Digital State of the HVAC Industry 2026104 Contractor Websites Scored

Executive summary

Across the 104 HVAC sites we scored, the average Fervor Score came in at 65.32. Median 65. Standard deviation 11.02. The lowest was 31, the highest 90. On the same scale your kids get report cards on, 12 brands (12%) earned a B or better. And 63 (61%) earned a D or worse. The geography breakdown: 78 brands in the US (or with a non-Canadian web address) and 26 in Canada. Those numbers are the whole story. Everything below just explains them.

Summary statistics

Where the trade sits across the full score range. Quartile and decile cuts on the same 104-site sample as the rest of this page.

Fervor Grade distribution

Category performance

The 100 points break across six categories. Each row below shows the average score against the most a brand could earn in that category, plus how many brands hit a perfect score there (n at max) and how many took a zero (n at zero). The six categories don't add to a clean 100 because some carry more weight than others. Trust & Credibility is worth 22 points. Accessibility is worth 8. We weighted them that way on purpose. A trust gap costs a brand more work than a missing alt tag does.

The trade does best on Mobile Experience, at 73% of what's possible. It does worst on Accessibility, at 44%. The gap between those two is where the easy lift is hiding. So if you're looking at your own grade, that's the category to fix first.

Performance: Google Core Web Vitals

These numbers come from Google's own performance tool, Lighthouse 13.3.0. We ran one browser at a time, in order, so anyone can rerun and get the same answer. Google's thresholds for a mobile page: load the biggest visible piece in 2.5 seconds or less and it's “good.” Over 4 seconds is “poor.” Hold the page layout steady (under 0.1 shift) and it's “good.” Over 0.25 is “poor.” And this matters for two reasons. Buyers leave slow pages, and Google quietly drops them in rankings:

"Page experience signals, including Core Web Vitals, are part of how Google Search ranks pages. Sites that meet Core Web Vitals thresholds have a better user experience and can perform better in Search."

Understanding page experience in Google Search results, Google Search Central (2024)

So when the median HVAC site's biggest visible element takes longer than four seconds to load on a phone, that's hurting both rankings and conversions at the same time. Google drops it down the page, and buyers bail before the form even renders.

Accessibility

We ran axe-core 4.10.2 (the open-source accessibility checker the WCAG community trusts) against every brand's homepage, plus four other pages on each site, on both mobile and desktop. We counted every violation instance across those pages, and the per-brand rates below are the share of brands with at least one. We measured against WCAG 2.1 AA, the same standard the U.S. Department of Justice points to when applying the ADA to commercial websites.

Why overlay widgets don't earn credit

The accessibility overlay tools (AccessiBe, UserWay, AudioEye, EqualWeb, Recite Me, all of them) don't earn a brand any Fervor Grade™ credit, even when installed. Why. An overlay sits on top of a broken page instead of fixing what's underneath. That doesn't stop ADA Title III lawsuits, and courts and federal regulators have said so directly:

"The Ninth Circuit held that Title III of the ADA applies to a business's website and mobile app when those have a nexus to a physical place of public accommodation. In October 2019 the Supreme Court declined review, leaving the Ninth Circuit decision intact and exposing brick-and-mortar businesses nationwide to website-accessibility lawsuits."

Robles v. Domino's Pizza LLC, 913 F.3d 898 (9th Cir.) (2019)

"A blind plaintiff sued Eyebobs over website inaccessibility despite the site running an accessiBe overlay. The case settled by consent decree requiring the U.S. portion of the website be made accessible within 24 months, plus an accessibility coordination team, a published accessibility policy, regular reviews, training, and retention of an accessibility consultant."

Murphy v. Eyebobs LLC, W.D. Pa. (2021)

"The FTC's final order requires accessiBe to pay $1 million and bars the company from representing that its automated product can make any website WCAG-compliant or can ensure continued WCAG compliance over time, unless it has evidence to support such claims."

FTC Approves Final Order Requiring accessiBe to pay $1 Million, Federal Trade Commission (2025)

So we scored each page the way a visitor actually lands on it. Or the way a plaintiff's attorney does. No overlay credit. No “this would work if a screen reader user knew to click the floating widget.”

Why the accessibility numbers matter legally

The accessibility findings above translate directly into legal exposure under two frameworks. Most of the 104 HVAC brands in this sample would fail a compliance review under either.

ADA Title III · United States

The Americans with Disabilities Act applies to public accommodations, and federal courts have repeatedly held that includes commercial websites with a nexus to a brick-and-mortar place of business. The Ninth Circuit's ruling in Robles v. Domino's Pizza (2019), which the U.S. Supreme Court declined to review, is the controlling precedent. The Department of Justice's March 2022 guidance reinforced that public-accommodation websites must be accessible regardless of whether an overlay widget is installed. Critical and serious axe-core violations under WCAG 2.1 AA are exactly the line items that surface in demand letters: missing form labels, low colour contrast, image-alt failures, broken landmark regions. A 64.4% rate of brands shipping at least one critical violation isn't a theoretical defect. It's what plaintiffs' counsel screenshot.

AODA · Ontario, Canada

Canadian brands carry an additional layer. The Accessibility for Ontarians with Disabilities Act requires designated organizations to meet WCAG 2.0 Level AA on public-facing websites. Compliance reporting is mandatory under Ontario Regulation 191/11. Administrative monetary penalties reach $50,000 per day for an individual and $100,000 per day for a corporation, and they accumulate while a violation goes unresolved. The 26 Canadian HVAC brands in this sample sit inside AODA jurisdiction in addition to whatever cross-border ADA exposure they carry from U.S. visitors.

The overlay trap

Third-party accessibility overlay widgets (AccessiBe, UserWay, EqualWeb, AudioEye, Recite Me, and similar) bolt a settings panel on top of an existing site to expose limited accessibility controls. They do not repair the underlying source code. The WebAIM Overlay Fact Sheet, signed by hundreds of accessibility professionals and developers, documents the consensus position: overlays do not satisfy ADA, AODA, or WCAG. Courts have not treated overlay installation as a defense in digital-accessibility cases. Multiple lawsuits have proceeded against overlay-equipped sites.

"A blind plaintiff sued Eyebobs over website inaccessibility despite the site running an accessiBe overlay. The case settled by consent decree requiring the U.S. portion of the website be made accessible within 24 months, plus an accessibility coordination team, a published accessibility policy, regular reviews, training, and retention of an accessibility consultant."

Murphy v. Eyebobs LLC, W.D. Pa. (2021)

"The FTC's final order requires accessiBe to pay $1 million and bars the company from representing that its automated product can make any website WCAG-compliant or can ensure continued WCAG compliance over time, unless it has evidence to support such claims."

FTC Approves Final Order Requiring accessiBe to pay $1 Million, Federal Trade Commission (2025)

So a contractor reading “98%+ ADA compliant” inside an overlay vendor's dashboard is reading marketing copy, not a legal assessment. axe-core measures the rendered DOM at page load, what an assistive-technology user actually encounters. That's the number that matters when a demand letter arrives.

Conversion potential adoption

Five things every contractor site needs in order to turn a visitor into a phone call or a form submission. “Phone in persistent header” means a clickable phone number in the top section of the page, where buyers actually reach for it. A phone number buried in the footer doesn't count here, and it shouldn't. “Inline form on hero” means a form in the first thing visitors see, not below the fold. “Chat widget” here catches both the visible chat box and the named tools we look for, like Intercom, Drift, Tawk, and HubSpot. (That's the broad read. The intake section further down reports the stricter DOM-signature-only count, which runs lower.) “Online scheduling” means a real booking tool (Calendly, Acuity, Square Appointments), plus any “book online” copy that backs it up. And every percentage in this section is out of the same 104 brands we inspected.

How visitors actually reach the trade

Every brand was scanned for the four ways a visitor can hand off to a contractor: phone (tel: links on the page), web form, embedded chat widget, and a click-through CTA. The numbers below come from DOM extraction across each captured page, not self-reporting.

Phone is still the dominant intake mode for home-service trades. A single-location brand with one tel: link buried below the mobile header is leaking calls that would otherwise close. Forms are the inverse story, the heaviest brand in this HVAC sample ships 46 forms across one site, most of them duplicate intakes embedded across location pages, service pages, and footers. Each duplicate is another CRM-mapping liability.

Trust signals: the strategic five

Five trust signals that actually move buyers. We made a point of catching every way they can show up on a page. Take “Google reviews surfaced.” A brand qualifies if their site links to their Google Maps profile, if the page carries a structured review block Google can read, if they run a review widget (NiceJob, BirdEye, Podium), or even if they just write “4.9 stars on Google” somewhere in the copy. Same idea for license numbers. We catch “fully licensed,” state-specific contractor license prefixes (Oregon's CCB, Washington's WACL), and bare license numbers in body copy. The deeper trust signals (manufacturer credentials, financing partners, warranty language, BBB) live further down this page. Keeping them separate so the front-and-center signals don't get diluted by the structural ones.

Positioning & service-mention coverage

Mentions on a website, not specialty. An HVAC brand that lists a given service as one of ten counts here, same as a brand that's built its entire identity around that service. The labels say “mentioned” for a reason. You're seeing what the trade talks about on its websites, not what any single brand uniquely owns. So if you want to know who genuinely specializes in a topic, that's a click-through into each site. This section is a coverage map, not a specialization map.

Manufacturer credentials in the wild

What's actually visible on the page. We read each brand's site and only credit a manufacturer credential if it shows up where a buyer can see it. A footer badge, “certified installer” copy, a link out to the manufacturer's brand page. We don't credit credentials buried in a PDF or claimed in a sales call. So if a brand is GAF Master Elite but their site never says so anywhere a buyer would find it, they don't show up here. And that's the right call. A credential nobody can see isn't earning trust.

Social platform & review presence

Live, clickable links on the page. Not text mentions of a platform name, not a hashtag. So when a platform's share reads 78% below, that means 78% of the sites we could read carry a working link to it in their navigation or footer. It does not mean their Facebook page gets updated, or that anyone follows them, or that they've posted this quarter. A live link is the floor. What gets done with the channel is a separate question.

What search engines see on the page

98% of HVAC brands publish a page title. That's the floor. Google reads it as the headline of the search result, the first thing a buyer sees before they click. The typical page title runs 61 characters, and 53% of brands blow past Google's 60-character truncation point. So the title shows up clipped on phones, the surface most HVAC buyers are searching from.

Meta descriptions are messier. 15% of brands publish no meta description at all, leaving Google to scrape one from the page (rarely the sentence you'd have chosen yourself). When one does exist, the typical meta description runs 146 characters. And 88.5% of HVAC brands publish structured-data markup. That number is adoption only. Most stop at Organization or LocalBusiness, missing the nested Service, Review, AggregateRating, and FAQPage entities Google needs before any of the rich-result formats can render.

Form complexity

Field counts collapse Gravity Forms composite inputs (one logical "Name" field = one count, not two sub-inputs) and exclude framework hidden inputs (Gravity Forms gform_*, WPForms wpforms[recaptcha], Contact Form 7 _wpcf7*, reCAPTCHA tokens, honeypots). Counted via scripts/lib/visible_form_fields.py.

How the trade structures its pages

H1 is the single most weight-carrying heading element on a page. Best practice is one descriptive H1 per page, with a clean H2 / H3 hierarchy underneath. The HVAC trade doesn't follow that practice. 14.4% of HVAC brands ship homepages with multiple H1 elements, usually because the page builder marks every section heading as an H1. And 7% ship pages with no H1 at all, typically because a visual hero replaces it with display text styled to look like a heading but rendered as a div.

Heading hierarchy is a shared SEO and accessibility signal. The heading-order defect on this page also surfaces on the accessibility section (axe-core's heading-order rule is one of the most-frequent violations). Fixing the H1 problem cleans up two scorecard categories at once.

The cheapest unfair advantage in the trade

The median HVAC brand ships 183 images across their captured pages. The mean is 279. The heaviest brand ships 3320 images on a single site, most of them un-optimized. Image weight is the dominant contributor to mobile LCP in this trade, and the fix is mechanical: convert to WebP, enable lazy-load, ship srcset. None of those require a redesign. All three are afternoon-of-developer-time fixes. And none of them are happening at scale.

The mean image count per brand is 279, and lazy-load coverage averages 33.7% across those images. So a typical brand ships roughly 185 images that load eagerly on every page request, most of which the visitor never scrolls to. Modern-format adoption at 11.5% means the same hero photo a contractor ships as a 2MB JPEG could be a 200KB WebP without any visible quality difference. The alt-text picture is the bright spot: mean coverage runs 93.4%, and 44% of brands ship 100% coverage.

AI integration adoption

Two buckets, kept separate so the numbers stay honest. Confirmed AI means the site is talking to a real AI service in the background (OpenAI, Anthropic, or Google Gemini), or running an AI-first chatbot like Ada, Chatfuel, ManyChat, Landbot, Botpress, or Kommunicate. Or the page just says “AI-powered” or “powered by GPT” in plain English. Chat platforms with optional AI covers HubSpot Conversations, Intercom, Drift, Tidio, and Crisp. Those tools sell AI as a paid add-on, but a chat widget being installed doesn't prove the AI is switched on. So we count them in their own bucket. Otherwise every HubSpot install would inflate the AI number and this whole section would lose its meaning.

Confirmed AI integration: 1% of brands

Chat platforms with optional AI features: 2% of brands

Marketing & analytics stack

We read every page on every brand's site and looked for the named tools they had running. Numbers below describe what's actually installed on the typical HVAC contractor's website. Not what an agency proposal claimed they were going to install.

Tech stack profile

Tech-stack detection runs on the rendered HTML we captured for every brand. Each percentage below is out of all 104 HVAC sites in the sample.

Trust badges and credentials the trade actually shows

One unified read across each whole site. Financing partners by name, warranty language, certification tiers, ownership claims. We anchored every pattern specifically so we wouldn't catch false positives. For example, "Best of Houzz 2025" or "Best of Angi 2024" qualifies. A loose "best of" floating in a paragraph doesn't.

Content patterns

What kind of content lives on the site, based on URL paths and how each page is laid out.

What separates the top quartile from the bottom

The top-28 brands averaged against the bottom-37. The bigger the gap, the more that category separates winning sites from losing ones.

How the trade structures its websites

Architecture detected mechanically: URL discovery, page-role coverage, and HTML patterns. The bulk of the HVAC trade uses the canonical multi-page layout (homepage + service + location + contact). Single-page brochures score meaningfully lower because the framework rewards page-role coverage. Multi-step quiz capture and chat-widget conversion architectures appear in small numbers, often outscoring single-page sites because they handle intake more deliberately.

Who they are, where they are

Tenure is derived from “since YYYY” / “established YYYY” patterns in body text on each captured site. A brand without such a phrase is excluded from the tenure aggregate, not “new,” just “didn't display founding year prominently.” 51 HVAC brands (49% of the trade sample) surface a founding-year marker on their public site, and the median one among them dates to the 1980s. The breakdown by founding decade is below.

An addressable-opportunity projection

Read this carefully. The Revenue Loss Estimate is a modelled industry projection, not a measurement of any single brand's actual revenue loss. The formula uses uniform per-trade traffic benchmarks. Each brand's RLE assumes the same range of monthly visitors regardless of brand size, the same close rate, and the same average project value. This matches what comparable industry research bodies do (NRCA, IBISWorld, JCHS, Verified Market Reports) and exists for inter-brand comparability inside the trade sample, not for per-brand financial planning.

The right way to read this is as the category-level addressable conversion-uplift opportunity. So if every HVAC brand inspected received its benchmark monthly traffic and improved its on-site conversion rate by the amount its Fervor Score implies, this is the modelled total of new revenue the category would capture. The figure should not be summed against any single brand's books, and small local operators sit inside the same traffic assumption as national franchises.

How we measured this

Same process on every HVAC brand. We captured full-page screenshots at desktop and mobile width across every page our crawler could reach. Then we read each underlying page. We looked at forms, the structured markup Google uses to build rich results, phone and email links, social-media links, manufacturer badges, review widgets, chat widgets, and common body-copy patterns. We ran axe-core 4.10.2 (open-source accessibility, configured against WCAG 2.1 AA). And we measured Core Web Vitals using Google Lighthouse 13.3.0. Then we scored what we captured against the Fervor Grade™ Framework. 100 points across six categories, and a letter grade A through F.

Base Score and Grade Adjustments

Every Fervor Score on this page is an Adjusted Score. That's the Base Score the framework returned, plus any Grade Adjustments. A Grade Adjustment is a small, documented correction we apply where the standard rubric penalty overstates the real conversion impact for the HVAC trade. Across the HVAC trade sample, the Base Score mean is 63.84 and the Adjusted Score mean is 65.32, a net Grade Adjustment of 1.48 points up. The headline figures on this page, including the 65.32 mean, are the Adjusted Scores.

What “header-prominent” means

A few of the signals on this page depend on where something sits on the page, not just whether it exists. “Phone in persistent header” means a clickable phone number in the top section of the page, the part that stays visible as a visitor scrolls. A phone number buried in the footer doesn't earn the credit. “Inline form on hero” means a form in the first thing a visitor sees when the page loads, before any scrolling happens.

What we count as a real tool

For the marketing-stack, accessibility, and AI sections, we only credit a named tool if its signature shows up in what the page actually loads. Intercom, Drift, Tawk, HubSpot, OpenAI, axe-core, Lighthouse, and so on. We don't take a brand's word for what's installed. And we don't ding a brand for things they choose not to advertise. So if we can't find the named tool in the page, it doesn't count.

The averages on this page (means, medians, distributions) are reproducible. Anyone with the same captures can compute the same number. Per-brand callouts aren't published here because they're still in the integrity-review queue. Individual scores live on each brand's published report under the Contractor CRO Index.

Citation

"HVAC Contractor Website Performance: State of the Industry 2026." The Contractor CRO Index, Fervor Studio. 2026-05-22. Sample size: 104. Methodology: Fervor Grade™ Framework.

Short form: (Contractor CRO Index, Fervor Studio, 2026) · Publisher: Fervor Studio · fervorstudio.ca

Glossary of terms

Definitions for terms used on this page. Where a term is also industry-standard (WCAG, axe-core, Lighthouse), the definition reflects how the Fervor Grade™ Framework applies it.