How AI decides which businesses to recommend.

Every recommendation is a judgment made with incomplete information.

That’s true of a machine. It was also true of the Chicago Bears in 2017.

The Bears traded up from third to second that year — giving San Francisco extra picks to do it — and took Mitchell Trubisky. Patrick Mahomes went tenth, to Kansas City. Deshaun Watson went twelfth.

The interesting part isn’t the miss. Everybody knows about the miss. The interesting part is that it was reasonable at the time.

Mahomes was not quiet. He led all of college football in passing yards his final season and threw for 734 yards in a single game against Oklahoma. The numbers were enormous. Evaluators looked at them and mostly shrugged — Texas Tech threw the ball fifty times a game in a wide-open scheme that made numbers easy to pile up, and college quarterbacks coming out of it had a long history of not making it in the NFL. Trubisky started thirteen games in a pro-style offense at North Carolina. Fewer numbers. Cleaner ones. Easier to trust.

So the loudest signal got discounted as noise, and the tidy signal got believed. Everyone had the same information. What separated them was how it got weighed.

That is the whole job an AI is doing when someone asks it who to hire.

It does not know your business. It cannot visit your shop, and it has never met you. What it has is a pile of claims about you, of wildly different quality, and a decision to make about which ones to trust. It is doing what a scout does with a highlight reel — trying to work out which numbers mean something and which ones are a product of the system that produced them.

Which is why the loudest marketing is often the least persuasive to it.

A page that says award-winning, industry-leading, trusted by thousands is production without context. There’s no way to check it, and nothing else in the world confirms it. It reads exactly like inflated stats. Meanwhile a business that says specifically what it does, for whom, in what town — and has customers, partners, and other websites saying the same thing in their own words — is a smaller number in a harder league. It travels.

The two questions underneath this haven’t changed since long before any of it was automated. Do they do the thing I need? Should I believe them? Relevance and authority. Every search engine and AI tool ever built is a machine for guessing at those two answers at a scale no person could manage.

So the work isn’t getting louder. It’s making yourself easy to verify.

Be specific enough to be checkable — real numbers, real names, real detail. Say the same thing everywhere you appear. And tend the reputation you’ve already earned, because what other people say about you is the only part of this a machine treats as evidence rather than as a claim. Everything you say about yourself is a highlight reel. Everything they say is a scouting report.

Kansas City didn’t find something nobody else could see. They weighed what everyone could see, differently.

You cannot control how you get weighed. You can control how much there is to weigh.

What would someone find, if they went looking to check your claim?