Stop Optimizing for the AI. It Already Decided.

There is a new panic making the rounds, and it goes like this: AI assistants are now choosing local businesses for your customers before those customers ever land on your website. Someone asks ChatGPT or Google's AI answers for the best plumber, the best dentist, the best law firm near them, and a short list appears. Three names. Maybe four. If you are not on it, you basically do not exist for that person.

That part is true. The panic is the problem.

Because the very next thing everyone says is: so how do we optimize for the AI? How do we get the robot to like us? And I want to push on that question hard, because I think it is the wrong one, and chasing it is going to waste a lot of good money over the next two years.

The robot is not the customer. It is the receptionist.

Here is the mental model I keep coming back to with clients. The AI assistant is not a new audience you need to seduce. It is a receptionist standing between you and the audience you already had.

When someone asked Google for a plumber five years ago, they got ten blue links and made their own decision. Now they get a paragraph and a shortlist. The decision got made for them, or at least most of the way made. That is a real shift and I am not waving it off.

But think about what a good receptionist actually does. She does not invent an opinion about you out of thin air. She repeats what she has heard. She tells the caller what everyone already says about you, and she does it fast.

The AI is not forming a fresh opinion of your business. It is summarizing the reputation you already built, or the one you neglected to.

That distinction matters, because you optimize a search engine and you earn a reputation. Those are two completely different jobs. And most of the advice floating around right now is quietly assuming they are the same.

Why you cannot game this the way you gamed Google

Let me be honest about the old game, because we all played it. For twenty years, ranking on Google was partly a contest of who understood the mechanics. Keyword density. Backlinks you bought or begged for. A site structure that pleased the crawler. Plenty of mediocre businesses outranked better ones because they were better at the plumbing behind the page.

You could whisper to that algorithm. You could nudge it.

The AI answer works differently, and the difference is not subtle. When a model builds a local recommendation, it is pulling together a cluster of signals: your Google Business Profile, your reviews and what they actually say, mentions of you across directories and local publications, how consistent your basic information is everywhere it appears, and how often the wider web treats you as the obvious answer to a specific question. Then it compresses all of that into a few sentences.

You cannot stuff a keyword into that. There is no meta tag for "trusted by the neighborhood." The model is reading the consensus about you, and consensus is a slow, distributed thing that lives in hundreds of places you do not control.

You can trick a page. You cannot trick a consensus.

So when someone sells you a package to "optimize your business for AI search," ask them the uncomfortable question. Optimize what, exactly? Because if the answer is a fresh batch of thin blog posts and some schema markup, they are selling you a hammer and telling you it is a house.

The signals are old. The stakes are new.

Here is the part nobody wants to hear, because it is not exciting. Almost everything that makes an AI recommend your local business is stuff good marketers have been telling you to do for a decade.

A complete, accurate, actively maintained Google Business Profile. Reviews that are recent, specific, and answered like a human wrote them. Consistent name, address, and phone number everywhere you appear, so the machine never has to guess whether two listings are the same company. A real presence in local coverage and directories that matter in your area. Content that answers the actual questions your customers ask, in plain language, so a model can lift a clean sentence and hand it over.

None of that is new. I have been saying versions of it since before "AEO" was an acronym anyone bothered to invent.

So what changed? The cost of ignoring it.

In the old world, a customer with ten links could still stumble onto the good business with the weak online reputation. They might click past the polished mediocre option and find you anyway, because they were doing the sorting themselves. There was slack in the system. Room to be underrated.

The AI removed the slack. Now the reputation you earned is the reputation you get, with no chance for the customer to overrule it.

That is the real story here. Not that the rules changed, but that the penalty for having neglected the old rules just went way up, and it now happens before anyone ever meets you.

What the AI actually rewards, and it is not what you think

I want to be specific, because "build a good reputation" can sound like a fortune cookie. So here is what I actually see moving the needle when a business starts showing up in AI shortlists, based on the accounts I have watched change.

Specificity beats volume. A business that is clearly the answer to a narrow question does better than a business that is vaguely good at everything. The model is trying to match intent. If someone asks for an emergency plumber who works weekends in a specific part of town, and your reviews and your site both make that concrete, you are an easy pick. If you describe yourself as a "full service solutions provider," you are noise. The machine cannot recommend a fog.

Reviews are read, not counted. I know we all learned to chase the star rating. But the models are reading the text now. A wall of "Great service, highly recommend" tells a machine almost nothing. A review that says you fixed a burst pipe at midnight and cleaned up after yourself gives the model something to actually repeat. That matters more than the difference between a 4.7 and a 4.8.

Consistency signals that you are real. This is boring and it is everything. When your details match across every platform, the model trusts that you exist and that it knows who you are. When they conflict, it hedges, and hedging means it recommends the business it is more sure about instead. You do not get the benefit of the doubt from a machine. You get the benefit of the evidence.

Being cited elsewhere beats talking about yourself. A model weighs what other sources say about you more heavily than what you say about you, for the same reason you trust a friend's recommendation over an ad. Getting mentioned in a local roundup, a community publication, a genuine partner's site, that is worth more than another self published page bragging about your values.

The trap of writing for the machine

Now for the mistake I am watching smart people make in real time.

They hear that AI assistants scrape and summarize content, so they start writing content designed to be scraped. Flat, over explained, keyword heavy pages that sound like they were built for a crawler because they were. The logic feels sound. Feed the machine what it eats.

Here is why it backfires. The model is not just extracting text. It is trying to figure out which sources are trustworthy enough to repeat. And content that was obviously manufactured to be extracted reads as exactly that. Thin. Derivative. The same fourteen points every competitor also published because they read the same advice.

Okay, so if everyone in your category publishes the identical AI friendly page about "5 signs you need a new roof," which one does the assistant pick? None of them, usually. It generates its own generic answer and cites nobody, because nobody said anything worth attributing.

Writing for the machine makes you interchangeable, and interchangeable is the one thing that guarantees you never get named.

The businesses that get recommended are the ones that said something a human found worth trusting. The machine is downstream of that. It always has been.

GA4 is going to make you feel like you are dying

One practical warning, because it is going to rattle people. As more of this decision happens before the click, your analytics are going to look worse even as your business does fine. The customer who asked an assistant, got your name, and called you directly does not show up as a tidy conversion path in GA4. There is no keyword, no landing page, no clean attribution line.

You are going to see traffic that feels softer and wonder if something broke.

Nothing broke. The work just moved to a place your dashboard cannot see. This is the moment to stop treating the analytics platform as the scoreboard and start treating it as one instrument on a wider panel. Ask the actual question that matters: are we getting picked? Track how often you appear when you ask the assistants yourself. Ask new customers how they found you and write the answers down. Watch your calls and your booked jobs, not just your sessions.

The map is not the territory. It never was. GA4 is about to make that painfully obvious, and the people who worship the dashboard are going to make bad decisions off numbers that stopped telling the whole story.

So what do you actually do

Not optimize for the AI. I will say it plainly one more time. You do not court the machine.

You become the business the machine cannot avoid mentioning, and you do it by being the answer that is already true. Get your profile complete and current. Earn reviews that describe specific things you did well, and answer them like a person. Fix your inconsistent listings this week, because that is free and it is holding you back. Get named by real sources in your community. Write things worth attributing instead of things built to be scraped.

Every one of those is a reputation move, not a technical trick. And that is the whole point.

The AI assistant did not create a new game with new rules. It hired a receptionist who repeats what the world already says about you, and it fired the customer who used to give you a second look. That is the shift. Not a new algorithm to please, but an old truth with the volume turned all the way up.

The businesses that spend the next year trying to whisper to the robot are going to lose to the ones that quietly became impossible to leave off the list. The robot is not choosing your customers for you. Your reputation is. The robot just says it out loud now, faster and colder than any human ever would.

You cannot argue with a receptionist about what people say about you. You can only change what they say.