There's a new acronym in your inbox.

GEO. Generative engine optimization.

Or AEO. Or LLMO. Depends which agency is pitching.

The pitch is usually the same. Search is dying. AI answers are the future. You need a new strategy, a new team and, conveniently, a new retainer.

Half of that is true.

The shift is real.

Pew Research found that when Google showed an AI summary, users clicked a traditional result in about 8% of visits, compared with 15% when no AI summary appeared. Fewer clicks. More answers delivered without a visit.

Buyer behaviour is changing too. People ask ChatGPT, Perplexity, Gemini and Claude for shortlists:

"Best payroll software for a 40-person company in Canada."

"Best CRM for a small manufacturing company."

"Which cyber insurance provider should a startup consider?"

If you're not in the answer, you may never make the consideration set.

So yes, you should care.

Here's the part the pitch decks tend to skip.

AI systems don't form brand recommendations in a vacuum. They learn from the information environment around them, and when they search the live web, they often rely on the same sources search engines have rewarded for years: authoritative publications, review sites, comparison pages, forums, product pages and structured information.

Which means the brands showing up in AI answers today are often the brands that did the boring authority work years ago. Earned coverage in publications people actually read. Reviews on sites buyers actually check. Clear pages that answer specific questions. A name that appears repeatedly, in the same context, across independent sources.

That's not a brand-new discipline. That's good SEO, plus PR, plus reputation, plus clear positioning. The stuff many companies stopped investing in when budgets got tight.

GEO isn't SEO's replacement.

It's SEO's revenge.

There are real changes, though, and they're worth understanding.

Mentions matter more. A traditional SEO mindset focused heavily on links. AI systems care about the broader context in which a brand appears: where you're mentioned, what claims are consistently associated with you and who's making them. A credible industry roundup, analyst piece, Reddit discussion or customer review can shape how an AI system understands your brand even when the SEO value of the link itself is modest.

Specificity matters more. Models need concrete information to work with. Who are you for, what does it cost, where do you operate, what integrations do you support and how do you compare with the obvious alternatives? Vague brand copy gives a model very little to repeat. The academic work that popularized the term GEO found that adding statistics, quotations and cited sources could improve visibility in generative answers.

That shouldn't be surprising.

Useful facts are easier to reuse than marketing adjectives.

Measurement gets messier. There's no exact equivalent of Search Console for ChatGPT. Answers vary by prompt, model, user, location, browsing mode and time. You can measure patterns, track whether your brand appears for important prompts and monitor which sources get cited. But anyone selling you a single precise "AI visibility score" is selling you a model of reality, not reality itself.

Treat it accordingly.

Everything underneath.

Be the obvious answer to a specific question.

Earn third-party proof.

Make your pages easy to understand for humans and machines.

Show up consistently in the places your buyers already trust.

If you were already doing that, you're in decent shape. If your old SEO strategy was buying links and publishing 2,000-word articles stuffed with keywords nobody cared about, AI didn't create your problem.

It exposed it.

Before you hire a GEO agency, run this yourself. It takes an hour.

Write down the ten questions your best customers asked before they bought. Ask them in ChatGPT, Perplexity, Gemini and Google. Then note who gets recommended, who gets cited, and which publications, review sites and communities keep shaping the answer.

You'll usually find the same handful of sources appearing again and again.

That's the opportunity.

Not "How do I manipulate an LLM?"

The better question is:

How do I become the obvious answer in the places the model already trusts?

Get mentioned in those sources. Give them specific, defensible claims to repeat. Make your own site painfully clear. Build enough third-party proof that the same story about your company shows up everywhere.

That's GEO.

No new acronym required.