Skip to content
Deep AI VisibilityA Deep AI Solutions Company
← The AI Visibility Index Live-measured · July 2026Edition 01

The HVAC & Plumbing
AI Visibility Index

When a homeowner asks ChatGPT for a plumber — or a contractor asks which software to run — the AI returns a short, confident shortlist. We measured that shortlist. Every one of the five major engines, 38 real buyer questions, 1,103 brand mentions, zero brands fed into a prompt.

5
AI engines
38
Buyer prompts
379
Queries run
1,103
Brand mentions
38
Brands ranked
10.0%
Leader
Roto-Rooter
The central finding

In the largest home-service category in America, the AI answer is decoupled from the industry it describes. Roto-Rooter and Mr. Rooter Plumbing — two franchised networks — win more AI citations than the entire independent trade combined.

39%
Top-5 citation share
3.2×
Engine spread
ChatGPT vs Gemini
24%
Franchise plumbing
33%
Deferred to directories
The Index

Top brands by AI citation share

Share of all brand mentions the five engines volunteered, across every buyer intent. Filter by category, sort by consensus or first-named, and open any row for the per-engine breakdown and the exact language the models used.

Model divergence

Which model names whom

A single blended number hides the most important fact: the five engines do not agree. The same brand can dominate one model and vanish from another. This is the map a blended ranking erases.

ChatGPT
parametric
Claude
parametric
Gemini
parametric
Grok
parametric
Perplexity
web-grounded
21
34
13
24
18
16
30
8
9
16
20
23
3
15
9
16
16
5
16
14
15
14
7
14
15
20
20
2
12
10
14
16
5
13
13
12
16
2
21
8
18
24
2
6
3
11
7
5
5
6

Cell = brand mentions by that engine · intensity normalized per column · hover for share

ChatGPTparametric
304

brand mentions volunteered

Claudeparametric
273

brand mentions volunteered

Grokparametric
227

brand mentions volunteered

Findings

Six things the data makes undeniable

39%

Five brands own a third of every AI answer

Across 1,103 brand mentions, the top five names captured 38.5% of all citations and the top ten captured 66.2%. AI does not return a directory — it returns a shortlist, and the shortlist is short.

3.2×

The engines disagree more than they agree

ChatGPT volunteered 304 brand mentions; Gemini named just 94 — a 3.2× gap on identical questions. Which engine your buyer opens changes the answer they get.

24%

Franchise plumbing owns consumer intent

Roto-Rooter, Mr. Rooter, and Benjamin Franklin Plumbing — three franchised networks — together take 24.0% of citations. National brand recall, not local reputation, is what the models reproduce.

33%

When unsure, AI names a directory or a furnace

Directories (Yelp, Angi) and equipment makers (Carrier, Trane, Lennox) took 33.3% of citations. When a model can't confidently name a local pro, it defers upward — the exact gap a well-structured contractor can capture.

21%

Contractor software is a three-horse race

In trade-intent queries, 21.0% of citations went to field-service platforms — concentrated in ServiceTitan, Jobber, and Housecall Pro. A software choice inherits its platform's citation surface.

82%

Position, not just presence

Roto-Rooter isn't just mentioned — it's named first 81.8% of the times it appears. Being on the list and leading the list are different games, and the Index measures both.

By category

Where the citations concentrate

Every tracked brand rolls up into one of seven categories. This is the shape of the AI answer for the trades.

National Plumbing Services
116 mentions
10.5%
National HVAC Services
106 mentions
9.6%
Franchise Networks
279 mentions
25.3%
Contractor Software & CRM
232 mentions
21.0%
Equipment Manufacturers
219 mentions
19.9%
Directories & Aggregators
148 mentions
13.4%
Lead & Marketing Tech
3 mentions
0.3%
Methodology

Measured, not estimated

The Index is generated by a reproducible harness, not hand-assembled. Here is exactly how, so you can weigh it — and, if you like, reproduce it.

The five engines

ChatGPTOpenAI
gpt-4o
parametric
ClaudeAnthropic
claude-haiku-4-5-20251001
parametric
GeminiGoogle
gemini-2.5-flash
parametric
GrokxAI
grok-3
parametric
PerplexityPerplexity
sonar
web-grounded
  1. 01  Ask 38 open-ended buyer- and trade-intent questions — never naming a brand.
  2. 02  Put every question to all five engines, 2× each (380 planned queries).
  3. 03  Detect every brand each answer volunteers, alias-aware and de-duplicated per answer.
  4. 04  Citation share = a brand's mentions ÷ all 1,103 mentions. Report per-engine, first-named, and consensus.

From the prompt bank

38 prompts across purchase, emergency, comparison, research, reviews & local intent

  • Who are the best HVAC companies in the United States?· Purchase
  • My air conditioner stopped working. What's the best HVAC repair service to call?· Emergency
  • Who should I hire to replace my furnace? Recommend a reputable HVAC contractor.· Purchase
  • Best heating and cooling companies for a new central AC system installation?· Purchase
  • Which HVAC company has the best reputation and reviews nationally?· Reviews
  • I need emergency AC repair today. Which large HVAC service should I contact?· Emergency
  • Compare the top national HVAC service companies. Which is most reliable?· Comparison
  • Who installs heat pumps? Recommend a trustworthy HVAC company.· Purchase
  • What are the most trusted heating and air conditioning contractors in America?· Reviews

On honesty: we measure the default answer each engine gives through its standard API — the recommendation a real user receives. Parametric engines answer from training memory; Perplexity grounds on the live web. We report which is which and never present an estimate as a measurement. 379 of 380 queries returned; the rest are excluded, not imputed.

The web sources behind the grounded answers

The domains Perplexity actually cited while answering — the corroborating surface that propagates AI trust in this category.

reddit.com54yelp.com42youtube.com27facebook.com17bbb.org14servicetitan.com14angi.com13rotorooter.com12hookagency.com11fieldpulse.com10thumbtack.com9housecallpro.com9
Now measure yourself

The Index shows the field.
The Scanner shows you.

Run the same class of buyer-intent prompts against your business — across all five engines, free, in under a minute. See your AI Visibility Score, the competitors named instead of you, and the specific reasons why.

No signup wall · five engines · real evidence

FAQ

About the Index

What is the HVAC & Plumbing AI Visibility Index?+

It is a first-party measurement of which HVAC, plumbing, and contractor-software brands the five major AI engines — ChatGPT, Claude, Gemini, Grok, and Perplexity — actually name when consumers and contractors ask them for recommendations. Deep AI Visibility ran 38 open-ended buyer- and trade-intent prompts across all five engines and tallied every brand each one volunteered. No brand is ever inserted into a prompt.

How is 'AI citation share' calculated?+

Each of the 38 prompts was asked to all five engines, 2 times each (380 planned queries). Every brand named in a reply is counted once per answer. A brand's citation share is its mention count divided by all 1,103 brand mentions in the study. We also report a per-engine breakdown, how often each brand is named first, and how many of the five engines name it at all.

Why does this Index differ from other AI visibility rankings?+

Most published rankings estimate 'citation share' from trade-press coverage or opaque proprietary blends. This Index measures the answer a real buyer receives: the default recommendation each engine gives to an ordinary question, captured through each vendor's standard API. That is why franchised consumer plumbers rank above enterprise software here — it reflects consumer intent, not press volume.

Can I see where my own brand ranks?+

Yes. The Index measures the field; the live AI Visibility Scanner measures you. Run the same class of buyer-intent prompts against your business across all five engines, free, in under a minute — and see your score, the competitors named instead of you, and the specific reasons why.

How often is the Index updated?+

This edition was measured in July 2026. The Index is a living measurement — because it is generated by a reproducible harness rather than hand-assembled, it can be re-run on demand as models change, and it will be refreshed as the AI answer layer shifts.