Why AI Recommends Some Businesses and Ignores Others

Here’s a conversation I have almost every week now.

An owner opens ChatGPT, types in the exact thing their customers would type, and watches three competitors get named. Not them. They read it twice. Then they call me and ask the obvious question: how do I get in there?

The honest answer is that there’s no there. No listing to claim, no auction to enter, no rep to call, no box to check. An AI recommendation isn’t a placement you buy. It’s a conclusion a model reaches about whether you’re trustworthy, based on everything it can find about you across the open internet.

Which means the question has to change. It’s not “how do I get in there.” It’s “what would have to be true about my business for a reasonable stranger to name me?”

Trust isn’t a Feeling Here. It’s a Pattern. 🧠

We use the word trust loosely in marketing. Usually we mean something warm and hard to pin down. That’s not what’s going on inside these systems.

What a model is doing is closer to corroboration. It’s checking whether multiple independent sources agree that you exist, that you do what you say you do, and that the people who hired you were glad about it afterward.

One source saying you’re excellent is a claim. Twenty unrelated sources implying it is a pattern.

Your website is a claim. Your reviews across a few different platforms, a mention in the local business journal, a listing in an industry directory that’s actually accurate, blog posts that answer questions in the words real customers use, your hours matching everywhere they appear, people talking about you on forums you’ve never visited — that’s a pattern.

Most businesses have spent fifteen years perfecting the claim and almost no time on the pattern. Gorgeous website. Thin, scattered, half-abandoned everything else. And then they’re surprised the machine won’t vouch for them.

The Signals are Boring, Which is Exactly Why Nobody Builds Them 📊

This is the part people don’t want to hear. The inputs to AI trust are unglamorous almost across the board.

How many reviews you have and how recent they are. Whether your business information is consistent everywhere it appears. Whether anyone outside your company has ever written about you. Whether your own content answers real questions in plain language or just performs for a keyword. Whether your presence looks tended or abandoned.

None of that is a campaign. None of it gives anybody a good feeling on a dashboard next Friday. All of it compounds.

Which is precisely why it’s a real advantage. The work is unfashionable enough that most of your competitors won’t stick with it, and slow enough that they can’t catch up in a quarter once you’re genuinely ahead.

I’ve started calling it the trust stack, because the layers hold each other up. Reviews make mentions more believable. Mentions make your content more believable. Consistent information makes all of it easier to attribute to you and not to the similarly named business two towns over. Pull one layer out and the whole thing wobbles.

You Can’t Buy In. But you Can Earn In. 🏆

When a channel gets hard, everybody starts hunting for the paid shortcut. There will be paid surfaces inside AI products, and some of them will be worth buying.

But the recommendation itself — the moment a machine tells a person you’re a good choice — isn’t for sale. I don’t think it’s going to be either, because the moment it is, it stops being worth anything to the person asking.

That should cheer you up rather than worry you. For twenty years the top of the funnel rewarded whoever could outspend everyone else. This rewards whoever built the most credible pattern, and pattern isn’t proportional to budget. A well-run operation with excellent recent reviews, a consistent presence, and real answers published in plain English can absolutely get named ahead of a competitor spending ten times as much on ads.

That was a lot less true in the paid auction era.

The Rejection you Never Find Out About 🚫

The cruel part of all this is how quiet the failure is.

When your ads underperform, you see it. When your rankings slip, you see it. When an AI leaves you out of a recommendation, nothing happens at all. No notification. No line on a report. No dip in any metric you’re already watching. Someone who would have hired you simply never learns you exist, and you never learn they were looking.

That silence is why so many owners feel busy and visible and stuck at the same time. The marketing looks active. The reach numbers look fine. And some unmeasured slice of the market is getting filtered out before the funnel even starts.

So run the crude version of the test yourself. Open the machines, ask the questions your customers ask, phrase them ten different ways like a normal person would, and write down who gets named.

That list is your real consideration set. It costs nothing, it takes twenty minutes, and almost nobody does it.

Start Where the Corroboration is Thinnest 🧩

If you want somewhere to begin, don’t begin with content. Begin with an inventory.

Find every place your business shows up online and check whether the picture agrees with itself. Then find the thinnest layer, which for most businesses is third-party mentions and review recency, and put steady pressure there for a couple of quarters.

It isn’t exciting work. It’s the work.

The businesses getting recommended today didn’t find a clever hack last quarter. They’ve been quietly, consistently trustworthy in public for years, and a machine finally noticed and started paying them for it.