Why Does AI Recommend Roofers Less Often Than HVAC?

ChatGPT, Gemini and Perplexity named roofers less than half as often as HVAC companies when asked who to hire: 18.3% of answers against 49.1% in the Cheers AI visibility grader, March to October 2026. The gap persisted when we matched review bands and question types. Set a baseline for each trade, then measure the work against comparable businesses in the same market.

Last verified: November 1, 2026.

This historical comparison covers 94 roofing and 123 HVAC companies that ran our free grader. It uses developer APIs with web search enabled. Use the March-to-October findings to understand trade differences and a fresh matched sample to set today's targets.

Far more people now ask AI to recommend a local business: BrightLocal's 2026 survey found 45% of US adults had used AI for local business recommendations in the past year, up from 6% in 2025.

Roofers came last of the trades we could measure

Of the five home-service trades with at least 10 graded businesses, roofing came last. Plumbers (26.1% of answers), pest control companies (23.8%) and restoration companies (28.3%) all sat closer to roofers than to HVAC.

The gap shows up on all three engines. Asked one question per request, ChatGPT named roofers in 16.9% of buyer answers and HVAC companies in 41.2%, while Gemini came in at 23.9% against 53.9% and Perplexity at 13.8% against 45.3%.

The engines also left 39 of the 94 roofers (41.5%) out of all 12 of their answers, against 17 of 123 HVAC companies (13.8%), and named roofers first in 9.7% of answers against 22.4% for HVAC. Missing from a shortlist isn't the same as being unknown, though; our study of businesses AI knows but won't recommend measures that gap across the whole grader.

Five explanations we tested

Do roofers just have fewer reviews?

The median roofer had 184 Google reviews and the median HVAC company 1,332, but comparing companies with similar counts doesn't let roofers catch up. Under 200 reviews, the engines named roofers in 12.2% of answers and HVAC companies in 42.7%, a 30.5-point gap, and roofers stayed under half the HVAC rate on all three question types we could match there.

The gap shrank further up, to 17.5% against 35.6% at 200 to 499 reviews and 32.4% against 39.1% at 500 to 999. That last gap is under 7 points, but the band holds too few businesses to call it a difference, and fewer than 10 roofers had 1,000 or more reviews.

Within roofing, more reviews went with more naming, but those roofers also got fewer of the questions that named a neighborhood (76.5% of answers under 200 reviews, 61.8% at 500 to 999), so we can't hand reviews all the credit.

HVAC naming didn't climb with review count until the four figures, where it reached 52.9% at 1,000 to 2,999 reviews and 59.7% at 3,000 or more. The trade-by-trade medians are in how many Google reviews you need for AI recommendations.

Pest control points the same way: its companies had a median of 1,075 Google reviews, far nearer HVAC than roofers, yet a naming rate closer to roofers'.

Did roofers get harder questions?

They got different ones. Roofers drew far more of the questions that named a neighborhood, 72.1% of their answers against 33.9% for HVAC, and those questions named the graded business less often in both trades. We can't tell whether the wording or the businesses drove that, so we matched on question type instead.

Matching didn't close the gap. On each of the five question types that at least 10 businesses in both trades received, roofers trailed by 23 to 29 points, and weighted to the same mix of city-wide and neighborhood questions they came out at 22.6% against 46.7% for HVAC.

Two of those types put roofers a little over half the HVAC rate, as does weighting both trades to HVAC's own question mix. Roofers trail in every comparison we ran; how wide the gap looks depends on which one you pick.

Was it when they were graded?

Most roofers were graded between March and early July, before our September 24 question change, and naming across the whole grader rose after it, from 38.0% of buyer answers to 49.4%. That overall change cannot rule out time or method effects within different trade cohorts. Fewer than 10 roofers were graded after the change, too few for a reliable within-trade comparison.

ChatGPT's combined request was the other method change in the window. Leaving those answers out gives 18.5% for roofers against 47.6% for HVAC, and Gemini and Perplexity on their own give 18.9% against 49.6%.

Are roofers' websites thinner?

The crawl doesn't support it. On November 1 we read the homepages of 85 graded roofers and 108 graded HVAC companies: the roofers' pages ran a median of 1,621 words against 1,478, LocalBusiness markup showed up on 67.1% against 68.5%, and roofers were actually ahead on FAQ markup (29.4% against 13.9%).

The buying path is where the homepages split. Only 8.2% of roofer homepages embedded booking software, against 58.3% of HVAC homepages, and 54.1% linked to financing against 75.0%. That doesn't prove a booking widget earns a recommendation, and since the crawl ran after most grades, read it as a picture of the sites today.

Answers cited a roofer's own site in 10.9% of cases against 33.6% for HVAC, but that mostly follows from naming: once an engine named a roofer, it cited the roofer's own website in 58.3% of those answers, close to HVAC's 64.7%. Our September roofing playbook covers which directories and lists ChatGPT reads for roofing questions and what each branch page should say.

Do manufacturer finders and the BBB crowd roofers out?

Not the manufacturer pages, and the BBB only slightly. The domain gaf.com showed up in 16.0% of roofer answers and no HVAC answers, and manufacturer pages of any brand in 16.8% against 10.8%. Roofers were named in 24.9% of the answers that cited one, against 16.9% of the rest, which fits what GAF's contractor finder is: a list of certified roofers near an address with a star rating, review count, certification and phone number.

The BBB showed up in 37.9% of roofer answers against 21.1% of HVAC answers, and roofers were named a little less often when it did (16.6% against 19.3%), small next to the overall gap.

These are associations, not causes, and we didn't check which roofers hold a certification. Our articles on whether BBB accreditation is worth it for AI search and on which sites AI cites for each home-service trade go further into both sources.

Our interpretation of the trade differences

Hypothesis one: roofing is a job people shop around for, and the answers read that way. In the Cheers Market Baseline, our separate research panel, Gemini told people asking about roofing to get several quotes in 44.8% of its answers from October 9 to 31, 2026, against 15.9% for HVAC, counting general advice. In October, Google AI Mode linked a booking or scheduling page in 1.7% of its answers to US roofing questions, against 30.6% for HVAC.

AI Mode rarely linked one for pest control (4.2%) or lawn care (5.4%) either, though, so a missing booking link isn't a roofing quirk and may have little to do with shopping around. Our comparison of ChatGPT phone numbers and Google AI Mode booking links goes further into where those links come from.

Hypothesis two: storm work comes with fraud warnings. The National Insurance Crime Bureau said in May 2026 that reported contractor fraud rose 38% from 2023 to 2025, and it lists manufactured roof damage among the most common schemes. If engines lean on warnings like that, their roofing answers would steer people to verification pages before naming anyone, which would fit the BBB's bigger share of roofer answers. We didn't test this.

Hypothesis three: the engines agree less about who leads a roofing market. Across a roofer's 12 answers, all three engines named 1.34 other companies on average, against 2.07 for an HVAC company, and roofing answers listed 6.1 businesses where HVAC answers listed 6.9. Fewer slots per answer, and less agreement on who fills them, would leave any one roofer fewer chances to be named.

Roofing answers also lean hard on warranties and on the BBB; our pieces on the trust signals AI repeats about contractors and on BBB accreditation carry those numbers.

Benchmark each brand against its own trade

If you run several brands, use Figure 1 to understand the historical sample, then collect a matched, current sample before setting targets. One portfolio target will flag every roofing brand and flatter every HVAC brand. Compare each brand with its own trade and review band, and measure it the same way every time, so the change between checks is what you read. A single HVAC company can check its own number against its band in Figure 1, too.

Don't call a roofing brand's low AI share a website failure until you've compared it with other roofers; the website checks above didn't point there. If you also run HVAC brands, the HVAC playbook covers the trade that set the high bar in this study.

What a roofing company can do this month

Start with one location this week, then repeat it for the rest.

• Get a like-for-like number. Run the free AI visibility grader on the location and save the "Named in X of Y AI answers" line it reports; it asks the same three engines this study used. Roofers here averaged about 2 of 12 and HVAC companies about 6 of 12, though a roofer graded today may land higher, since the grader has named businesses more often since the September 24 change. Track your own number month to month, run the same way, and don't hold it up against a search you ran by hand in a signed-in app.
• Check the third-party pages roofing answers cite. Claim and correct your BBB profile first, since the BBB turned up far more in roofer answers than HVAC ones. If you hold a manufacturer certification, open that contractor-finder profile too and check the phone, service area, rating and the certification shown. Show the certifications the branch holds and make the record easy to verify alongside its website and reviews.
• Keep a steady review ask inside Google's rules. Google says more reviews and positive ratings "can help your business's local ranking" and publishes no target number. Its rules for merchants say to offer nothing in return for a review, not to "selectively solicit positive reviews," to set no staff review quotas, not to have staff ask for specific content and not to require or pressure anyone to review while on the premises. Ask every homeowner in the same neutral way and make it easy to respond whenever they choose. A growing record of real customer experience supports the branch's reputation; our roofing review bands show why it belongs in the visibility programme alongside the other work.
• Fix each branch page with the checklists we've already published. The roofing playbook above covers what goes on it, and since roofing answers bring up warranties constantly, spell out which warranty is your workmanship warranty and which is the manufacturer's, the way this warranty guide lays out. The FTC tells storm victims to confirm a contractor's license with the state or county and ask for proof of insurance, so put both where an engine can read them; here's how to publish them. List a license number where your state licenses roofers.
• Keep storm and insurance copy inside your state's rules. Describe how your inspections run and what the homeowner does with their insurer, without promising to negotiate or interpret the claim and without offering to waive or cover a deductible. Texas bars a roofer from acting as a public adjuster on its own jobs or advertising that it will, and from offering to waive or absorb a deductible; Florida bars contractors from offering a deductible waiver or anything of value for a roof inspection or claim, and from interpreting policy provisions or adjusting a claim unless they're licensed as public adjusters. Check your own state's rules before you publish.

If you'd rather have someone run this across your branches, Cheers for roofing manages project review programs, business profiles and content, with visibility tracking by market.

Urgency changes the question too: the emergency-repair shortlist study compares the businesses named for urgent and general requests.

Method

The Cheers AI visibility grader asks ChatGPT, Gemini and Perplexity through their developer APIs, not the consumer apps, with web search on and one shared instruction: answer as a helpful assistant would for a real buyer.

Per business it asks four buyer questions from its category and city ("Who are the best [trade] companies in [city]?"), some naming a neighborhood, each a fresh request with no sign-in or history; we skip a fifth that names the business. ChatGPT and Perplexity have also gotten the city as the asker's approximate location since September 23, and from March 18 to June 8 ChatGPT answered all four in one combined request for about a third of businesses.

The sample is each business's latest complete run, minus test runs: 455 businesses graded March 17 to October 30, 2026, Pacific time, 387 before the September 24 rewording and 68 after. Roofing is 94 businesses (90 contractors plus 4 pairing roofing with solar or siding, excluding two waterproofing firms) and HVAC the 123 whose label includes it; every roofer and all but three HVAC companies (no country recorded) are US businesses. Without the 19 that also do plumbing or electrical, HVAC came out at 49.3%.

ChatGPT's single-request figures cover 172 answers for 43 roofers and 308 for 77 HVAC companies. Every published cell has at least 10 businesses; none is named.

Question types 1 to 5, roofers then HVAC: 15.2% and 43.7% (90 and 84 businesses), 31.5% and 54.5% (32 and 100), 9.7% and 37.7% (62 and 30), 35.0% and 60.1% (37 and 99), 12.4% and 35.5% (62 and 24), over 2,463 answers. The reweighting uses all 2,604.

Review bands use the 92 roofers and 121 HVAC companies with a review count: 50 and 25 under 200, 21 and 11 at 200 to 499, 17 and 16 at 500 to 999, then 38 and 31 HVAC companies. The under-200 gap is about 3.7 per-business standard errors; the 500-to-999 gap is inside one.

Agreement figures match company names exactly, so two spellings count as two.

The Market Baseline, our own panel, asks ChatGPT, Google AI Mode and Gemini 32 home-service questions weekly in 60 US and 40 Canadian metros. Booking links use 480 roofing and 2,640 HVAC US answers from October 2 to 31, 2026; several quotes, 600 and 3,414 Gemini answers from all 100 metros, October 9 to 31. We never chart it alongside, or subtract it from, grader numbers.

The grader sample is self-selected, its questions and models changed during the window, and every comparison is an association, not a proven cause.

How we check our numbers: How to Check an AI Visibility Report.

Sources

• Local Consumer Review Survey 2026, BrightLocal, February 11, 2026. The 45% of US adults using AI for local recommendations, up from 6%; the AI cut is in Nearly Half of Consumers are Asking AI for Business Recommendations, BrightLocal, March 10, 2026.
• Find a residential roofing contractor near you, GAF, undated. What a manufacturer contractor finder lists.
• NICB Warns Americans Contractor Fraud Continues to Rise Nationwide, National Insurance Crime Bureau, May 18, 2026. The 38% rise in reported contractor fraud and manufactured roof damage.
• Tips to improve your local ranking on Google, Google Business Profile Help, undated. Reviews can help local ranking; no target number.
• Prohibited & restricted content, Google Maps User Generated Content Policy Help, undated. Google's rules for asking customers for reviews.
• How To Avoid Scams After Weather Emergencies and Natural Disasters, Federal Trade Commission, September 2025. Checking a contractor's license and insurance.
• Roofing and insurance: Know the law, Texas Department of Insurance, March 24, 2025. Texas limits on public adjusting and deductible offers.
• Florida Statutes 489.147, Prohibited property insurance practices, Florida Senate, 2026. Florida limits on deductible waivers, inducements and claim interpretation.
• Web search, OpenAI API docs, undated and Grounding with Google Search, Google AI for Developers, October 28, 2026. How the grader's API answers use web search.

Sources read November 1, 2026.

Dylan Allen-Arnegård is the CEO & Co-Founder of Cheers, the done-for-you local visibility partner for established home-service companies and PE-backed portfolios.