14 HVAC Marketing Mistakes, Each One Documented

Scorpion's 2026 State of Home Services Marketing Report — 2,000 U.S. homeowners aged 25 and older who hired a trades professional within the past 18 months plus 944 home services operators, marketers and executives, surveyed late 2025 with Dynata across HVAC, plumbing, electrical, lawn, pest and roofing — found that 56% of homeowners want 24/7 scheduling or a way to communicate after hours, while 66% of businesses say providing after-hours customer service is their top challenge. Scorpion sells marketing technology, so read that as vendor-collected survey data rather than neutral fact.

Google adds a harder cost. Its Local Services Ads documentation states that "If you regularly fail to answer calls or respond to messages, your ad ranking may be affected". In that channel the missed call is charged as a valid lead, loses the job, then degrades the position that produced it. Three penalties, one ringing phone.

Mistake 1 — buying demand before the phone is covered. Ranking this first is our reading of the published evidence, not a measured finding; no study we could source isolates the revenue lost to unanswered HVAC calls. The arithmetic is why it leads anyway: fixing it changes the return on every item below, and none of them changes it back. Pricing that loss — which benchmarks survive being traced to a primary source and which collapse — is a separate page's job, worked through in the missed-call cost evidence, traced claim by claim.

Mistake 2 — paying a retainer whose published floor excludes your shop. Hook Agency publishes its prices, which is rare and useful, and also publishes who it is not for: "we are usually the best fit for growth-mode contractors over 3M+ because of the significant investment required", with a revenue selector whose first option reads "Under $1M – Not a Fit at This Time". A two-truck shop signing at that tier buys a programme designed around a much larger revenue base.

Mistake 3 — treating "AEO" as a comparable line item. Two acronyms need pinning before the prices make sense: AEO is answer engine optimization and GEO is generative engine optimization — both meaning citation inside AI-generated answers, neither meaning geographic targeting. Hook Agency lists AEO at $4,000/mo, footnoted "*$2k when added to an SEO package". Footbridge Media includes both Answer Engine Optimization (AEO) and Generative Engine Optimization (GEO) inside a "No Contracts. No Hidden Fees. Just $249/Month" package. Those two prices sit sixteen times apart under the same acronym, so comparing them is the mistake; asking each vendor to define the deliverable in writing is the fix.

Mistake 4 — signing without knowing which line items are term-locked. Scorpion's HVAC FAQ states: "Yes, Scorpion typically requires a 12-month contract for our marketing technology and certain marketing services like search engine optimization (SEO)", and separately that "you own your website after the length of your contract with Scorpion is complete." Footbridge publishes the opposite posture: "No long-term contracts here. We work month-to-month." Hook Agency sits between the two and shows why the question needs asking twice: its price card names no contract term for SEO, AEO or PPC, but prices both website tiers as a twelve-month payment schedule — "$12K Total - Split into 12 Months" and "$24K Total - Split into 12 Months". A payment schedule is not a contract term, and only the agreement says whether it also functions as one. All three are legitimate models. Not knowing which one you signed is the mistake.

Mistake 5 — hiring from a roundup nobody updated, or an agency running a retired playbook. Blue Corona's announcement dated October 1, 2024 states it and RYNO Strategic Solutions are "unifying under the RYNO Strategic Solutions brand" — yet bluecorona.com still ranks top-three on several HVAC agency queries. KickCharge Creative published on February 19, 2025 that "we are thrilled to announce our acquisition of Levergy", and levergy.io returned HTTP 503, "Site Currently Unavailable", when we fetched it on 2026-08-17. Tactics go stale the same way: Google now issues a single Google Verified badge and is discontinuing the Money Back Guarantee, with reimbursement available only for services booked before December 7, 2025, and has replaced manual lead disputes with automated credits that no longer cover "job type not serviced" and "geo not serviced" leads. A pitch built on the old badge names and a dispute process quotes a programme Google has retired.

Mistake 6 — accepting a guarantee of rankings or AI citations. Google Search Central is unambiguous: "No one can guarantee a #1 ranking on Google", and on audits, "If they guarantee you that their changes will give you first place in search results, find someone else." The same page asks whether a provider's AEO or GEO advice aligns with Google's published guidance. Audits, programmes and measurement are sellable; outcomes are not — a rule we hold ourselves to, with our own number printed where anyone can check it instead of a promise — $2,500 a month, GEO and AI-search work included, and no outcome attached to it.

Mistake 7 — buying a pitch built on uncited numbers. Three of the largest HVAC agency sites carry statistics that name no study anywhere on the page. Hook Agency asks: "Did you know that HVAC businesses that invest in effective marketing strategies are 43% more likely to experience significant growth?". Thrive Internet Marketing Agency states that "The sad truth is that 90 percent of HVAC contractors do not understand the basic online marketing concepts". Blue Corona opens its HVAC page with "97% of users search online to find a local business, making the internet the number one place homeowners search to hire a heating and air conditioning company". All three pages were fetched on 2026-08-17; none of the three sentences names a study, a sample size or a date. Ask for the study; if the answer is a link to another agency blog, the number has no floor under it.

Mistake 8 — buying an AI-visibility file instead of a page. Google's guide to its generative AI features states that "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn't use them", that "Structured data isn't required for generative AI search", and, as a summary: "For Google Search, you can ignore tactics like 'chunking' content, creating unnecessary AI text files (like llms.txt), or pursuing inauthentic mentions." Keep schema for rich-results eligibility; stop paying for it as an AI unlock. What the file does and does not do outside Google Search — which engines read it, and what it costs to keep — is settled elsewhere, in the llms.txt guide for contractor sites.

Mistake 9 — blanket-blocking the crawlers you are paying to be found by. OpenAI's two agents do different jobs, and OpenAI publishes the difference: "Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links", while disallowing GPTBot only signals that content should not train foundation models. A template that disallows both removes an HVAC site from ChatGPT search answers as a side effect of a training-data decision. Google's equivalent token is narrower still: Google-Extended "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search" — it governs Gemini Apps and Vertex AI, not AI Overviews.

We audited this rather than assuming it, fetching the robots.txt of 14 HVAC marketing agency domains on 2026-08-17 and counting named AI/LLM user-agent groups:

Two readings follow. The "your robots.txt is blocking AI" upsell is priced against a problem this field does not have — zero of fourteen block anything. And silence is not a decision: because Google "supports the following fields (other fields such as crawl-delay aren't supported)" and crawlers ignore rules without a path, an empty disallow: blocks nothing, while the Crawl-delay: 10 on kickcharge.com and mediagistic.com is discarded by Google and may be honoured elsewhere. Auditing the file takes fifteen minutes; paying a retainer for it is the mistake. Running that audit on your own domain — the directives to read, in order, and what a pass looks like — is the job of the five-minute AI crawler access check.

Mistake 10 — marking up your own testimonials to chase stars. Google's Local Business structured data reference requires only address and name, and frames the review property for sites that capture reviews about other businesses. Star ratings in the SERP for a contractor's own testimonial wall are not an expected outcome of that markup.

Which page elements actually convert a visitor is a separate question with its own evidence base, worked through in the conversion elements an HVAC page has to carry.

Mistake 11 — generating a page for every suburb. Google's spam policies define doorway abuse to include "Having multiple domain names or pages targeted at specific regions or cities that funnel users to one page" and "Creating substantially similar pages that are closer to search results than a clearly defined, browseable hierarchy." The adjacent policy on scaled content abuse names generative AI tools used "to generate many pages without adding value for users", then adds the operative phrase: "no matter how it's created." AI authorship is not the violation; forty near-identical city pages funnelling to one contact form is.

Mistake 12 — incentivising reviews. The FTC's final rule on consumer reviews and testimonials, announced August 14, 2024 on a 5-0 Commission vote, sets out six prohibited conduct categories and authorises courts to impose civil penalties for knowing violations. Two reach HVAC shops directly. On buying reviews, the rule "prohibits businesses from providing compensation or other incentives conditioned on the writing of consumer reviews expressing a particular sentiment, either positive or negative." On suppression, it prohibits "unfounded or groundless legal threats, physical threats, intimidation, or certain false public accusations to prevent or remove a negative consumer review." Read the FTC's own limits before rewriting a review process — the rule took effect October 21, 2024, does not ban incentives outright where no sentiment is required, and provides no private right of action. None of this is legal advice. Building the request sequence that stays inside those limits is handled in the FTC-compliant review request system.

Mistake 13 — buying leads whose conversion rate nobody can substantiate. The clearest documented case in this market is regulatory. On January 23, 2023 the FTC issued a proposed order requiring HomeAdvisor to pay up to $7.2 million and stop deceptively marketing its home-improvement leads. Providers generally paid a $287.99 annual membership plus a per-lead fee; the complaint alleged HomeAdvisor "often tells service providers that its leads result in jobs at rates much higher than it can substantiate." Redress was capped at up to $30 per provider, and by November 28, 2023 the FTC had sent more than $3 million across 110,372 checks to eligible home service providers.

"Today's order requires HomeAdvisor to refund home service providers millions of dollars and stop misleading them about the quality of its leads. Even as the nature of work and the economy change, the FTC will continue to combat dishonest commercial practices aimed at consumers, workers, and small businesses."

— Samuel Levine, Director of the FTC's Bureau of Consumer Protection, January 23, 2023

The ratio a contractor is left with — a $287.99 annual fee, per-lead charges on top, a $30 redress ceiling — is the argument. Not "never buy leads": a lead source that will not show its conversion denominator is selling a number it cannot defend.

Mistake 14 — treating one ranking, or one AI citation, as the result. Ahrefs — an SEO tool vendor publishing its own study, so read it as vendor-collected data rather than a Google-reported metric — re-ran its analysis for 2026 and found the presence of an AI Overview correlates with a 58% lower average clickthrough rate for the top-ranking page on informational keywords, December 2023 against December 2025 — correlation, position one only. Pew Research Center, which sells nothing in this market, studied the browsing behaviour of 900 U.S. adults on google.com and found users who met an AI summary clicked a traditional result in 8% of visits versus 15% without, and clicked a link inside the summary in just 1% of visits (March 1-31, 2025). Position one is not pipeline; a citation is not a click. What that click loss does to HVAC lead volume specifically, and where the evidence stops short, is argued in the honest read on AI Overviews and HVAC leads.

SISTRIX supplies the other half, and is likewise an SEO tool vendor reporting from its own index. It analysed 82,619 prompts over 17 weeks and measured weekly source churn of 5% in AI Overviews, 56% in Google AI Mode and up to 74% in ChatGPT Search, concluding that "with a 56% weekly drift, a single citation placement is not a reproducible result but merely a snapshot." Semrush, a third SEO tool vendor testing on its own blog, ran a citation-speed test that carries the caveat mattering most here: 29 of 81 pages (36%) were cited by Google AI Mode within 24 hours and 21 of 81 (26%) were still cited by day 30 — on a domain with Authority Score 84, which Semrush states gave those pages "the best possible conditions to succeed". A new HVAC site should not be sold that curve.

1. Cover the after-hours window before adding spend (mistake 1).
2. Get the deliverable defined in writing before comparing prices — a $249 AEO line and a $4,000 AEO line are not one product (mistakes 2, 3).
3. Ask which line items are term-locked, and who owns the site, content and tracking numbers at the end (mistake 4).
4. Verify the agency still exists as named — check for a merger notice, load the domain (mistake 5).
5. End the call on any guarantee — rankings, page one, AI citations (mistake 6).
6. Ask for the source of every number in the pitch; an agency blog is not a source (mistake 7).
7. Refuse llms.txt or "AI markup" as a visibility product, and keep schema for rich results (mistake 8).
8. Read your own robots.txt — confirm OAI-SearchBot is not disallowed, and treat Google-Extended as a Gemini decision (mistake 9).
9. Drop self-serving review markup; put reviews where AI engines and homeowners read them (mistake 10).
10. Delete near-duplicate city pages in favour of fewer pages with real local substance (mistake 11).
11. Rewrite the review request so no incentive is tied to sentiment (mistake 12).
12. Demand the conversion denominator from every lead source (mistake 13).
13. Report booked jobs, not positions, and track AI visibility as presence over weeks (mistake 14).

Two neighbouring questions sit just outside this page. On how much should move through these channels at all, ACCA's Contractor of the Future Study of more than 1,000 HVACR contractors found that the average contractor spends 6% of annual revenue on marketing and advertising, and contractors who invest at least 12% reported net profits of 9% versus 5% — correlation in a survey, not a proven return, and the closest thing to a primary HVAC benchmark that exists. Sizing builds on it in setting the annual number at each company size; allocation is handled in which channels return the money you put into them. If the conclusion is that the current agency is the problem, how each of these firms scores once the published evidence is lined up applies this same sourcing standard to the vendors themselves.

One account of our own makes mistake 14 concrete, and it is published here with the half most case studies cut. A Winston-Salem HVAC and refrigeration contractor we run SEO for is up 234.4% in organic clicks and 426.2% in top-10 ranking keywords since engagement start, measured in Google Search Console to 14 August 2026. It also plateaued: clicks peaked around 24 July 2026 at 2,486 for the month and stood at 1,886 on 14 August. The constraints behind that flattening are technical and measured rather than mysterious — largest contentful paint of 4,260ms against Google's 2,500ms "good" threshold, and 411 unindexed URLs against 268 indexed. Both halves are the point: a percentage with no window and no plateau attached is the same uncited number mistake 7 warns about, only it is ours.