What HVAC buyers do before they call.
The calls you get are survivors. Between the moment a homeowner decides they need help and the moment your phone rings, a quiet filtering process runs its course. Most contractors are eliminated during it without ever knowing they were considered. This is the model we use to map that process, checkpoint by checkpoint.
This is not a survey and we did not poll homeowners. It is the working model behind our examinations, built from how search, reviews and AI assistants demonstrably behave, and from the patterns we keep meeting in client markets. Test every step against your own experience as a buyer. That is the standard it has to survive.
A referral used to end the search. Now it starts one.
Most HVAC owners will tell you their best work comes from word of mouth, and they are right. The mistake is the conclusion drawn from it: that a strong referral pipeline makes online presence optional.
Watch what actually happens when a neighbor recommends you. The homeowner does not dial. They type your company name into a phone. They look at your rating before they look at your number. They skim three reviews, maybe open your website, maybe ask an AI assistant whether anyone has complaints about you. The referral got you into the process. It did not get you through it.
This is the single most expensive misunderstanding in the trade. Word of mouth has become the opening argument, not the verdict. Every checkpoint that follows can still overturn it, and each one runs silently, on a screen you never see, before your phone has any chance to ring.
You are not competing for the call. You are competing to survive until the call.
The checkpoints never change. The clock does.
HVAC demand arrives in three distinct moods. Each one runs the same filtering sequence, but at a different speed and with a different level of scrutiny. A marketing system that treats them as one buyer fails all three.
No cooling, house at 90 degrees
Searches from a phone, scans whoever appears first, gives reviews a five-second glance and calls the first company that looks credible and answers. The vetting is shallow but brutal: if you are not visible and reachable at that moment, you were never in the running.
A system near end of life
The big ticket buyer. Collects two or three quotes, reads reviews properly, compares websites, asks about financing and looks for a reason to trust one company over the rest. Every checkpoint gets its full weight. This is where the pre-call journey earns or costs the most money.
Tune-ups and peace of mind
Low urgency, price sensitive, loyal once settled. Chooses whoever makes booking effortless and gives a clear price signal. Quietly the most valuable filter to win, because the maintenance customer is standing in the house when the replacement decision arrives.
Four gates between "I need help" and your phone ringing.
We model the pre-call journey as a sequence of gates. At each one, the homeowner is not choosing a winner. They are removing candidates. You do not need to be the best at any gate. You need to not be eliminated at all four.
Search
The homeowner searches the problem or your name, almost always on a phone, and considers only what surfaces in the first screenful: the map results, the ads, the top organic listings.
Reviews
Whoever survived the search gets their reputation read: the rating, the count, how recent the latest reviews are, and how the company responds when something went wrong.
Website
The shortlist gets a site visit, on cellular, between other tasks. The buyer wants three answers fast: do you do my job, do you serve my area, and what will this roughly cost or how do I find out.
The question stage
Increasingly, the tiebreak happens in an AI assistant or a comparison search: "is X reputable", "X vs Y", "who should I hire". The machine summarizes everything public about you, whether you participate or not.
The candidate pool narrows at every gate. The widths above are illustrative, not measured; the direction is the point. By the time a phone rings, most of the market has already been removed from consideration.
What we examine at each checkpoint.
When we run this model against a real company in the diagnostic, these are the questions each gate has to answer. They work as a self-inspection too.
Search: do you exist where the buyer stands?
Rankings are not one number. They are a map. A company often ranks well beside its own office and vanishes four suburbs over, inside its actual service area.
- Visibility across the full service radius, not just the office zip code
- Business profile complete, categorized correctly, photographed recently
- Presence on the searches that spike when the weather turns
Reviews: does your reputation read as alive and human?
Buyers compare you to the companies listed beside you, not to an abstract standard. Recency and responses carry more weight than owners expect, because they signal what happens after the invoice.
- Volume and rating relative to the named competitors around you
- A steady drip of recent reviews, which means a process that asks for them
- Owner responses on the bad ones. A well-handled complaint is proof, not damage
Website: does it answer the three buyer questions fast?
The site is not a brochure. It is the moment consideration becomes a decision, and it happens on a phone. We treat pricing silence as the biggest single leak at this gate: the buyer knows you know the number.
- Loads fast on cellular, on a mid-range phone, not just in the office
- Says what you do, where, and how pricing works, in the first screen or two
- One-motion contact: a tappable number, a short form, a booking path
The question stage: what does the machine say about you?
Ask an AI assistant about contractors in your area and it will answer from your public record: profiles, reviews, citations, your own site. That answer is already shaping shortlists, and it cannot be bought, only earned in the open.
- Name, address, services and service area consistent everywhere they appear
- Enough verifiable public substance for the model to describe you confidently
- Your own site stating plainly what you want the machine to repeat
The defining property of a filtered buyer: no trace.
A missed call shows up in a log. A lost quote shows up in a CRM. A homeowner who read your reviews and moved on shows up nowhere. There is no report of eliminations, no dashboard of people who almost considered you. The losses at these four gates are structurally invisible from inside the company.
This is why busy companies with full boards can still be losing badly at the pre-call stage. The work that arrives feels like the whole market. It is actually the residue that survived a filter you have never inspected.
It is also why we insist on examining companies from the outside, the way a buyer meets them, before connecting to any internal dashboard. Your internal data describes the survivors. The gates are only visible from the street.
Your CRM shows who survived the filter. It cannot show the filter.
Run the gauntlet on your own company tonight.
Thirty minutes, a phone, and honesty. Do each step as a stranger would, not as the owner hoping to pass.
-
Search your trade from the edge of your service area
Not your company name. "AC repair near me" from a neighborhood you serve but do not sit in. If you are not in the first screenful, that neighborhood does not know you exist.
-
Read your last ten reviews as a skeptic
Check the dates. Check whether anyone from your company replied. Then open the profile of the competitor listed next to you and ask which company you would call.
-
Load your website on cellular, standing in your driveway
Count the seconds. Then try to answer three questions in thirty more: do they do my job, do they cover my street, what happens about price. If any answer is missing, the buyer supplies their own.
-
Ask an AI assistant who to hire in your city
Then ask it directly about your company. What it says, and what it cannot say, is a reading of your public record. That record is now part of your sales team whether you manage it or not.
The rest of the machine.
Find out which gate is eliminating you.
The Marketing Diagnostic runs this examination on your actual market, with numbers and named competitors. $1,000, credited in full to month one of any program.