Research Study
What AI Assistants Actually Recommend When a Homeowner Asks for an Electrician
When a homeowner asks ChatGPT, Claude, or Gemini for the best electrician in their city, do the three assistants recommend the same companies?
Key Finding
No. Across eight metro markets, 89% of the electricians named (94 of 106 company names) were recommended by only one of the three assistants, and no company was named by all three in any market.
- Dataset
- LeadValets AI Answer Panel — electrician recommendation sweep
- Sample Size
- N = 120 AI recommendations across 3 engines and 8 metro markets
- Study Period
- Collected November 9, 2026
- Author
- Jonathan Floyd
Last updated:
How we measure thisSummary
We asked three leading AI assistants the same question a homeowner asks — who should I call for an electrician in my city — in eight metro markets, once per assistant per market. Each answer produced five company names, 120 recommendations in total. The assistants barely agreed with each other: 94 of the 106 company names that appeared were named by only one assistant, 12 by two, and none by all three. One in four recommendations (28 of 120) was a national franchise network brand, and every market included at least one.
Findings
Finding 1: The three assistants named almost entirely different electricians
Bar chart of 106 distinct company names: 94 were named by only one AI assistant, 12 by two assistants, and none by all three.
View source data table
| Assistant agreement | Company names |
|---|---|
| Named by 1 assistant | 94 |
| Named by 2 assistants | 12 |
| Named by all 3 | 0 |
Of the 106 distinct company names across the eight markets, 94 were named by a single assistant and only 12 were named by two. Not one company was named by all three. In Kansas City the overlap was zero: three answers, fifteen recommendations, fifteen different companies. For an electrician, this means a recommendation in one assistant is not visibility in AI search — each engine is arriving at its list independently, so a business has to earn its place three times over.
Finding 2: National franchise brands took one in four recommendation slots
Bar chart of 120 recommendations: 28 were national franchise network brands and 92 were independent or regional companies.
View source data table
| Company type | Recommendations |
|---|---|
| Franchise network brand | 28 |
| Independent or regional | 92 |
Twenty-eight of the 120 recommendations belonged to a national franchise network, and every one of the eight markets included at least one franchise brand in its answer. In Nashville and Indianapolis, five of the fifteen recommendations were franchise brands. Franchise systems publish standardized service pages, hours, and location data at scale, which is exactly the material answer engines draw on — an independent contractor with a single well-built page is competing against that infrastructure, not just against the franchise down the street.
Finding 3: Even within one market, the shortlist barely repeated
Bar chart of distinct company names per market out of 15 recommendation slots: Houston 14, San Antonio 14, St. Louis 13, Kansas City 15, Nashville 12, Louisville 12, Oklahoma City 13, Indianapolis 13.
View source data table
| Market | Distinct companies named |
|---|---|
| Houston | 14 |
| San Antonio | 14 |
| St. Louis | 13 |
| Kansas City | 15 |
| Nashville | 12 |
| Louisville | 12 |
| Oklahoma City | 13 |
| Indianapolis | 13 |
Each market offered fifteen recommendation slots across three assistants. Depending on the market, those slots went to 12 to 15 different companies, and in half the markets only one or two names appeared in more than one answer. A homeowner who asks two assistants gets two different shortlists, which means the practical question for a contractor is not "am I ranking" but "which answers am I missing from, and why".
What This Means for Plumbing and HVAC Owners
For electrical contractors: being the obvious choice in Google is not the same as being named by an AI assistant, and being named by one assistant is not visibility in the others. Each engine pulls from different signals, so the work is to make one business unmistakably legible — consistent name, address, and phone data, license and service-area facts stated plainly, and pages that answer the exact question a homeowner asks — and then to check, assistant by assistant, whether the name actually shows up. That is what the Search Authority Audit measures.
Limitations
Three models were used as stand-ins for the three assistants (GPT, Claude, and Gemini families), collected on a single day, in eight metro markets, with one prompt per market per assistant. Answers vary run to run and by location, so these figures describe this collection, not a fixed ranking. Franchise classification counts national franchise network ownership (Mister Sparky, Mr. Electric, Benjamin Franklin Plumbing, One Hour Heating & Air) and does not assess any company's quality. This study measures what the assistants named; it does not yet compare those names against Google's local rankings in the same markets.
How to Cite This Study
Floyd, J. (2026). What AI Assistants Actually Recommend When a Homeowner Asks for an Electrician. LeadValets Research Lab. https://leadvalets.com/research/studies/ai-electrician-recommendations-2026