AI Is Starting to Make Choices for Your Customers. Is Your Brand Getting Named?

A few days ago, a friend of mine who's in the smart-hardware business invited me to dinner.

Halfway through the meal, he pulled out his phone and showed me his dashboard: search rankings, top three; traffic, steady.

He said, the data all looks good, but lately at trade shows I keep hearing customers say the same thing: I went back and asked AI, and it recommended another company.

He asked me, what's going on here? My search numbers haven't dropped.

I told him, this isn't a traffic problem. The rules have been rewritten.

First, Let's Be Clear: What Is GEO

Doing business used to be like opening a shop on a busy road. Customers walked along the road, and if you made your sign bright and your storefront good, they could find you. That's the job SEO does.

Now, many customers have stopped walking that road.

They've hired a private advisor instead. ChatGPT, Perplexity, Gemini, Claude — they all play this role. Ask it a question, and it doesn't hand you a map or ten links. It hands you one answer.

This shift has minted a new term: GEO, Generative Engine Optimisation.

What it solves can be said in one sentence: optimization used to aim at getting your shop found. Now it aims at getting your brand to be the one AI points to and says, "Just go with this one."

Being cited is being seen.

The Funnel Has Been Flattened

Think about it: people used to buy through a funnel. First search, then compare, and finally place the order.

Start with "search."

A need pops up in a customer's mind. In the past, he would open a search engine, and you would spend budget waiting for him on the results page. Now, he asks AI. AI doesn't give ten blue links, only one answer.

So your goal has changed: it's no longer "rank at the front." It's "become the name AI reads out."

Now look at "compare."

Where did comparing and researching happen before? Your official website, review pages, search results pages. Now, most of that has been swallowed whole by AI: it reads your specs for the customer, reads user reviews, forum discussions, and expert articles, runs the comparison rounds itself, and serves up a single conclusion.

And here's the killer: this research is invisible to your website analytics.

Someone can compare you and your competitors back and forth for three rounds inside AI, and your dashboard won't show a single visit.

Finally, "buy."

"Just buy this one." "Book it for Thursday." "Subscribe for a year." You can say these things straight to AI now, and it will place the order, book the seat, and start the subscription. The customer may never have set foot on your website at all.

So the dashboard wasn't wrong. The ledger was.

I gave him an analogy: you've been counting the number of people walking into the store. Now customers are all at home letting their advisor run the errands, so of course store visits look bad, but business hasn't dropped. Keep staring at store visits, and you'll misjudge whether you live or die.

The influence is real, even when the clicks are invisible. Someone says good things about you on a forum, your specs are written clearly, an expert cites you — all of this is quietly changing AI's answers. It just never becomes a click in your dashboard.

So What Do You Do? Claim the Questions

The second question follows right away: questions are everywhere. Why would AI read out your name?

My answer is, don't be greedy.

There's a concept for this called Category Entry Points, abbreviated CEP. What is a category entry point? To put it plainly, it's the few key questions that pop into a customer's mind right before they take out their money: to buy this kind of thing, what's most suitable? Between these two, which one should I choose?

These customer questions are like the questions on an exam. You don't have to answer every paper. Pick the one or two worth the most marks, and make yourself the standard answer.

An example. A fitness brand doesn't shout "our product is the best." It locks onto one question: for home fitness equipment in small spaces, what's best? Then all of its content is built around this question, until AI can't get around them whenever that question comes up.

Big brands should still show up in the authoritative places. But the real winning move is answering those one or two key questions so completely that you become the only answer.

You Have to Appear Wherever AI Places Its Trust

You might ask, how does AI know you answered well?

AI doesn't lack information. What it lacks is information it can trust. Where it reads, you have to be.

Which places? Official pages, clearly structured FAQs, Wikipedia, review platforms, and communities with real discussion like Reddit. Your product showing up on your own FAQ page, in encyclopedia entries, and in back-and-forth discussions on Reddit is worth far more than showing up in a pile of no-name blogs nobody reads.

Working with people has its rules too. Don't just chase follower counts; find people who truly understand. For a tech brand, one YouTube reviewer who seriously breaks down "X or Y — how do you even choose?" beats ten influencers reading ad copy. Because AI learns from exactly this kind of content: the clearer the explanation and the more concrete the comparison, the easier it is for AI to learn it.

The content itself also has to be easy for machines to read. JSON-LD, FAQ schema — these words sound intimidating, but they come down to one thing: give AI a neatly formatted manual so it can cite with confidence. Specs, pros and cons, use cases — write them clearly. Don't make AI guess.

And bind yourself to reliable platforms. A fitness app that plugs into Apple Health, plugs into Garmin — AI sees those pairings and naturally trusts you a little more. Established partners are still chips you can play at the new table.

The last rule is the plainest: write content that hands over the answer directly. A software company honestly publishes a piece — "Which CRM should a small team choose?" — laying out each option's use cases and pros and cons. What AI cites is always the passage that can be served straight to the customer.

Change the Ledger

The playbook has changed, so the ledger has to change too.

Traffic and clicks can only tell half the story from now on. You have to start keeping new books:

In AI's answers, how many times does it mention you; among relevant questions, how high is the share where you appear; for the one or two key questions you've locked onto, are you there every single time; in forums, reviews, and communities, how deep does the goodwill under your name run.

And the hardest one of all: when AI places the order for someone, does it choose you?

The others are trends. This one is life or death. AI agents are becoming the middleman between customers and the internet: they filter the options, give advice, and close deals directly. If it trusts you, you're always in the room. If it doesn't, you don't even get considered — and there's no second chance.

VML Intelligence has a prediction in The Future 100: 2026: by 2030, 40% of consumers plan to hand comparison shopping over to AI.

Forty percent. By then, buying from you without ever opening your website won't be the exception. It'll be the norm.

Finally, a Checklist

One thing first: don't rush to move your whole budget away. Performance ads on the usual platforms — Google, Meta, Amazon — are still putting food on the table today. Keep investing, and watch the efficiency. The land hasn't gone barren. Rain has simply started falling from the sky — you have to learn to catch it.

Then, put these on the schedule:

Lock onto one or two key questions and make yourself the standard answer to them; rebuild your website and product pages so AI can read them easily, and keep the information fresh; go to Wikipedia, Reddit, review platforms — the places AI trusts — and leave real traces; lock in long-term partnerships with genuine experts and credible platforms; and run an LLM audit on a regular cycle, to see whether the various AIs, when they talk about your category, are talking about you and getting it right.

Back to That Dinner

As we wrapped up that night, my friend said, first thing when I get back, I'll rewrite the FAQ on our website.

I said, right. But what really needs rewriting isn't the FAQ.

It's that old assumption: customers will always walk up to you.

They won't anymore. They ask AI first now, and going forward, even the buying will be done by AI for them.

May your brand be the one AI is willing to name.