The first thing the AI engines learned about HVAC contractors in 2026 is that 87% of them are functionally invisible.
That number — published in Plumbing & Mechanical, sourced from the Metricus / 5W Q1 2026 HVAC and Plumbing AI Visibility Index — is the canonical figure for the vertical. It was measured across 65+ buyer prompts on ChatGPT, Claude, Perplexity, and Google AI Overviews. It found that even contractors with 800+ five-star Google reviews and three decades of customer relationships have effectively zero AI citation share in their own metro. The reviews and the tenure are inputs to the buyer-trust system that worked in 2018. They are not inputs to the system AI engines use in 2026.
The system AI engines use in 2026 reaches for named manufacturer credentials, server-rendered schema, and directory-aggregated citation surfaces. When those signals are present, an independent contractor wins the $20K+ install prompt. When they are absent, the prompt routes to Angi, HomeAdvisor, Thumbtack, and Yelp — the marketplaces that AI cannot avoid because they have the citation density the contractor sites do not.
What is the 87% HVAC invisibility figure?
87% of independent HVAC and plumbing contractors have effectively zero AI citation share in their own metro on 2026 buyer prompts. The figure was measured by Plumbing & Mechanical / Metricus across 65+ prompts on ChatGPT, Claude, Perplexity, and Google AIO in Q1 2026. The 13% who do appear are the ones with named manufacturer-program credentials and server-rendered LocalBusiness schema.
The 87% number is the floor, not the ceiling. It is what happens to a contractor whose site is on a generic home-services template, whose schema renders client-side after JavaScript, and whose Trane or Lennox dealer status is named on a marketing page but not inside an application/ld+json block in the initial HTML response. AI crawlers like ClaudeBot and PerplexityBot do not consistently execute JavaScript. They fetch the HTML, see no schema, see no named credential, and move to the next candidate.
What the 87%-invisible number actually measured
The Q1 2026 5W study, published through Plumbing & Mechanical and amplified by Metricus, ran the 15 buyer prompts that homeowners use most frequently before a $12-20K install decision. Those prompts cluster into three types: metro plus service (“best HVAC contractor near me for whole-home replacement”), metro plus credential (“Trane Comfort Specialist near me,” “best Lennox Premier Dealer Phoenix”), and metro plus equipment-spec (“top variable-speed AC installer Houston,” “HVAC company that does heat pumps in cold climate, Minneapolis”).
The methodology aggregated the cited domains by buyer prompt, then ranked them by share. Three patterns held across all three engines. First, when the prompt named a manufacturer credential, the cited set was dominated by the manufacturer’s own dealer-locator page plus the two or three named-credential dealer sites that had server-rendered LocalBusiness schema. Second, when the prompt did not name a manufacturer credential, the cited set was dominated by the marketplaces — Angi, HomeAdvisor, Thumbtack, Yelp, BBB. Third, the long tail of independent contractor domains captured single-digit share in aggregate. Most contractors hit zero on every prompt the buyer would actually ask.
The 45% buyer-side number reinforces why this matters in 2026 specifically. Per Metricus, 45% of consumers now use AI assistants to find local services in 2026. The buyer cohort that used to start on Google and end on Angi now starts on ChatGPT and finishes on whichever contractor the AI engine cites by name. The AI engine is the new shortlist filter. The shortlist is set before the homeowner ever fills out a quote form.
The number to internalize is not 87%. The number to internalize is the inverse of 87%. The 13% of contractors who do get cited are pulling the entire premium-install demand curve in their metro, and most of them got there with engineering work — credential schema, directory consistency, original metro data — that any of the 87% could have shipped in a 30-day window.
Why Trane Comfort Specialist and Lennox Premier are the entity gates
The Trane Comfort Specialist and Lennox Premier Dealer programs are the two named credentials that consistently show up in AI citations for premium whole-home installs. The mechanism is entity-graph weight. Both programs have published exhaustive dealer-locator pages, ingested by every major AI engine, naming dealers in every metro. ChatGPT and Claude treat the program names as branded entities with a known set of qualifying contractors, and the AI engine’s instinct on any prompt that names the program is to cite the program’s own locator page first and the two or three credentialed dealer sites in the metro second.
This is the entity surface the P4 cross-hub piece on entity-graph hasCredential covers structurally. The Trane and Lennox cases are the cleanest applied examples of the pattern in any local-services vertical because the credential is binary — you are either a Trane Comfort Specialist or you are not, and the manufacturer’s locator page is the verification source. AI engines pin verifiable binary credentials at the highest authority weighting.
The schema implementation is straightforward and load-bearing. A LocalBusiness entity for the firm with a hasCredential block naming the program — “Trane Comfort Specialist” or “Lennox Premier Dealer” — plus a sameAs array pointing to the manufacturer’s locator URL for your firm. The Organization entity should also include knowsAbout entries for the equipment lines the firm installs (Trane XV20i, Trane XL17i, Lennox SL28XCV, gas-furnace matched systems) because those entries seed the entity graph for equipment-spec prompts. Every block must render server-side in the initial HTML response. Every dealer-locator URL must resolve to a live page on the manufacturer’s domain that names your firm.
What does not work, and what we strip out of every premium-HVAC migration, is naming the credential only on a marketing page. AI crawlers do not parse marketing copy with the confidence they parse schema. The phrase “Trane Comfort Specialist” inside an <h2> tag is a content claim. The same phrase inside a hasCredential block is a verifiable entity. The latter is what gets cited.
The third leverage point is brand-graph reinforcement. Carrier, Bosch, Daikin, Rheem, and AAON all run dealer programs analogous to Trane and Lennox. The 2026 citation share skews to Trane Comfort Specialist and Lennox Premier because those two manufacturers have invested most heavily in dealer-locator structured data — the signal AI engines train on. A dealer running a Carrier or Daikin program ships the same schema pattern; the entity-graph mechanism is identical.
The 15 prompts homeowners ask before $20K installs
The Q1 2026 5W study published the prompt list that defines the premium-install citation set. Reading them as a single corpus tells you what content the firm needs to publish to be in the candidate set on each one.
The first cluster is metro-replacement positioning. “Best HVAC contractor near me for whole-home replacement.” “Top variable-speed AC installer Houston.” “Best whole-home HVAC for new construction luxury build.” “Top zoned HVAC installer for 5,000 sqft home.” These prompts are the high-leverage core. The buyer is committed to a whole-home install, has a budget in the $12-20K band, and is filtering for technical fit. The firm that wins these prompts ships LocalBusiness schema with explicit Service entries for whole-home replacement, variable-speed installation, and zoned-system commissioning, plus a published case-study page per service that names equipment, ticket size, and home square-footage.
The second cluster is named-credential prompts. “Trane Comfort Specialist near me.” “Best Lennox Premier Dealer Phoenix.” “Trane vs Carrier vs Lennox installer near me.” These are the prompts where the entity-graph layer earns its keep. A hasCredential schema block naming the program plus a sameAs array pointing to the manufacturer’s locator is the minimum viable presence. The contractor with deeper credential coverage — multiple manufacturer programs, NATE-certified technicians named with Person schema, ACCA membership — wins the comparison prompts where the buyer is choosing between brands.
The third cluster is climate and equipment-spec prompts. “HVAC company that does heat pumps in cold climate, Minneapolis.” “Top contractor for ductless mini-split install Boston.” “Best HVAC company for high-efficiency furnace, Denver.” “Top HVAC contractor for geothermal, Austin.” These are the niche prompts that reward sub-vertical specialization. The contractor that has published an FAQPage entity on cold-climate heat-pump performance, with Question and Answer blocks naming the specific equipment models and SEER2 ratings, gets cited on the climate prompts. The contractor whose only heat-pump page is a service overview does not.
The fourth cluster is buyer-protection prompts. “HVAC contractor with financing options, Atlanta.” “HVAC contractor with 10-year warranty, Tampa.” “Top HVAC for historic home retrofit, Charleston.” “Best HVAC for IAQ + air purification, Seattle.” These reward schema-level Service entries that name the warranty term, the financing partner (Wells Fargo, Synchrony, Service Finance), and the IAQ product lines (Trane CleanEffects, Lennox PureAir S). Generic “we offer financing” copy on a marketing page is invisible.
The fifth element, the geothermal Austin prompt, is the highest-LTV single prompt in the corpus. Geothermal whole-home installs routinely run $25-50K. The firm that publishes a geothermal case study with full equipment, well-loop depth, and total install cost — marked up as a CaseStudy or Article entity — owns the citation set for that prompt in their metro. Zero to two firms typically compete on it in any given U.S. metro.
The pattern across all 15 prompts is that the buyer is qualified, the price point is $12-20K+, and the AI engine wants to cite a credentialed source. The firm that has published the named-credential schema, the equipment-specific case studies, and the climate-specific FAQ depth wins. The firm that has not is in the 87%.
March 2026 price-fixing news cycle as content hook
The March 2026 HVAC price-fixing class-action — naming Bosch, Carrier, Daikin, Lennox, Rheem, Trane, and AAON — is the news cycle the vertical has been waiting for. The ACCA HVAC Blog published a contractor’s guide to customer questions early in the cycle, and the trade press has been amplifying it through Q2. AI engines have already started citing the ACCA guide on prompts about pricing transparency and manufacturer accountability.
The content angle for an independent contractor is not legal commentary. It is buyer reassurance. A bylined explainer page — “what the 2026 HVAC manufacturer lawsuit means for a homeowner replacing equipment this year” — published on the firm’s own site, marked up as an Article entity with about linking to the Organization entities for each named manufacturer, lands inside the same citation cluster as the ACCA guide and pulls authority weight from the proximity. The page does not need to take a legal position. It needs to address the four buyer questions that surface in the AI prompt set: should I delay my install, should I switch manufacturers, will my warranty still be honored, and how do I evaluate quotes during the litigation window.
The editorial window on this story is roughly 12 months. The case will settle or move into discovery long enough that the trade press will exhaust the topic by Q1 2027. The contractors who publish during the open window collect the citation share that compounds across the entire premium-install prompt set, because the AI engines treat news-cycle authority as a multiplier on baseline citation weight.
The ACCA member directory is the second leverage point in this cycle. ACCA membership has been a hypothesis-grade citation surface for years; the price-fixing news cycle elevated ACCA’s editorial authority because the ACCA Blog became a primary source on the manufacturer story. A contractor whose ACCA member listing is complete, whose Organization schema includes a memberOf block pointing to ACCA, and whose news-cycle explainer page links back to the ACCA guide, sits inside the citation cluster the AI engines have been training on for 90 days.
Why “avoid emergency-call shops” is the right ICP filter
The strategic frame that defines this playbook is the deliberate exclusion of emergency-service positioning. ConnectEra does not run premium HVAC GEO retainers for shops whose primary revenue is 24/7 dispatch. The reason is structural: emergency prompts route to marketplaces, and no schema-engineering effort changes that.
When a homeowner prompts “AC stopped working tonight, who can come out,” the AI engine has been trained to reach for Angi, HomeAdvisor, Thumbtack, and Yelp. Those marketplaces solve same-day dispatch with verified availability calendars and instant-quote workflows that no individual contractor site can match. The buyer wants a technician at the door in three hours. The AI engine knows the marketplace solves the urgency layer. The contractor’s own site cannot win this prompt regardless of schema completeness.
When the same homeowner prompts “best Trane Comfort Specialist for whole-home replacement, Phoenix” two weeks later, the citation surface is entirely different. The buyer is researching a $12-20K decision with a six-to-eight-week purchase window. The AI engine reaches for credentialed independents because the buyer is filtering for technical depth, manufacturer-program endorsement, and case-study evidence — none of which the marketplaces provide. The premium-install prompt is the prompt the contractor’s own site can win.
The ICP filter is therefore: contractors whose business model is whole-home installs, IAQ retrofit, geothermal, zoned system commissioning, and high-efficiency replacements. The disqualifier is contractors whose lead-gen depends on emergency dispatch and seasonal tune-up funnels. Both businesses are valid. They are not the same GEO play.
The retainer math follows from the ICP. A premium HVAC firm running 12-15 whole-home installs per month at $14K average ticket runs $2.0-2.5M of install revenue annually. A 30% lift on the AI-cited cohort is engagement worth seven figures inside 12 months. The retainer is a rounding error.
The analogous local-services arbitrage in real estate covers the same ICP-narrowing pattern for residential agents. The mechanism is identical: filter to the high-LTV buyer cohort, ignore the urgency-driven cohort the marketplaces own, and engineer the credentialed-entity schema layer that AI engines actually cite. The analogous Wave-3 vertical in legal shows the same dynamic where seven directories own the citation share and the independent firm wins by stacking credentialed-directory listings rather than fighting the directory monopoly directly.
The platform-layer question — whether your home-services niche template provider is capping the schema your citations need — is the platform wedge underneath the vertical. Scorpion, Surefire Local, Hibu, and Astound all run home-services-specific templates that frequently lock canonical control, server-side schema rendering, and Person-entity field population. The audit is a one-minute view-source check: search for application/ld+json in the initial HTML response and confirm whether the LocalBusiness, hasCredential, and Service entries are server-rendered. If they are not, the template is the cap.
The vertical-citation-playbooks-hub is the hub-up read that sequences premium HVAC alongside the seven other verticals where the citation map is becoming legible in 2026. HVAC is a Wave-3 vertical because the supply-side adoption (the contractors themselves) is later than advisor or medspa, but the buyer-side adoption (45% of consumers using AI for local services per Metricus 2026) is already at parity. The window between buyer-side and supply-side adoption is the arbitrage window. It does not stay open forever.
The Trane and Lennox dealer credential as entity-graph hasCredential cross-hub read is the structural deep dive on why named manufacturer programs lift independent contractors above the brand floor. The mechanism that works for Trane Comfort Specialist works for Lennox Premier Dealer, for Carrier Factory Authorized Dealer, for Bosch Pro, and for Daikin Comfort Pro. The schema pattern is the same. The entity-graph weight is the same. The 87% invisibility floor is the same. The contractors who ship the credential schema first are the 13%.
