The words were fine. What was underneath them was not.

Earlier this year I asked AI to describe a client's firm. It got almost everything right.

The founder's credentials. His twenty years in the work. A result he is proud of, quoted back nearly word for word.

Then I ran the searches a buyer would run. 15 of them. He came up once, and that one time, the tool got his firm's name wrong.

The words on his site were fine. His homepage explained what he did more clearly than most I have read.

What was wrong sat underneath the words. A field nobody looks at had cut the firm's name off mid-word. His city appeared nowhere except the area code of a phone number in the footer. There was nothing in the code behind the page telling a machine this was a firm, with an address and a set of services. The site still described the business he used to run, not the one he runs now.

None of that is copy. You cannot fix it by explaining yourself better.

We rebuilt the site from the structure up. 30 days later the same searches found him 13 times out of 20, and one tool now recommends him for exactly the kind of project he wants.

The advice everyone gives is to describe your business clearly. It is good advice. It is also where most people stop, because the rest is invisible, and it is not interesting to talk about: how the site is built, the data that tells a machine what the firm is and where it is, whether the firm's name and details match everywhere a machine might look. None of it is written for people. All of it decides whether you are found.

The machine was never confused by what he said. It was confused about who was saying it. That is the part of this work I find hardest to explain, yet it is the part that matters most.

This is the kind of problem the AI Visibility Audit is built to find.