13 First-Page Plumbers Were Named in 48 of 260 AI Answers

This is the third study in the series. In August we tested thirteen roofing companies. On October 7 we tested thirteen HVAC companies that already hold Google's first page in their metro. For plumbing we kept everything the same, down to the cities: thirteen plumbing companies, each the first plumber's own site on Google's first page for "plumber [city]", in the same thirteen metros as the HVAC study.

The plumbers did worse. They were named in 48 of 260 answers, 18 percent, against 31 percent for the HVAC companies in the same cities. And one name came up more than all thirteen of them together: Roto-Rooter, a national franchise, was named in 73 of the 260 answers, in every one of the thirteen markets.

13

plumbing companies, one per metro, each on Google's first page for its city

48 of 260

answers named the tested plumber, 18 percent

73

answers named Roto-Rooter, across all 13 metros

3 of 13

first-page plumbers were named in none of their 20 answers

2 of 13

company sites answer Claude's crawler with 403 Forbidden

As with the earlier studies, this is not a judgement on anyone's plumbing. It is a measurement of what four assistants do with a local plumbing question, and of what they build the answer from.

How we picked the 13 plumbers and what we asked

We pulled Google's organic results for "plumber [city]" from Ahrefs SERP data on October 5, 2026, in the thirteen metros of the HVAC study: Boston, Atlanta, Seattle, Minneapolis, Columbus, Charlotte, Tampa, Dallas, Phoenix, Chicago, Nashville, Denver and Philadelphia. In each we took the first plumbing company's own site. We skipped directories, Reddit, national franchise pages, HVAC-first companies, and any company that was already in the HVAC study.

The testing tool wrote five buyer questions for each company from what its website says, so the questions follow the work each plumber advertises: drain cleaning and sewer camera inspections in the Twin Cities, slab leak repair in the Phoenix Valley, water heater repair in Columbus. Across the 65 prompts the shapes were:

1. 01Who are the top [plumbers / drain cleaners / emergency plumbers] in [metro] in 2026? The plain list question.
2. 02Which [plumbing company] in [metro] offers the best value?
3. 03Which plumber in [metro] do you recommend for [a job]? Slab leak repair, same-day repairs, recurring sewer backups.
4. 04Which plumbing company in [metro] has the best reviews? or is the most recommended?
5. 05Who does the best [job] in [metro]? or Where can I get [job] in [metro]? Water heater repair, sewer camera inspection, emergency drain cleaning.

Each prompt went to ChatGPT, Gemini, Claude and Perplexity through the same visibility-testing connector as the two earlier studies: fresh sessions, no account history, no location set, the smaller tier of each assistant. Twenty answers per company, 260 in total, collected on October 5, 2026.

Two numbers per company, as in the HVAC study. Linked, of 5: in how many of the five prompts did at least one assistant link the company's own website. Named, of 20: in how many of the twenty answers did the company's name appear at all. New this time, we also checked whether each company's website lets the AI assistants' crawlers read it.

The results, all 13 plumbers

Sorted by prompts linked, then answers named. Mention counts in the names-instead column are across that company's twenty answers.

The table is available as a spreadsheet: download the study data as CSV, 13 rows, one per company. Every one of the 260 answers is published too; the data appendix at the end explains the file.

What the 260 answers say

1. A national franchise out-named every local plumber

Roto-Rooter was named in 73 of the 260 answers, and it appeared in the answers for all thirteen companies, from Boston to Phoenix. The thirteen local plumbers that hold Google's first page were named 48 times between them. When an assistant has to name a plumber in a city it knows little about, it reaches for the name with a location page in city after city: rotorooter.com had a city page on Google's first page for "plumber [city]" in eight of the thirteen metros.

The franchise's own site was among the most-cited sources too: rotorooter.com was cited 67 times. In the HVAC study, equipment makers like Carrier and Trane were among the names given most often. In plumbing the gap is filled by a franchise with a page per city, which is the clearer lesson: a page about your service in your town is what the assistants reach for.

2. Plumbers trail HVAC companies in the same cities

Same cities, same method, two weeks apart. The plumbers were named less often than the HVAC companies in nine of the thirteen metros, more often in three, and the same in one. In Boston, Atlanta and Phoenix the HVAC company was named 11, 8 and 5 times; the plumber in each was named in none of its twenty answers.

Philadelphia runs the other way: the HVAC company there was named in none of its answers, and the plumber in ten. A first-page ranking in one trade says nothing about the next trade over, even on the same street.

3. Three first-page plumbers were never named, and other plumbers' sites wrote their answers

The Boston Plumber, Turner Signature Plumbing in Atlanta and Pink Plumbing & Sewers in Phoenix were named in none of their twenty answers. Look at what those answers were built from and the pattern is plain. In Boston the most-cited source was a competing plumber's website, cited 11 times. In Phoenix it was another competitor's site, cited 13 times. In Atlanta it was an AI-era plumbing directory and another plumber's site, 12 and 10 times.

The Boston Plumber's own site was cited once in twenty answers; Turner's and Pink's were not cited at all. Holding Google's first page put them in front of people searching Google; it did not put them in front of the assistants.

4. Directories and lead marketplaces are the wiring

The same AI-era directory topped the HVAC study, and it tops this one: serviceagent.ai was cited 165 times. Lead marketplaces were cited more in plumbing than in HVAC, 166 times against 113, and Reddit threads 56 times. The tested plumbers' own sites were cited 46 times across all 260 answers.

5. Gemini names local plumbers three times as often as ChatGPT

Gemini named the tested plumber in 22 of its 65 answers, ChatGPT in 7. Gemini led in the HVAC study too. The ordering matters less than the spread: the same thirteen companies and the same prompts produced very different answers depending on which assistant a homeowner happened to open.

6. New: two sites turn away Claude's crawler

This time we also asked each company's website for its homepage the way the assistants' crawlers do, from two networks: an ordinary office connection and a data center, the kind of network the real crawlers use. We count a block only when both agree. McNatt Plumbing in Tampa and Christopher's Plumbing in Denver answer Claude's crawler with "403 Forbidden" from both, while letting ChatGPT's, Perplexity's and Google's crawlers in. Nine sites answered every crawler normally. Two (Goodman Plumbing and Pink Plumbing & Sewers) didn't answer our data-center checks at all, so we report them as inconclusive rather than blocked.

Claude named neither McNatt nor Christopher's in any answer; Gemini named both. Two companies are far too few to prove cause, and most owners don't know their site does this. A security plugin, a firewall rule or the host can switch it on without anyone deciding to. One more lesson from doing this: a robot-check screen we saw on one site from the office connection didn't appear from the data center, so a check from a single network can mislead. It takes a few minutes to check properly, and it's the one thing on this list that can rule a company out entirely.

What this means for a plumbing company

1. 01Make sure the assistants can read the site. Check robots.txt for GPTBot, ClaudeBot and PerplexityBot, and ask whoever runs your hosting or security plugin whether those crawlers get through. Two of thirteen first-page sites were refusing Claude's crawler. The free AI crawler checker does both checks in ten seconds.
2. 02One page per job and per town. The prompts were built around specific work: slab leaks, sewer camera inspections, water heaters, emergency drain cleaning. A franchise wins those questions with a page for each city. A local plumber can win them with a page for each service in each town it actually works.
3. 03Put the proof on the page. Review count and rating, license number, years in business, the towns by name. That is what an assistant can quote back to a homeowner.
4. 04Be in the places the answers are built from. The directories and review sites in the chart above, with a complete profile and the same name, address and phone everywhere.
5. 05Write answers, not slogans. A short FAQ in the words homeowners use ("how much does it cost to clear a main line", "do you do same-day water heater replacement") is the shape assistants lift.

To see where your company lands, run the AI visibility test on your own business, free, and we will send the twenty raw answers. The pages per job and per town are plumber SEO, and the site they sit on is plumber website design. The earlier studies: HVAC companies and roofing companies.

Data appendix

Two files sit behind this article. The summary has 13 rows, one per company, including the crawler-access result. The answer-level file has 260 rows, one per answer, for all thirteen companies. Every number in the article can be recomputed from them.

How it was collected

• When: October 5, 2026, for the AI answers, the Google results and the crawler check.
• How: a third-party visibility-testing tool, called through its API. It writes the prompts from each company's website, sends each one to the four assistants with no account history and no location, and records what comes back.
• Models, as recorded: ChatGPT gpt-4o-mini, Perplexity sonar, Gemini gemini-3.1-flash-lite, Claude claude-haiku-4-5. These are the smaller API models, not necessarily what the consumer apps answer with.
• Crawler check: each homepage requested once per crawler, identifying as that crawler, from an office connection and again from a data-center connection. A block counts only when both agree. Real crawlers come from their companies' own networks, which a host can treat differently, so a block here means the site's rules turn that crawler's name away.
• One run per company. Nothing was repeated, so the file is one day's answers, not an average.

What the columns mean

• own_site_linked: the tool's flag that the answer put the company's own website in front of the reader. A prompt counts toward linked, of 5, if any of its four answers has it.
• company_named: the company appears in the tool's list of brands named in that answer, link or not. This is named, of 20.
• cited_domains and cited_urls: the sources the tool recorded for the answer, the pages the assistant drew on. A source is not a click.
• answer_excerpt: the first 400 or so characters of the answer. The tool stores only that much.

Checks and edits

• Recomputed from the answer file, all thirteen companies match the published scores on both counts, and the total is 48 of 260 answers naming the tested company.
• The three lead marketplaces appear as [lead marketplace] in the sources, the links and the answer text, as in the HVAC study. Marketing-firm pages appear as [marketing firm, withheld]; we don't name other agencies. Nothing else is edited.
• Roto-Rooter's 73 counts answers where any form of the name appears ("Roto-Rooter", "Roto-Rooter Plumbing & Water Cleanup").

Both files are free to use under CC BY 4.0, with credit to Riptide Growth.