The AI Search Gap: What We Found After Testing 1,000 Businesses

We ran an experiment. We took the top Google results for plumber, electrician, dentist, accountant, and locksmith queries across 7 cities and asked ChatGPT the exact same question. Then we compared the two lists.

87% of Google's top businesses didn't appear in ChatGPT. Not ranked lower. Not on page two. Completely absent.

That number held up across 143 batch experiments and over 1,000 individual businesses. We tested Manchester, London, Austin, Leeds, Birmingham, and suburbs within each. The gap was consistent everywhere.

Some categories did better than others. Dentists had 29% overlap between Google and ChatGPT results. Electricians, 16%. Plumbers, 13%. Accountants scraped by at 5%. Locksmiths hit zero. Not a single locksmith that ranked on Google appeared in ChatGPT's recommendations.

This matters because people are actually using ChatGPT to find local services. The BBC reported that 45% of consumers now use AI chatbots for local recommendations, up from 6% a year ago. That's not a future trend. That's right now.

The gap exists because ChatGPT doesn't use Google. It runs its searches through Bing, reads the top results, and synthesizes an answer. The businesses that spent years climbing Google's rankings are fighting a battle that ChatGPT doesn't even know about.

We also found something we didn't expect. Different phrasings of the same question return almost entirely different businesses. "Best plumber in Manchester" and "emergency plumber in Manchester" had 5% overlap. Ninety-five percent different businesses for essentially the same need.

And suburbs are their own world. "Best plumber in Didsbury" and "best plumber in Manchester" returned 0% of the same businesses. Every suburb is a separate competition that nobody is tracking.

The traditional advice for "AI SEO" didn't hold up either. Schema markup, robots.txt tweaks, FAQ pages, blog content, content freshness -- we tested all of them across 1,000 businesses. None showed more than 10% correlation with ChatGPT visibility. Several showed negative correlation. The things that actually predict whether ChatGPT recommends you are simpler: having a website at all (the single biggest factor), being listed on multiple platforms beyond Google, and linking your social profiles.

We're publishing all our findings over the next few weeks. The data on what works, what doesn't, and what we got wrong along the way. Specific numbers, specific businesses, specific experiments. No hedging.