Most of the conversation about "AI SEO" is still pretty vague. People say things like "make sure your content is helpful" or "optimize for conversational queries" — which isn't wrong, but it's not actionable either.

Over the past year, I've been doing deep audits on local businesses that are getting recommended by AI assistants (ChatGPT, Perplexity, Gemini, etc.) and businesses that aren't — even when both seem similarly established on paper. I started cataloging every difference I could find and testing whether those differences actually correlated with recommendation frequency.

The result is a working index of 150+ measurable factors. Here's what I learned building it.

The five categories where the factors cluster

1. Profile completeness

This sounds obvious but goes much deeper than "fill out your Google Business Profile." AI assistants pull structured data from multiple sources simultaneously. A business that has complete, consistent information across its GBP, its website structured data, its Bing Places listing, Yelp, and industry directories is dramatically easier for an AI to "trust" and surface confidently.

Factors here include things like:

• Whether hours are consistently formatted across sources
• Whether the business category matches across platforms
• Whether a physical address is present and consistent (even for service-area businesses)
• Whether a business description exists that directly names what the business does in plain language

2. Review signals

Not just star rating and volume — those matter, but the index gets more specific:

• Recency distribution (a business with 200 reviews but none in 8 months scores differently than one with 80 reviews and steady monthly additions)
• Response rate and response quality (AI systems appear to weight whether owners engage)
• Whether reviews use natural service/product language vs. generic praise
• Cross-platform review presence — reviews only on one platform are a weaker signal than reviews distributed across multiple sources

3. Corroboration across the web

This is the one most people underestimate. AI assistants aren't just reading your website. They're effectively asking: does the broader web agree this business exists, does what it says it does, and is credible?

Corroboration factors include:

• Third-party mentions (news coverage, blog posts, local directories, association listings)
• Whether the business is cited in context — not just listed, but mentioned in relevant content
• Whether the business name appears naturally alongside its service category and location across independent sources
• Chamber of commerce, BBB, and professional association listings

The pattern I kept seeing: businesses that got recommended had a "web presence surface area" that was 3-5x larger than businesses that didn't, even when the non-recommended business had a better-looking website.

4. Page readability for machines

This is distinct from SEO in the traditional sense. The question isn't "will Google rank this page" — it's "can an AI system extract accurate, structured facts from this page quickly and confidently?"

Factors here include:

• Whether the homepage clearly states what the business does, who it serves, and where it operates — within the first visible content block
• Whether service pages name specific services (not just categories)
• Whether structured data markup is present and accurate
• Page load behavior and whether content is rendered in a way that's accessible to crawlers
• Whether FAQs exist and are written in natural question-and-answer format

5. Credentials and trust markers

• Licensing information (where applicable to the industry)
• Named staff/ownership — businesses where a real human is identifiable score better
• Years in business, stated clearly
• Awards, certifications, and affiliations — but only when they're verifiable and linked to the issuing organization
• Whether an About page exists and tells a coherent, specific story

What surprised me most

The factors that mattered least were the ones people obsess over most — domain authority, backlink counts, social media follower numbers.

The factors that mattered most were boring and operational: consistency, completeness, and corroboration. A business with a basic website but rock-solid profile data across 15 platforms and steady recent reviews consistently outperformed businesses with polished sites and weak data hygiene.

Also: named humans matter. Businesses where the owner or key staff are identifiable — on the website, in reviews, in third-party mentions — got recommended more reliably. Anonymous "we" language throughout a site appears to be a trust gap.

Why 150+ factors and not just 10?

Because the AI systems themselves are nuanced. A business missing one or two factors doesn't necessarily get dropped from recommendations. But businesses that scored below threshold across a cluster of related factors consistently underperformed.

For example: weak review recency plus no response to reviews plus reviews only on one platform = a compounding trust deficit that no amount of website polish fixes.

The index lets you see where the clusters of weakness are, not just individual gaps.

Happy to discuss methodology, share what I found by industry vertical, or talk through specific factors

This is genuinely a new enough space that I think collective knowledge-building is valuable. What are you all seeing on the ground with AI recommendation visibility for local businesses?