Why AI picks them and not you

Last week I talked about the Lost Decade. The marketing many stopped doing because it didn't help Google rankings.

This week: what AI actually looks for when it decides who to recommend.

Because "AI visibility" is still vague for most people. You know it matters.

You've checked whether you appear. But what determines whether you show up or not?

Here's what I've found after running hundreds of these queries across dozens of verticals.

The fundamental difference between GEO and SEO

Google answers: "Here are businesses that match your search, sorted by relevance and proximity."

AI answers: "Here's who I'd recommend for your specific situation, and why."

That "and why" is the key. AI needs to justify its recommendation. It needs material to work with. It needs something to say about you that makes you the right answer for this particular query.

Businesses that give AI that material get recommended. Businesses that don't, don’t.

Signal 1: Entity clarity

AI needs to understand what you are before it can recommend you.

This sounds obvious, but most businesses fail here. They assume AI knows who they are because Google knows who they are.

Different systems. Different knowledge bases.

Entity clarity means:
→ Consistent NAP (name, address, phone) across the web
→ A clear description of what you do across the web
→ Your business mentioned by name in third-party sources it uses

A regional auto dealer group I work with had a problem: their individual dealership names appeared nowhere except their own websites. Corporate press releases mentioned the parent company. Reviews mentioned salespeople by name but not the dealership. Local directories had inconsistent naming.

Google figured it out through GBP. ChatGPT wasn’t connecting the dots. The dealerships were invisible in recommendations until we fixed the fragmentation.

AI needs sources to verify against. Give it them!

Signal 2: Review content (not just rating)

Google uses review rating and review count as ranking signals. AI uses review content as recommendation material.

A 4.8 rating with 500 reviews tells Google you're credible. But if all 500 reviews say "great service, highly recommend," AI has nothing to work with.

Compare these:

Review A: "Great experience. Would definitely recommend!"

Review B: "We came in with a complex estate planning situation as we need a business succession for a family-owned company plus some tricky trust structures. Sarah walked us through the options over three meetings, explained the tax implications clearly, and got everything documented properly. Really knew her stuff with multi-generational family businesses."

Review A helps your star rating. Review B helps AI recommend you for "estate planning attorney for family business succession."

AI is reading the text. It's extracting what you're good at, who you help, what makes you different. Generic reviews are not helpful to this process.

Signal 3: Service specificity

AI knows a lot more about context than traditional search provides.

This allows AI to be more exact in matching specific queries to specific capabilities. It’s looking for what differentiates you as a business from your competitors doing the same things.

Which means vague service descriptions and serving everyone don’t work well.

"We offer comprehensive financial planning services" matches nothing specific.

"We specialize in retirement planning for corporate executives navigating concentrated stock positions and deferred compensation" matches exactly those queries.

The more specific your service descriptions, the more specific queries you can match.

This is the opposite of traditional SEO thinking, where you'd keep descriptions broad to rank for more keywords. AI doesn't work that way. Specificity is what gets you into the recommendation.

For multi-location businesses, this often means the same service described differently by location because the market needs are different, and the queries will be different.

Signal 4: Third-party corroboration

This is big. AI trusts what others say about you more than what you say about yourself.

Your website says you're "the region's leading provider of wealth management services."

Okay. Says who?

If your managing director was quoted in the local business journal about market trends, if your firm won a regional "Best Wealth Manager" award, if you're mentioned in a financial planning subreddit as someone who actually knows what they're doing well, AI has something to verify your claims.

Third-party mentions:
→ Press coverage (local and trade)
→ Awards and recognition lists
→ Industry association features
→ Community organization mentions
→ Expert quotes in relevant publications
→ User-generated content (Reddit, forums, social mentions)

This is the Lost Decade stuff. The things we stopped doing because they didn't move Google rankings. AI eats them up.

Signal 5: Content clarity (not hidden behind friction)

AI can parse natural language. It doesn't need schema to understand what you do.

But it does need to find the information. And it needs the information to be clear.

Common problems:

→ Key services only described in PDFs that AI may not crawl
→ Important details buried in accordion menus or tabs that require interaction
→ Information spread across dozens of pages with no clear summary
→ Marketing fluff that sounds good but says nothing specific

A financial advisory firm I audited had 47 pages of content. Great information! But it was spread thin across the site, buried in expandable sections, and written in vague marketing language.

AI couldn't efficiently extract what they actually did or who they served.

Rewriting their core service pages in plain, specific language made them visible within 60 days. No schema changes. No links. Just more clarity.

The question isn't "is this machine-readable?" It's "if AI reads this page, will it understand what we do and who we help?"

Signal 6: Completeness beyond your website

AI isn't just looking at your website. It's synthesizing information from everywhere it can find you.

→ Your GBP (especially important now that Ask Maps pulls from it)
→ Your directory listings
→ Social profiles
→ Press mentions
→ Reviews across platforms
→ Any third-party content that mentions you

Gaps and inconsistencies hurt you. If your website says you offer 24/7 emergency service but your GBP doesn't mention it and no reviews reference it, AI may not trust the claim.

The businesses that win here have complete, consistent, corroborated information across all surfaces. Not just an optimized GBP and a good website.

Everything aligned and everything saying the same thing.

What doesn't matter (as much)

Some things that matter a lot for Google matter less for AI recommendations:

→ Map pack rank — AI doesn't care if you're #1 or #7 in the local pack
→ Keyword optimization — AI understands semantic meaning, not keyword density
→ Domain authority — AI doesn't weight backlink metrics the way Google does
→ Citation volume — 50 directory listings don't help if they're just NAP with no context
→ Proximity — AI considers location but isn't as proximity-weighted as Google

This is why some businesses dominate Google but disappear in AI, and vice versa. Different systems, different signals.

The audit checklist

Here's how to evaluate your current position:

Entity clarity:
□ Is your business name consistent everywhere it appears?
□ Can AI connect your brand name to your locations?
□ Do third parties mention you by name?

Review content:
□ Do your reviews mention specific services, outcomes, or situations?
□ Could AI extract "what you're good at" from your review text?
□ Or are your reviews generic praise with no detail?

Service specificity:
□ Are your services described specifically enough to match specific queries?
□ Would AI know to recommend you for "retirement planning for executives with stock options" vs. just "financial advisor"?

Third-party corroboration:
□ Does anyone besides you say you're good?
□ Press mentions? Awards? Industry recognition? Community features?
□ When was your last piece of earned media?

Structured data:
□ Do you have LocalBusiness schema implemented?
□ Are your services, hours, areas, and contact info machine-readable?
□ Can AI parse your site efficiently?

Cross-platform completeness:
□ Is your information consistent across GBP, website, directories, social?
□ Are there gaps where AI might find conflicting information?
□ Does your GBP have the depth to support Ask Maps queries?

For multi-location: run this audit per location, not just at brand level. The variance across your portfolio is probably significant.

Where to go from here

Being findable on Google is not the same as being recommendable by AI.

Findable means you show up in search results. Recommendable means AI has enough material to confidently suggest you for a specific need.

Most businesses are findable. Few are recommendable.

The signals above are what close that gap. Entity clarity. Review content. Service specificity. Third-party corroboration. Structured data. Cross-platform completeness.

Next week: the playbook. Specific fixes, prioritized by impact, for single locations and multi-location portfolios.

Talk soon,
Garrett

P.S. — If you want to see where you stand: amiagentready.com checks structured data and site readiness. localseodata.com/tools checks AI visibility across platforms. Run both and you'll know which signals need work.