Background so you know where this is coming from: I've been doing search marketing since before Google was the answer. Back when Yahoo, Lycos, and AltaVista were the battleground, I was the guy reverse-engineering the top-ranked pages to figure out what the algorithm actually rewarded. No courses existed. No gurus. You just pulled apart what was winning and built a model.

That same approach is what I applied here — except instead of search rankings, the question was: why does an AI assistant recommend THIS plumber, THIS dentist, THIS restaurant when someone asks for the best one nearby?

Here's what the research actually showed:

The volume of factors is bigger than most people realize. We identified roughly 150 measurable signals across the businesses that consistently showed up in AI recommendations. Not all of them matter for every business type — a restaurant has a completely different relevant factor set than a contractor — but the breadth surprised even me.

The factors aren't mysterious. The gap is in execution. Things like: whether your business license is verifiable and linked to a state database (huge for contractors, irrelevant for restaurants), whether your hours are consistent across every surface AI might read, whether the language on your site matches the way people actually describe the problem they're trying to solve, whether third-party review platforms corroborate what your own site claims about you.

None of this is magic. It's alignment. AI synthesizes sources. If your sources contradict each other or leave gaps, the model loses confidence in you as an answer.

Most local businesses scored poorly — not because they're bad businesses, but because nobody told them these signals exist. The average score across the businesses we analyzed was genuinely low. These weren't neglected businesses. Many had been doing SEO for years. But the criteria AI uses to build trust in a local recommendation is different from what got you ranked on page one in 2019.

The competitive intelligence angle is underused. The most actionable thing any marketer can do right now is pull the top AI-recommended businesses in a client's category and geography, map their signals, and do a direct comparison. Where your client has gaps that competitors have filled — that's your work order. It's the same reverse-engineering logic that worked in 1999 and it works now.

My honest take on where the industry is stuck: There are a lot of smart people having very sophisticated conversations about whether backlinks still matter for AI, what "GEO" means vs "AEO" vs "SEO", whether citations are dead, etc. Those are real questions. But while we're debating the finer points, local businesses are losing recommendation share right now. An 80% correct strategy executed today beats a perfect strategy delivered in Q3.

The pattern I keep seeing: businesses that are winning AI recommendations aren't doing anything exotic. They've just closed more of the basic gaps than their competitors have.

Happy to go deep on any specific factor category if that's useful. What are you seeing in your own testing?