Pillar 1 · AI visibility

Does ChatGPT recommend your business? We asked 750 times.

When someone asks ChatGPT "who should I call for pest control near me," somebody gets recommended. Over two weeks, our measurement panels asked ChatGPT, Gemini, and Google's AI Mode roughly 750 buyer-style questions about two ordinary local trades: a junk-removal company in the Pacific Northwest and a pest-control company in the Deep South. And we stored every answer word for word. Not a score. The actual words.

What we found

Business one is living the dream and doesn't know it.

The junk-removal company was named in 18 of 22 analyzed answers on buyer questions; ChatGPT listed it first on most of them. Its owner has no idea. Nothing on its website mentions AI. It has simply earned the profile, the reviews, and the citations that assistants lean on, and it is quietly collecting customers who never saw a Google results page.

Business two is invisible while a rival owns the conversation.

The pest-control company appeared in 9 of 71 analyzed answers. One competitor, same city, same services, was named in 57 of 71. When a thousand residents ask an assistant who to call this year, that gap is a phone that rings across town, every time.

How we measured this (and why "AI scores" can't)

Each business got a panel of ~25 questions real buyers ask ("who should I call for...", "best ... near me that answers fast"), asked daily across the three assistants. We store every answer verbatim, then count who was named: mention share = answers naming the business ÷ answers analyzed. Failed fetches are excluded from the denominator, not papered over.

No model guesses a score; the assistants' own words are the dataset. That is the difference between measurement and the "AI visibility score" widgets appearing everywhere: a score you can't expand into the words behind it is an opinion.

What actually moves these answers

From reading thousands of these answers, the assistants lean on the same substrate local SEO always did, plus recency: a complete, active Google Business Profile; review volume AND velocity; consistent directory presence; and a website that plainly says what you do and where. The playbook isn't magic: it's the honest fundamentals, now with a second search engine reading them.

Three things to do this month

Find out. Ask the three assistants your own buyer questions, or run a grade that includes it (ours does, free, in about a minute).

Fix the substrate. Claim the missing directories, answer the reviews, make the site say the city out loud.

Watch it monthly. These answers move; a business that measures them sees the drift while it's cheap to correct.

Three questions we get

Is AI really sending customers? Assistant answers increasingly front the buying journey the way the map pack did in 2015. Early movers banked years of cheap wins then too.

Can you pay to be recommended? No. Which is exactly why earning it matters.

Does this replace Google rankings? No, it stacks on top; the map still matters, and both read the same fundamentals.

Want the 60-second version for your business?

The free Lighthouse Grade measures where you stand on Google, the map, your reviews, AND whether the assistants recommend you. Real numbers, not guesses.

Get your free grade

Related on the data desk: Your rank isn't a number. It's a map. · How we measure everything