We Asked AI About 600 Local Service Businesses. Here’s What It Said.

If a homeowner in Nashville asks ChatGPT for a plumber, there are exactly two businesses in the entire metro that every major AI platform will recommend. Two. In a city with hundreds of licensed plumbers.

We know this because we tested it. We ran 600 real queries across four AI platforms, ten U.S. cities, and five home service trades. We asked ChatGPT, Perplexity, Google AI Overviews, and Gemini the same questions a homeowner would: who is the best plumber, who should I call for AC repair, where do I find a reliable roofer. Then we documented every answer.

This is the first published study of what AI actually says when you ask about local service businesses. Not what it could say. Not what SEO experts predict it will say. What it says right now, today, when a real person types a real question.

What We Did

We selected ten metros ranging from large (Dallas-Fort Worth, Nashville, Portland) to mid-size (Tucson, Omaha, Columbus). Five trades: plumbing, HVAC, electrical, roofing, and landscaping. Three query types per trade, designed to mirror natural homeowner language. That gave us 15 queries per city, times four AI platforms, times ten cities. Six hundred data points.

We coded every response: did the AI name specific local businesses (Type A), give general advice without naming anyone (Type B), recommend only national franchises (Type C), or just send you to a platform like Yelp or Angi (Type D)? In some cities, Google AI Overviews didn’t trigger at all, meaning the query produced no AI-generated answer (Type E).

What We Found

Across all 600 queries, 87% returned specific local business names. That number was even higher on some platforms: Perplexity and ChatGPT both named local businesses on nearly every query. The fear that AI only gives generic advice is wrong. When someone asks for a plumber, AI usually answers with names.

This is the headline finding. Of 1,571 unique businesses named across the study, 80% were recommended by only one AI platform. Just 1.5%, 23 businesses total, appeared on all four. Ask ChatGPT for a roofer and ask Perplexity the same question: you will almost certainly get different names. Each AI tool is pulling from different data sources, weighting different signals, and arriving at different answers.

In Portland, Nashville, and six other cities, Google AI Overviews triggered on every query and named local businesses every time. In Raleigh-Durham and Tucson, they returned zero results on all 15 queries. Not low results. Zero. A homeowner in Portland gets AI-powered recommendations from Google. A homeowner in Raleigh gets nothing. Same queries, same format, completely different experience.

When we measured which platform pairs recommended overlapping businesses most often, Google AI Overviews and Gemini agreed 61% of the time. Both pull from Google’s index. Perplexity and ChatGPT, which rely on different data pipelines, agreed only 37% of the time. The average Jaccard similarity across all pairs was below 0.10, meaning even when two platforms both answer with businesses, fewer than 10% of their recommendations overlap.

Omaha had eight businesses appear on all four platforms. New Orleans had three. Sacramento and Columbus each had four. Dallas-Fort Worth and Raleigh-Durham had zero. In smaller metros with fewer competitors, the strong local brands rise to the top everywhere. In larger markets, the AI tools diverge more because there are more businesses competing for attention.

BBB was the most-referenced third-party source in seven of ten cities, with up to 24 mentions per metro. Yelp dominated on the West Coast. Angi appeared consistently everywhere. ChatGPT in particular leaned heavily on BBB data for its recommendations. Businesses without profiles on these platforms are effectively invisible to the AI tools that cite them.

Franchise brands appeared in 15% to 42% of responses depending on the city. Roto-Rooter was the most frequently cited franchise in the study. But in every single metro, independent local businesses made up the overwhelming majority of AI recommendations. Nashville had the highest franchise penetration at 42%. Portland had the lowest at 15%. Local operators are not being crowded out by national brands in AI search. Not yet.

What This Means If You Run a Service Business

The shift is real. BrightLocal’s 2026 data shows consumer use of AI for local business recommendations jumped from 6% to 45% in a single year. But the data from this study shows something most people haven’t grasped yet: there is no single AI search to optimize for. There are four, and they barely agree on who to recommend.

If you only show up on Google, you are invisible on three out of four AI platforms. That’s not a future problem. It’s a current one.

The 23 businesses that made it onto all four platforms share a pattern: strong review profiles on multiple sites, BBB accreditation, consistent business information across directories, and websites with specific service pages. These are not new ideas. But they used to be optional extras for local SEO. In the AI search era, they are the difference between being recommended and being unknown.

The first step is finding out where you actually stand. Not where you think you stand. Not where your website says you should stand. Where you actually appear, right now, when someone asks an AI for help.

About This Study

This research was produced by Local Vitals, a visibility diagnostic platform for service-area businesses. The study covered 600 queries across ChatGPT, Perplexity, Google AI Overviews, and Gemini in 10 U.S. metro areas and 5 trade categories. The full dataset, including metro-level breakdowns and trade-specific findings, is available at hello@getlocalvitals.com.