Two weeks ago in this newsletter I asked whether the AI tools your customers are using know that you’re a Maine company. It was a fair question, but after I sent it I realized I’d left out the more uncomfortable one: what happens if they don’t? Not in the abstract, but in the specific, way that sales shift on you before you’ve noticed it moving. The phone rings a little less. You blame the season. You blame the economy. You don’t blame a conversation that happened somewhere you can’t see, because you didn’t know that conversation was happening at all. So over the last couple of weeks, we went and actually checked. We sat down and asked ChatGPT, Gemini, and Claude the questions a real buyer asks before they’ve decided who to call.

What surprised me wasn’t the results. It was how easy it is to check this badly. I sort of pride myself on great AI prompts, but this took a bit of finesse to determine the best way to find out how your business is doing.

My first instinct, and I suspect it would be yours, was to type a company name in and see what came back. Tell me about this business. The answer was almost always reassuring. Regardless of the model, it produced a competent, faintly flattering paragraph about who they are and what they do, and I could have closed my laptop feeling fine about the whole thing.

But in reality, that test was worthless. These systems will describe any company you name, because naming it has already done the hard part for them. The question that actually matters is the one a buyer asks before they know who you are. Who does commercial fencing in central Maine? Who should I hire to build a warehouse near Bangor? Which financial institution is good for a small business loan in Maine? Where can I get the best options for boat insurance in Maine? That is where the shortlist gets made, and a business can be entirely absent from that answer while still generating a perfectly nice paragraph when you type its name in.

That’s the difference between recognition and findability. Most established Maine businesses have earned the first one. Considerably fewer have the second, and, even more importantly, almost nobody has actually looked.

The problem is, there is no easy tell for this. When website traffic drops, you see it. When a campaign underperforms, the numbers tell you, and you can go fix something. But nothing notifies you when an AI tool just answered a buying question in your category and didn’t mention you. That absence is invisible to the person it affects (you). You find out much later, indirectly, because the phone rings a little less than it used to, and by then it’s reasonable to blame the season, economy, or a competitor who seems to be advertising more.

There are ways of seeing hard evidence: looking across our own client base at website traffic referred specifically by AI-powered search tools over a thirty-day window, one client received a single visit. Three comparable Maine businesses, similar in size, all selling the kind of considered purchase people research before they buy, each received at least fifty-five in the same period.

Yes, the absolute numbers are small, the channel is young everywhere, and thirty days is thirty days. But a gap that wide is not a rounding error, and gaps like it tend to widen rather than close, because the underlying systems are self-reinforcing. Getting cited is part of what makes you, well, more citable.

Muck Rack published a study in May that analyzed more than 25 million citations across ChatGPT, Claude, and Gemini in seventeen industries. Eighty-four percent of what those systems cited was earned media. Paid and advertorial content accounted for only three-tenths of one percent. That pattern has now held across three studies going back to the middle of last year.

This means the material these tools are actually drawing on is largely not your website. It’s news coverage, trade association records, industry directories, interviews, forum threads, and the accumulated evidence that other people have said something about you somewhere a machine can find it. Which means becoming findable looks a great deal more like traditional public relations than like search engine optimization, and it explains why businesses with a quiet approach to business but a genuinely good website may turn up missing.

If you want to test this properly before you finish your coffee, open a private browser window or a temporary chat so you’re not being shown results shaped by your own history. Ask the question your best customer would have asked the week before they found you, and don’t name your company. Then pay attention to two things: whether you appear at all, and what sources the system cites when it answers. The second one is more useful than the first. Whatever it cited is the kind of place your business needs to exist.

It’s not that there’s a new channel to add to the list. It’s that a stage of the buying process that used to happen in front of you has moved somewhere you can’t see. The shortlist used to form after someone found you. Increasingly, it forms before, in a conversation you aren’t part of and can’t see. Being good at the work doesn’t put you in that conversation. Being on the record does.

Introducing Findable

We built a program around the problem in this issue’s lead article. Findable is our managed search and AI visibility service — SEO, answer engine optimization, and generative engine optimization, grounded in Maine Insights Lab research on how decisions actually get made in this market. Every engagement starts with a foundation month that establishes where you currently stand and what you should be known for.

Learn more →