The question we asked
We opened ChatGPT and typed one prompt, exactly as a homeowner would:
“Who is the best plumber in Prescott, AZ? List 3-5 specific businesses by name with a short description of each.”
This isn’t a hypothetical. This is how a growing share of customers now find local services. Instead of scrolling through ten blue links and comparing review counts, they ask an AI assistant one question and get one answer. Bain’s 2025 consumer research found that about 80% of consumers now rely on AI-generated summaries for at least 40% of their searches — and many never click through to a website at all. If your business isn’t in the answer, you’re not in the conversation.
What ChatGPT said
ChatGPT gave the standard disclaimer that it can’t provide real-time reviews, then did exactly what we asked: it named five businesses.
The list: Yavapai Plumbing & Heating, Benjamin Franklin Plumbing, The Plumber Guy, Sundt Construction, and Mr. Rooter Plumbing of Yavapai, Coconino and Maricopa Counties.
Each came with a short, confident description — Yavapai Plumbing & Heating for its service range across residential and commercial, Benjamin Franklin Plumbing for punctuality, The Plumber Guy for affordability and speed, Mr. Rooter for emergency service and franchise consistency.
One entry stood out. Sundt Construction is primarily a large construction company, not a residential plumber you’d call about a water heater. ChatGPT even said so in its own description — and included it anyway. That’s worth remembering: the model isn’t verifying who actually answers a Saturday drain call. It’s pattern-matching on the strength of the signals it has seen.
Why these businesses got recommended
AI engines don’t crawl Prescott and inspect anyone’s work. They assemble answers from the sources they trust, and research has started to map what earns that trust.
The Princeton GEO study (KDD 2024) tested which content changes make a business more likely to be cited in AI-generated answers. Two findings matter most here: adding relevant statistics to your pages increased visibility in AI answers by roughly 41%, and citing credible sources boosted visibility by as much as 115% for sites that weren’t already ranking at the top.
Applied to this list, the likely signals are:
- Authoritative directory presence. These businesses appear consistently across the major local listing platforms with matching names, addresses, and service descriptions. That repetition across trusted sources is exactly what a language model learns from.
- Review density. Not just star ratings — volume of written reviews that describe specific services (“water heater,” “emergency,” “repipe”) in plain language the model can associate with the business name.
- Brand mentions beyond their own site. National franchises like Benjamin Franklin Plumbing and Mr. Rooter benefit from thousands of pages across the web mentioning the brand, which local independents rarely match.
- Content density. Sites that spell out services, service areas, and specifics give the model more to quote. Thin one-page sites give it almost nothing.
What the recommended businesses have in common
Four patterns show up across this list:
- Descriptive, findable names. Yavapai Plumbing & Heating and The Plumber Guy say what they do and roughly where. The model doesn’t have to infer anything.
- Franchise-scale citation footprints. Two of the five are national franchise locations. Their brand equity was built by content published far outside Prescott — and it carried the local branch into the answer.
- Years of accumulated web presence. Every business named has been listed, reviewed, and mentioned for a long time. AI answers reward accumulated history, not recent effort.
- Service-area breadth in writing. The businesses that got picked describe commercial work, emergency service, and multiple counties on their pages and profiles. Broader documented scope reads as authority.
What’s missing from the plumbers who weren’t recommended
Prescott has dozens of licensed plumbers. Most of them do excellent work. None of the others made this answer, and the gaps are usually predictable:
- Thin or inconsistent directory profiles. A listing with a bare name and phone number, or an address that doesn’t match the website, gives the model nothing reliable to cite.
- No statistics or specifics on the website. Princeton’s finding was blunt: pages with concrete numbers earned about 41% more visibility in AI answers. “Family owned, serving Prescott since 1998, 24-hour response” beats “quality service you can trust” every time.
- No credible sourcing or structured content. The 115% visibility gain from citing credible sources went disproportionately to sites that weren’t already dominant — meaning this is the one lever where a small local shop can out-signal a franchise. Almost none use it.
- Reviews that are sparse or vague. Ten reviews saying “great guy” teach the model far less than a hundred reviews describing actual jobs.
The frustrating part: none of these gaps have anything to do with plumbing skill. The best plumber in Prescott may have lost this answer to a construction company.
What this means for your business
If you own a plumbing company in Prescott — or anywhere — assume customers are already asking AI assistants this exact question. The winners are being chosen by citation signals you can influence: complete and consistent directory profiles, a website with real specifics and numbers, review volume that describes actual services, and content that gives the model something concrete to repeat. This is fixable, measurable work. But it only pays off if you know where you stand today.
Want to see your score?
We run this exact test for local businesses every day. Get your free AI Visibility Score at rankforward.ai/score — we’ll ask the AI engines about your industry in your market and show you whether you’re in the answer, who is, and why.