We ran 300 real buying questions across five AI engines and recorded every brand that came up. Three names dominate. Most of the industry is close to invisible. Here is exactly who gets named, and who doesn't.
Each score is the share of the 290 scored answers in which a brand was named at all. A brand counts once per answer, whether it's mentioned once or three times. Ranked by visibility.
Of the 259 distinct brands named across the study, 230 appeared in fewer than 7% of answers. The market for AI recommendation is not crowded at the top — it is nearly empty everywhere below third place.
Jobber, Housecall Pro and ServiceTitan are named in 70%, 63% and 52% of answers. The next brand — FieldPulse — sits at 23%, barely a third of the leader's rate. For a funded contender being named a quarter as often as Jobber, that gap isn't a ranking. It's a revenue leak nobody is measuring.
FieldPulse is named in 22 of ChatGPT's answers but only 5 of Google AI Overview's. ServiceTitan swings from 34 mentions on Claude to 21 on Google. A brand can be doing fine on one engine and be effectively absent on another — and no single dashboard shows you which. Visibility has to be measured engine by engine, or it isn't measured at all.
On questions like "how do I stop losing track of jobs?" or "how do I get paid faster as a contractor?", most engines recommended no specific product at all — they answered with generic advice. That's an uncontested lane: the first brand to publish content the engines will cite on these problem-first queries owns a space every competitor has ignored.
Mentions per engine, out of 58 scored answers each. Read across each row to see how unevenly the engines treat the same brand — the green cells are where a brand is strong, the red where it falls away.
| Brand | ChatGPT | Perplexity | Gemini | Claude | Google AI |
|---|---|---|---|---|---|
| Jobber | 35 | 37 | 39 | 38 | 34 |
| Housecall Pro | 35 | 32 | 34 | 35 | 30 |
| ServiceTitan | 26 | 32 | 25 | 34 | 21 |
| FieldPulse | 22 | 13 | 11 | 9 | 5 |
| Service Fusion | 14 | 9 | 9 | 20 | 8 |
| Workiz | 12 | 12 | 9 | 10 | 12 |
| FieldEdge | 10 | 7 | 3 | 15 | 4 |
| IFS | 6 | 3 | 9 | 9 | 1 |
Jobber is the only brand named consistently across all five engines. Everyone else has at least one engine where they quietly disappear.
AI engines don't invent their recommendations — they pull from pages they trust. Across the study, the same domains appeared again and again as the sources behind the answers. These are the pages a brand has to exist on to be recommended.
The pattern is consistent: third-party review directories and community threads, not brands' own websites, are what the engines cite. A brand can have a beautiful site and still be invisible to AI — because AI isn't reading the site. It's reading Capterra, G2 and Reddit.
Between 5 and 8 August 2026, we ran a panel of 60 buyer-intent prompts across ChatGPT, Perplexity, Gemini, Claude and Google's AI Overview — 300 queries in total. Prompts spanned six intents: category discovery, feature-led, comparison, trade-specific, problem-first, and pricing. Each prompt was run in a fresh, logged-out or unpersonalised session to avoid history effects. For every answer we recorded each brand named and each source domain cited.
Two of the 60 prompts were set aside from scoring because every engine read them as a different software category — one surfaced phone-repair tools, the other online-booking apps. That leaves 290 scored answers. A brand's visibility score is the share of those 290 answers in which it appeared, counted once per answer. Where an engine named no brand at all — common on problem-first questions — the answer still counts, because "AI recommended no one" is itself a finding.
AI answers vary between runs; this index is a dated snapshot, not a permanent ranking. That is precisely why it will be re-run monthly — the movement over time is the point.
This index shows the category at a glance. A Citation Desk audit goes deep on one brand: every prompt you're missing from, which competitors are named instead, the specific pages the engines cite in your space, and a 30-day plan to get into the answers. Five days, one fixed fee.
Request your brand's breakdown →