GEO, AEO, AIO: The Alphabet Soup Has One Main Ingredient

I asked three of the big AI tools (Gemini, ChatGPT, and Claude) to brief me on one of my clients. The same homework a buyer’s team now does before anyone books a call with you.

Know what came back? Three confident answers. Well written. Nicely formatted.

And wrong.

They described the company’s services in language it had moved away from. Details from a bigger, unrelated company that happens to share its name crept in too. Had I been a buyer, I’d have walked into the first meeting with the wrong picture.

None of the tools were broken. They did what they’re built to do: combine what they learned in training with what they could find on the web, and summarize it. The problem was the material they had to work with. Old pages. Stale profiles. Another company with the same name.

That’s what GEO is really about: making sure that when AI describes you, it’s working from information that’s accurate, current, and actually about you. Forget tricking the machine. Focus on what you’re giving it to read.

First, the alphabet soup

GEO (generative engine optimization). AEO (answer engine optimization). AIO (AI optimization). LLMO, if you want to sound like a smarty-pants at parties. Purists will tell you each one means something slightly different. They’re not wrong. But the terms overlap far more than they differ, and for a company your size, the work is the same. GEO seems to be the one that’s sticking, so I’ll use it here.

In plain terms, it’s about how your company shows up when someone asks an AI tool about you, your competitors, or the problem you solve.

I get asked this a lot: “Should we be doing something about AI search?” Fair question. Let’s start with its older sibling.

SEO (search engine optimization) is making your website easy for search engines to find, understand, and trust. In a lot of smaller B2B companies, it quietly lapsed somewhere between the last website refresh and the last marketer who left.

Why does that matter now? Because AI knows about you in two ways.

Memory. What the AI absorbed in training. Christopher Penn compares it to an old road atlas: redrawn only now and then, because new ones are expensive to make. So what it “remembers” about you may be well out of date.

Search. What the AI looks up when someone asks a question, using search engines and the pages they’ve indexed.

GEO is about both, though the balance is tipping toward search: the newest models are built to know less and look things up more.

The good news? Google’s own guidance says its AI features run on the same core systems as regular search, so optimizing for them is still SEO. Not a new discipline per se. Just an evolution of the old one. But with a higher price for being vague.

What it isn’t (despite what’s in your inbox)

It isn’t a ranking. AI answers are probability, not position. When SparkToro had hundreds of volunteers run a dozen prompts nearly 3,000 times between them, the same list of brands almost never came back twice. Anyone promising to get you “ranked #1 in ChatGPT” is talking out their arse.

It isn’t a bag of tricks. FAQ schema (code that earned special search listings) was a favourite until Google switched those listings off in May 2026. Then came llms.txt (a file to guide AI tools around your site), until Ahrefs found 97% of them didn’t get a single request in a month. And publishing your own “10 best CRM platforms” list with you miraculously listed at number one? In Lily Ray’s analysis of 100 B2B “best of” searches, Google’s AI cited those self-ranked lists hundreds of times, then mostly recommended someone else. You wrote the list. Your competitor won it.

And for most of you, it isn’t mainly about being discovered. In an established category, buyers usually know who the players are before they ask AI anything. (In a new or fast-moving category, AI can still put you on the list.) What counts is what it tells them about you. Which brings us to why it matters.

Where AI actually shows up

I’ve written about how B2B buying really works. The short version: a buying group of around ten people does most of its homework before talking to anyone at your company, makes first contact roughly 60% of the way through their process, and has the shortlist mostly set on day one.

So where does AI fit? Kerry Cunningham, head of research at 6sense, shared the numbers on the Shift & Thrive podcast: 94% of buyers use AI somewhere in the buying process. Just not where most marketers assume.

AI use is around 44% at the start, peaks at 58% in the middle, and drops sharply near the end. The middle is where AI does the heavy lifting: summarizing vendor material, sifting through reviews, and building comparison tables. By the end, they’ve switched to demos and talking to people.

Kerry points out that buyers don’t just ask for the comparison table. They tell the AI what they want the table to say. Read that again.

That’s your champion (the person inside the buyer’s company who wants you to win) building the case for the rooms you’ll never be in. The chat with the CFO. The IT review. The call with Gerald in procurement.

So when your champion asks AI to make the case for you, what does it have to work with?

Your website. Your reviews. Your LinkedIn page. That 2021 PDF nobody remembers publishing. Kerry’s early look at North American B2B vendor websites (outside regulated industries) found about 75% of the content is pure self-promotion, which AI largely shrugs off. And anything gated behind a form? To the AI tools doing that research, it’s essentially invisible.

Your website isn’t a brochure anymore. It’s source material.

(To keep this in proportion: people still ask people. Referrals from trusted folks carry the most weight. AI doesn’t replace that. Buyers use it to check what they’ve heard.)

What to do, in the right order

Biggest impact for the least effort, first.

Start here (hours, not months)

1. Ask AI about yourself. Open ChatGPT, Gemini, Copilot, and Claude in logged-out or temporary chats (your own history skews what you see). Ask each to brief you on your company, then to compare you with your two closest competitors for a specific type of customer. Flag everything that’s wrong, outdated, missing, or about another company. That’s your to-do list.

2. Fix the source, not the AI. You can’t correct the AI. You correct what it reads. Start with your website, then LinkedIn and whichever profiles turned up in step one (such as Crunchbase, G2, Clutch, or partner directories), so they all describe you the same way. Update or remove stale pages, retired service names, and old PDFs, and keep at it as things change. Corrections take weeks to months to show up, so start now.

3. Say it plainly on your homepage. Who you are, what you do, who you’re for, and how you’re different. In the first few lines. In the words your buyers use.

AI links companies to the words used to describe them. Penn’s analogy: every category is a city on a map. “Payroll software” is one city. “Managed IT services” is another. Describe yourself the way your buyers do and you’re downtown, right where they’re headed. Use your own invented language and you’re out in the boonies, miles from where anyone’s looking.

It’s also why I get nervous when a startup tells me it’s creating a new category. Everyone’s favourite example is HubSpot. Penn puts it at about ten years and $100 million to make “inbound marketing” a thing. I’m taking a wild guess that you don’t have the time or the money.

Think of the grocery store. Chili ketchup is still ketchup, so you know what you’re getting: thick, tomatoey, made for fries. The chilies are what make it different. Put it in the hot sauce aisle and it gets judged against thin, fiery sauces (where it’s a pretty weak hot sauce). On the ketchup shelf, it’s the interesting one.

And nobody walks in looking for the “tomato-forward heat experience” aisle. Not at the grocery store, and not in ChatGPT.

Pick the category your buyers already use, earn your place in it, then explain how you’re different.

Honest caveat: all of this is easy if you’re already clear on who you’re for, what makes you different, and which category you belong in. Ask three people on your team how you’re meaningfully different. If you get three different answers, that’s the real work. (And the real problem.)

4. Ungate what buyers need to compare you. Your capabilities, your approach, and at least a price range or what drives cost. If it’s behind a form or requires a call, AI can’t read it. Most humans won’t bother either. (When did you last fill in a form for a white paper? Right.)

5. Check the doors are open. Set up Google Search Console and Bing Webmaster Tools. Both are free, and both now show when you appear in AI answers. Don’t skip Bing. It powers Copilot, the AI built into Microsoft 365 (so likely on your enterprise buyers’ laptops), and it shows you the searches Copilot ran to find you. Then hand whoever runs your website one technical check: make sure nothing is blocking AI bots by accident.Search bots (like OAI-SearchBot and Claude-SearchBot) are how AI looks you up. Training bots (like GPTBot and ClaudeBot) shape what it remembers. Unless you have a specific reason to keep your content out of AI training, let them in. You want to be found.

Worth the extra effort

Get other people saying it. Podcasts, partner pages, reviews, a spot in other people’s roundups, and media coverage from good old-fashioned PR. Ahrefs found brand mentions track visibility in Google’s AI answers far more closely than backlinks. What others say about you beats what you say about yourself.

One page per service or product. Who it’s for, the outcomes, how it works, what drives cost, and the questions buyers actually ask, answered in plain view. Most of what’s good for AI is just good for buyers.

Case studies with real numbers and dates. Specifics beat superlatives.

The one ingredient

The acronyms will keep multiplying. The tools will keep changing. The hacks will keep having the shelf life of a bag of milk. (Yep, bagged milk is a thing.)

What won’t change: AI deals in probabilities. It repeats what the web says about you most often and most consistently. You can’t control the answer, but you can change the odds. Give it something clear, current, and true to repeat.

That’s the main ingredient in the alphabet soup. Clarity. It always has been.

So if you do one thing this week, make it step one. Ask the questions your best-fit prospect would ask. Then look hard at that comparison table: would they see a reason to pick you?