do the legwork. AI search does the reading for you, like a mate who's skimmed every review on the internet, and gives you one answer with a few sources tucked underneath.
| Traditional search | AI search | |
|---|---|---|
| What you get back | A page of blue links (plus ads) | One written answer with a few linked sources |
| What winning looks like | Ranking high enough to get the click | Being mentioned or cited in the answer |
| How people ask | Short phrases, around 4 words | Full questions, around 8 words |
| How many businesses get shown | Up to 10 on page one | Usually a handful |
| How you measure it | Rankings, clicks and enquiries | Mentions, citations and enquiries that are harder to trace |
Word counts from Semrush, which compared one brand's Google keywords with the AI prompts it showed up for.
There's also something sneaky going on behind the scenes. When you ask an AI tool a question, it often runs a load of related searches at the same time. Google calls this query fan-out. (Not a TikTok dance. I checked.)
Google's own example: ask how to fix a lawn full of weeds, and it might also search for the best weedkillers, how to get rid of weeds without chemicals, and how to stop them coming back.
What that means for you: one page chasing one keyword matters less than it used to. What matters is whether your website properly answers the questions your clients actually ask, in the words they actually use.
So do you need to rip it all up and start again with a shiny new AI strategy?
Mostly, no. And you don't have to take my word for it, because Google said so. In writing.
What Google says about AEO and GEO (and the hacks it tells you to ignore)
In May 2026, Google published its own guide to showing up in its AI features. I read it so you don't have to. The short version? Keep doing good SEO.
Google's AI Overviews and AI Mode pull from the same search index and ranking systems as normal Google. So the same stuff gets you in: pages Google can find and index, a site that doesn't fall apart on a phone, and content people actually find useful.
Then it does something that was not on my bingo card for 2026. It lists the AI search "hacks" you can safely ignore.
llms.txt files and other special AI files. Google Search doesn't use them. Having one won't hurt you. It just won't help you on Google either, so don't let anyone charge you a day rate for it.
Chopping your content into tiny chunks for AI. Not needed. Google can find the relevant bit of a normal page. It's not a toddler. It doesn't need its food cutting up.
Rewriting everything in an "AI-friendly" way. Also not needed. Write for people. Revolutionary, I know.
Special schema markup for AI. There isn't any. Normal structured data is still worth having for other reasons.
Chasing fake mentions. Google's spam systems feed into its AI features too. Buying your way into a load of dodgy "top 10" lists is still dodgy.
It also warns you off third-party tools that promise ranking success, because no outside tool has access to Google's internal systems. (Worth remembering next time someone guarantees you page one.)
One honest caveat, because I'm not here to blow smoke up your arse. Google's guide only covers Google: AI Overviews and AI Mode. It doesn't speak for ChatGPT, Perplexity or Claude. But when those tools answer live questions, they search the web for sources too, so the same foundations do most of the heavy lifting there as well.
ChatGPT, Claude, Perplexity and Google don't all pick sources the same way
Google's the only one that's actually published how its AI search works. The others are keeping their cards close, so what we know comes from people testing them and watching what gets cited. Treat this as patterns, not gospel.
| Platform | Where it gets its answers | What seems to help |
|---|---|---|
| Google AI Overviews and AI Mode | Google's own search index and ranking systems (Google confirms this) | Normal SEO, a properly filled-in Google Business Profile, and content built on first-hand experience |
| ChatGPT | What it learned in training, plus live web search for current questions | Being mentioned again and again by different independent sources, rather than one brilliant page. Comparison-style content ("X vs Y", "best X for Y"), because that's how people phrase their prompts |
| Perplexity | Live web search, with the sources linked under every answer | Being one of the handful of pages it cites for your topic. Clear headings, summaries and tables it can pull from. Showing up on review sites, Reddit and industry publications |
| Claude | What it learned in training, plus web search when it's switched on | Measured, specific writing with no hype. Honest trade-offs and "who this isn't for" sections. Real practitioners over generic copywriting |
| Gemini | Google's infrastructure, so normal SEO signals carry more weight here | The same as Google, plus YouTube videos with proper titles and descriptions |
The reassuring bit? The advice overlaps almost completely. Get mentioned in lots of real places, and write like someone who actually does the work. You don't need five separate strategies for five separate robots. You need one, done properly.
"My web designer did the SEO": what that usually means
First things first: you didn't do anything wrong here. Paying someone to build your site and trusting them with the SEO is a perfectly sensible thing to do. You hired an expert. That's the whole point of hiring one.
And there are some very good web designers out there. I've worked with some great ones who truly understand SEO. But there are far more who don't. Most web designers are creatives first. They build beautiful websites, and SEO usually isn't their strong point unless they're tag-teaming with someone whose it is.
So when a designer says they've "done the SEO", what that often means is they made sure there was some text in the metadata. The page titles and descriptions Google shows in search results got filled in. Box ticked. Invoice sent. Everyone's happy.
That's SEO set up. It's useful, and it's also only the start. (A bit like the gym membership you signed up for in January. Technically, you've got one.)
Then the business grows. You add a new services page. A blog. That new offer you dreamt up in the shower. Nobody told you those pages need the same treatment, so they go live without it. And the site that was "done" at launch slowly stops describing the business you actually run.
Meanwhile, AI tools trying to sum you up in a sentence will work with whatever they can find. Including the version of you from three price rises ago.
What your SEO already covers for AI search
Good news first. If your site was built properly, this lot is probably already done, and it's most of what AI search needs:
Google can find and index your pages
Your pages have clear headings and a structure that makes sense
It loads quickly and works on a phone
Your pages answer the questions clients actually ask
Your pages link to each other, so it's obvious what's related to what
If those are in place, you're not starting from zero. You've got a decent foundation. It just needs looking after.
What it probably doesn't cover
Now for the genuinely new bit. Plot twist: most of it happens off your website.
Other people talking about you
AI tools don't only read your website. They read what everyone else says about you too: reviews, directories, podcast show notes, guest posts, roundups, that Reddit thread you didn't know existed. Google's guide says its AI features can pull in what's being said about businesses across blogs, videos and forums.
Your website can say you're brilliant all day long. So can your mum. AI wants to know who else thinks so.
Your details matching everywhere
If your website says one thing, your Google Business Profile says another, and your LinkedIn still describes the job you had before you started the business, AI has to guess which one is true.
Make sure your business name, what you do and who it's for say the same thing everywhere you exist online. It's the least glamorous job on this list, and one of the most useful.
Being able to say what you do in one sentence
This is the one I see most. Most of us struggle with this. Describing ourselves and what we do feels icky, and I see people stumble on it all the time. The biggest thing is being consistent. Say you land on a label like "operations strategist". In some places you use it, but elsewhere you call yourself an online business manager. To AI, and to people, that looks like two different businesses.
If you can't say what you do in one plain sentence, AI can't either. It'll describe you vaguely, or worse, describe someone else.
Most founders describe their work in their own language: the tasks, the method, the clever name for the framework they came up with at 11pm. Their clients describe the problem. AI search is built around people asking about their problems, so if your website doesn't answer in their words, it's hard for anything to match you to them.
Keeping your pages and blogs current
Google's AI features look for relevant, up-to-date pages to build their answers. A services page still proudly listing the offer you retired last year is neither.
Same goes for your blog. A blog post isn't once-and-done. Stats go out of date, your opinions get spicier, and you've got better client examples now than when you wrote it. Updating your best posts every few months keeps them in the running, in Google and in AI answers.
And if you haven't got a blog at all? It's the most direct way to answer the questions your clients are asking AI, in your words, on your own website. (Yes, I would say that. I write blogs for a living. Doesn't make it less true.)
Knowing whether any of it works
Google Search Console now has a generative AI performance report showing how your site does in AI Overviews and AI Mode. Google Analytics can show you visits from ChatGPT, Perplexity and friends.
If nobody's set those up for you, you're guessing. Guessing is a lovely hobby. It's a rubbish strategy.
The 10-minute check: is your site ready for AI search?
Grab a brew. Tick each one off as you go. Nobody's marking this.
Check you're not blocking AI tools by accident. On Squarespace it's Settings, then Crawlers. On WordPress, ask whoever looks after your hosting or SEO plugin. Some people block AI crawlers on purpose, and that can be the right call. What matters is that somebody chose it, rather than it being a default nobody's looked at since launch day.
Read the first two lines of your homepage. Could a stranger tell you what you do and who it's for? Not your method. Not your vibe. What you do.
Ask AI about yourself. Ask ChatGPT, Perplexity and Google's AI Mode: "What does [your business name] do, and who is it for?" Is it right? Is it you? Is it even a real business?
Ask the question your client would ask. Something like "Who's the best [what you do] for [who you help] in [where you are]?" Who gets recommended? Are you in there?
Google your own business name. Do your website, Google Business Profile and social profiles all describe you the same way?
Count your mentions. How many places mention your business that aren't your own website or socials? Reviews, directories, podcasts, guest posts, press. Write the number down. Then don't panic about it.
Check where your visitors come from. In Google Analytics, look at referral traffic from chatgpt.com, perplexity.ai, gemini.google.com and Copilot. In Search Console, open the generative AI performance report.
Check when you last updated your three most important pages and your best blog posts. If it was before your last price rise, it doesn't take Scooby and the gang to work out the answer.
How did you do?
Mostly yes: look at you. Your foundations are solid. Keep them updated and put your energy into getting mentioned elsewhere.
Mostly no: don't shit the bed. That's an SEO job first. Fix the foundations and AI visibility comes along with them.
Somewhere in the middle: start with number 2. If AI can't work out what you do, nothing else on this list matters much.
So, do you need to do anything?
Probably a bit. Probably less than the internet's been telling you.
| If this is you | Do this |
|---|---|
| Your site's set up properly, kept up to date, and Google already sends you enquiries | Keep going. Add the off-site stuff: reviews, directories, guest spots, getting mentioned |
| Your site was built once and hasn't been touched since | That's an SEO job first. Fix the foundations and AI visibility comes with them |
| Someone's pitching you a separate AEO or GEO package | Ask what's different from SEO and how you'll measure it in enquiries |
| You can't say what you do in one sentence | Fix that before you spend a penny on anything else |
| You're fully booked and enquiries keep coming | Carry on. Just don't let the site go stale |
And here's the bit I actually love.
Google's guide says the thing most likely to get you into its AI search, more than anything else in the guide, is content with a unique point of view. First-hand experience. Stuff that couldn't easily be churned out by an AI model. It specifically tells you not to recycle what's already on the internet.
Which is the one thing the shortcut merchants can't sell you. No llms.txt file, paid listicle or "AI visibility package" can fake it.
AI can summarise everything that's already been said. What it can't do is have your opinions, tell your client stories, or notice the thing you've watched happen a hundred times that nobody else writes about.
That's always been good marketing. Turns out it's good SEO too. Who knew? (Me. I knew.)
Questions people ask about SEO and AI search
Is SEO dead now that AI search exists?
No. Google's AI features run on its normal search index and ranking systems, so SEO is what gets you into them. The bit that's dying is SEO done once and left alone.
Is there an SEO equivalent for AI?
That's what AEO and GEO are meant to be. In practice it's mostly the same work as SEO, with more attention on what other people say about you online.
Can ChatGPT do my SEO for me?
It can help with the typing: drafting page descriptions, suggesting questions to answer, tidying up headings. It can't decide what your business stands for, or know what a client said on last week's call. Google's guide warns against content that could easily be produced by an AI model.
Do I need an llms.txt file?
Not for Google. Google Search ignores them. Some other AI tools may read them, so one won't hurt, but it's nowhere near the top of the list.
Is AEO different from GEO?
Not in any way that matters to a small business. Both describe getting your business into AI-written answers.
Should I block AI crawlers on my website?
It depends what you care about more: keeping your content out of AI training, or showing up when people ask AI for recommendations. Different crawler settings affect different things, so check what yours actually does before you flip it.
How do I know if AI search is sending me traffic?
Check Google Analytics for referrals from chatgpt.com, perplexity.ai and similar, and Search Console's generative AI performance report for Google's AI features. Then add one question to your enquiry form: how did you hear about us? Low-tech, and it works.
Did the check and got more no's than yes's?
That's the work we do at Clowt: getting your SEO foundations right, making sure AI can tell what you do and who it's for, and tracking whether any of it actually brings in enquiries. (Because a mention that never turns into a client is just a very expensive compliment.)
Featured posts
