Your next customer is asking AI who to hire. Are you the name it gives?

What actually changed — and why your traffic chart is lying to you

For most of the last two decades, the path was simple: someone had a problem, they typed it into Google, they got ten blue links, and they clicked. Your job was to be one of those links. Now look at what a buyer actually does. They open ChatGPT, Gemini, Perplexity or Google's own AI answer at the top of the page, and instead of ten links they get one synthesised reply — a paragraph that says 'here are the three or four options that fit you, and here's why'. A large and growing share of real research now happens this way, especially for the considered purchases I care about: a treatment, a property, an agency, a school, a consultant. The buyer never scrolls to the old list. The decision narrows before a single website is opened.

Here's the part that catches founders off guard. Your analytics dashboard doesn't show any of this. It can only count the people who click through to you — so when AI answers the question without sending a click, that influence is simply invisible to your reports. You'll see 'organic traffic flat or down' and assume the market went quiet, when what really happened is that the conversation moved to a room you weren't standing in. The buyers are still there. They're just being introduced to a shortlist by an assistant now, and if you're not in the introduction, you don't get to make your case. That's not a traffic problem you can solve by spending more — it's a presence problem.

Being recommended is not the same as ranking

The instinct is to treat this like SEO — 'tell me the keyword and I'll rank for it'. But an AI engine doesn't rank a list; it reads the open web, decides what's true about you and your category, and then writes a recommendation in its own words. So the goal shifts from 'be link number three' to 'be the business the model is confident enough to name'. And confidence, for these systems, is built the same way trust is built between people: by hearing the same thing about you from many independent places.

That means the levers are different. The model is more likely to recommend you when there's a clear, consistent story about who you are, where you operate and what you're genuinely good at — on your own site, yes, but more importantly across the wider web: your Google Business Profile, real reviews, directories, mentions on other people's sites, comparison pages, even forum and social answers. It leans toward businesses that have published clear, specific, genuinely useful content that answers the real questions buyers ask, with actual numbers and detail rather than fluff. And it quietly penalises the opposite — a thin website, an inconsistent name and address, no third-party mention, vague 'we're the best' copy with nothing concrete behind it. None of that is exotic. It's the same reputation you'd want in the real world; the AI is just reading it faster than any human could.

What I'd actually do this quarter

Start by finding out what the machines already say about you, because most owners have never checked. Ask ChatGPT, Gemini and Perplexity the questions your buyers ask — 'best [your service] in [your city]', 'who should I hire for [problem]', '[competitor] vs alternatives' — and write down exactly what comes back. You'll usually learn one of three uncomfortable things: you're not mentioned, you're mentioned with wrong details, or your competitor is being described in words you wish were yours. That gap is your brief.

Then close it on the open web, not just on your own site. Get your Google Business Profile complete and consistent, get real reviews flowing steadily, and make sure your name, location and what you do are identical everywhere they appear — these systems trust consistency and get confused by contradictions. Publish content that plainly answers the questions buyers actually ask, with specifics: real prices or ranges, real timelines, real before-and-after numbers, honest comparisons. Put your proof in writing — case studies with figures, not adjectives. Where it fits, add the basic structured data that tells a machine in plain terms who you are and what you offer. None of this is a trick to game the model; it's giving the model accurate, quotable reasons to put you in the answer. The businesses doing this now, while it's still early, are going to be the 'default recommendation' in their category for years — the same way the people who took Google seriously in 2010 still benefit today.

Why this is a performance problem, not an SEO one

I'm a performance marketer at heart, so let me put it in those terms. A buyer who has just been handed your name by an AI assistant — as a considered, reasoned recommendation — is not a cold click. They arrive warmer and further down the funnel than almost any other traffic you can buy, because something they already trust has effectively vouched for you. That makes this visibility some of the highest-intent demand there is. Ignoring it isn't a tidy SEO oversight; it's leaving your best-qualified future customers to be introduced to your competitors instead.

Two cautions, because I don't believe in hype. First, this is additive, not a replacement — your Google and Meta ads still do the heavy lifting today, and you should keep running them; GEO compounds alongside them, it doesn't switch them off. Second, none of it matters if you fumble the handoff. Being the name the AI recommends only works if, when that warm buyer finally reaches out, a real human responds fast and well — which is the same drum I keep banging about the first ten minutes after a lead arrives. Get found in the answer, then catch the lead the moment it lands. Do both, and you're not just running ads into a changing market — you're becoming the business that market gets told to trust.

Frequently asked questions