Why AI recommends other contractors instead of you

The uncomfortable data on the trades and AI

We didn't theorize this. We ran the same free check we sell on 7 real U.S. home-services companies — plumbers, HVAC shops, and a roofer, across Austin, Dallas, Tampa, and Charlotte — asking each company's buyers' real questions ("best plumber in [city]," "who should I call for AC repair in [area]," "is [company] any good?") across ChatGPT and Perplexity. Here is what came back:

• Average AI Visibility Score: 30 / 100. The best company in the set scored 53; the most common score was 17. Not one company cleared 60.
• Nearly half were pinned at the 17/100 floor. Three of the seven landed at exactly 17 — they surface occasionally but are out-named everywhere they appear.
• Named in only half of answers. Across all 42 AI answers we collected, the subject company appeared in just 21 of them — a literal coin-flip to be in the reply when a customer asks AI about your market.
• A single rival often owns the whole metro. In our two Charlotte HVAC checks, the same regional competitor was the top-named company in both — and across all three plumber audits (in three different cities) one national plumbing brand kept surfacing as the recommendation. The engines aren't undecided; they've already picked someone.
• The field is fragmented for everyone else. Across these seven audits the engines named more than 25 different competing companies by name — specific, confident recommendations, all going to someone else. Ask twice and the shortlist often reshuffles, which is exactly why a single lucky screenshot proves nothing.

And the reach of a big brand doesn't save you: in a separate check, a national HVAC franchise in Phoenix scored 17/100 and a national plumbing franchise in Chicago went unnamed on ChatGPT entirely, which listed seven local independents instead. If billion-dollar brands lose the "best near me" answer to the local shop, the answer in your city is wide open — and right now it's being decided without you in the room.

Why the trades are a distinct AI-visibility problem

Home services sit in the same local-search reality as any "near me" category, but a few things make them their own case:

• The intent is urgent and winner-take-one. "Emergency plumber near me," "AC not working who to call" — the customer is in a hurry and frequently calls the first name the assistant gives. There's no page two of an AI answer. If it names three companies and you're not one, you never get the call.
• The jobs are big, so each miss is expensive. A new HVAC system is $5,000–15,000; a re-pipe, a panel upgrade, a sewer line, a roof — thousands each. A single AI-referred customer lost to a competitor isn't a missed click; it's a five-figure job booked by someone else.
• Directories and review consensus decide it. For service categories, AI engines weight third-party corroboration — Google, Yelp, Angi, HomeAdvisor, Thumbtack, Nextdoor, "best plumber in [city]" listicles — far above what your own site says. If the companies named ahead of you own those pages, you lose the answer.
• Your Google rank doesn't protect you. Plenty of contractors rank on page one of Google and still score badly with AI. The engines don't read your ad spend or your rank — they synthesize from the sources above, a surface most home-services owners have never looked at.

The upside hiding in the same data

Here's the part that should make a well-run local shop lean in: because AI favors local independents over national brands in these answers, this is a winnable category. The franchises with the biggest budgets are the ones getting left out — which means the "best {trade} in {city}" slot is up for grabs by whoever the engines can most easily corroborate. That's not the marketing money; it's the directory, review, and roundup footprint the engines actually read. It's fixable, and most of your competitors aren't even watching it.

What "loses to a competitor" actually means for your business

When an AI answer names three or four contractors and yours isn't among them, that's not a ranking you slipped a spot in — it's a job that went to someone else before the customer ever typed your name into Google. The customer asked an assistant for a recommendation, got a shortlist, and started dialing. Increasingly people treat that shortlist the way they used to treat the top of the search results: as the answer. In a trade where a single new system or repair is worth thousands, being absent from the shortlist for your own city and trade isn't a marketing nicety — it's booked jobs going to the company the AI trusts instead.

The five things that actually move AI recommendations for the trades

In priority order — the work, not the theory:

1. 1Measure your real frequency first — don't guess from one answer. Ask ChatGPT and Perplexity your customers' real questions several times and count how often you appear, and who's named instead. A single answer is noise (remember: 25+ different rivals surfaced across our audits). The honest baseline is a frequency, which is exactly what a check gives you.
2. 2Get into the third-party pages the AI already trusts. Being named in a "best plumbers / HVAC / roofers in [city]" roundup, the local service directories (Angi, HomeAdvisor, Thumbtack, Nextdoor), and cross-platform review pages does far more than any claim on your own site — because those are the pages the engine retrieves and quotes.
3. 3Build consistent reviews across more than one platform. Cross-platform review presence — Google and Yelp and the trade directories, not one perfect profile — is what these engines read as consensus for local service categories. The companies that win are corroborated in several places.
4. 4Publish answer-shaped, structured pages on your own site. Service and service-area pages with clear headings, plain factual statements, and LocalBusiness structured data are easier for a retriever to extract and cite. Keep the facts (company name, trades, service area, phone) identical everywhere.
5. 5Keep one consistent story across every listing. Same company name, same trades, same service area across your site, directories, and reviews. Contradictions make the engine hedge or pick a source it trusts more — and hand the answer to a rival with a cleaner identity.

Why the shop down the road gets named and you don't

The skills overlap with SEO, but the target moved: from "rank on the results page" to "be the company named inside the answer."

Why you can't fix what you haven't measured

Here's the trap that wastes the most effort: a single AI answer is not a measurement. These engines are nondeterministic — ask the same "best plumber in [city]" question twice and you frequently get a different set of companies, because the live search returns slightly different pages each time (we have the run-to-run data). That's why seven companies produced more than 25 distinct rivals in the answers. So one screenshot with your name is false comfort, and one with a competitor is false panic. The only honest baseline is a frequency: across many runs, how often does the AI name you — and which companies does it name when it doesn't?

That's the measurement AskedAbout exists to give you. Start with a free 60-second check — your company name, your trade, and your city — and you'll see your AI Visibility Score (0–100), how often ChatGPT and Perplexity mention you, and which competing companies they recommend in your place. Then the $79 one-time Full Audit tells you which fixes matter most: 25 buyer-intent questions across ChatGPT, Perplexity, Gemini, and Claude, sampled 3 times each — 300 answers in all — reporting your mention rate per engine, share-of-voice against the companies taking your slot, and the exact third-party sources the engines cite when they build a shortlist without you. One payment, no subscription.

A realistic timeline (so you don't quit a month in)

Moving your AI recommendation rate is a compounding effort, not a switch. A genuinely new source page — a fresh roundup mention, a corrected directory listing, a wave of new reviews — can start showing up in answers within weeks once search indexes it, but building enough corroborated presence to be named reliably takes months. The contractors that win treat it like SEO: get an honest baseline now, change the inputs deliberately (the directories, the reviews, the consistent facts), and re-measure on a cadence to see what actually moved. The ones that lose either never measure, or change one thing, screenshot it once, and give up while the shop down the road keeps compounding.

Where AskedAbout fits

AskedAbout doesn't do the marketing for you — it tells you, specifically and in priority order, what to fix and whether it's working, across the four engines your future customers actually ask. The free check gives you a baseline in 60 seconds. The $79 audit turns "why doesn't AI recommend us?" into a concrete, ranked to-do list tied to your real mention frequency per engine and the actual sources each one cites about your trade in your city — so you know which directories, review sites, and roundups to get into first. You hand the fix plan to whoever runs your marketing, do the work above, and re-check to confirm the shortlist changed. Measure, fix, re-measure — with an honest number instead of a screenshot.

Methodology

Benchmark figures are from 7 AskedAbout free checks of U.S. home-services companies (plumbing, HVAC, and roofing) across Austin, Dallas, Tampa, and Charlotte, run June–July 2026. Each free check submits three buyer-style questions ("best {trade} in [city]," "who should I call for [service] in [area]," "is [company] a good choice?") to two answer engines — ChatGPT and Perplexity — for six AI answers per company (42 completed answers in all across the set). Each answer is parsed for the subject company and competitor mentions by a judge model, and the AI Visibility Score (0–100) is computed over completed answers. "Named in half of answers" is 21 of 42 completed answers; "beaten by a competitor" means a rival named more often than the subject. The "more than 25 distinct competing companies" figure is the count of unique rival company names across the audits. The national-franchise figures (Phoenix HVAC 17/100, Chicago plumbing unnamed on ChatGPT) are separate free checks of two recognizable national franchises. The paid $79 Full Audit widens this to 25 questions across four engines (adds Gemini and Claude), sampled three times each (300 scheduled answers). We query official engine APIs, which approximate but do not perfectly match the consumer apps — we say so. Subjects anonymized; aggregate benchmark data only, never customer-identifiable; figures refresh as the audit corpus grows.

See also

• Built for your business: AI visibility for home services — the vertical page with the fix checklist.
• The category and the other engines: why AI recommends competitors instead of you · how to show up in ChatGPT recommendations · how to get recommended by Perplexity.
• The data: the local-business AI visibility benchmark · AI answers change every time you ask · AI visibility tools compared.
• Other local verticals: why AI ignores your med spa · why AI ignores your law firm.

FAQ