The HVAC CRM playbook.

Most CRM software was built for a salesperson with a pipeline of deals that move in a straight line. HVAC does not work like that. You have a service call that might become a repair, a repair that might become a replacement quote and a replacement quote that sits for three weeks while the homeowner gets two more prices. Meanwhile the same customer is on a maintenance agreement that renews in April, and none of that is a deal stage.

So the first thing to look for in HVAC CRM software is whether the record is built around the equipment and the address rather than the opportunity. You want to open a customer and see the system you installed, when you installed it, every visit since, what the tech wrote down, what you quoted and what they said no to. That history is what turns a summer breakdown into a replacement conversation instead of another patch.

The second thing is seasonality. A generic CRM assumes steady demand. Your year has two walls of work with quiet stretches between them, and the quiet stretches are where the margin gets made or lost. Software that only helps when the phone is already ringing helps you in July and abandons you in October.

The third is that your team is not at desks. Anything that needs a technician to log in and update a field at the end of a job will not get done, because at the end of a job they are driving to the next one. What works is a system that fills itself in from the call, the text and the booking, and asks the office for as little typing as possible.

When a system dies on the hottest night of the year, the homeowner does not read reviews. They work down the search results and hire whoever picks up. Everything else you do to win that customer, the truck wrap, the reviews, the years in business, only matters if you are the one who answers.

That is a staffing problem with no good staffing answer. Paying someone to sit by a phone through the night is expensive in the weeks it pays for itself and dead money in the weeks it does not. An answering service picks up but cannot tell an emergency from a filter question, so you either get woken for everything or you find out in the morning.

The AI agent answers every call, at any hour, and has an actual conversation. It asks what the system is doing, whether there is heat or cool at all, how old the unit is and whether anyone in the house is vulnerable to the temperature. A genuine emergency gets your on-call tech. A routine call gets booked into the next open slot with a confirmation text, and it is on the board before your dispatcher opens the laptop.

The part that matters more than the after-hours coverage is what happens at two in the afternoon in August. Your office is on another line, your techs are in crawl spaces, and the calls that ring out are the ones you never learn about. The agent takes those at the same time, as many at once as arrive, and none of them go to the contractor who happened to be free.

Every HVAC company is sitting on a list of people who already paid them, already trust them and already have equipment that is a year older than the last time anyone called. That list is the cheapest work available to you and it is almost always the least worked, because calling it is nobody's actual job.

There are three jobs in that list and they are all mechanical. Maintenance agreements come up for renewal and nobody rings, so the customer quietly lapses and you lose the recurring revenue and the twice-yearly foot in the door. Pre-season tune-ups need booking before the weather turns, not after everyone else has called. And systems past replacement age need a conversation while the homeowner is comfortable, not while they are standing in a hot house getting an emergency price.

The agent works that list on a schedule. It calls ahead of a renewal date, books the visit and writes the outcome to the record. It runs the pre-season calls in the weeks before you need them, so the shoulder months have work in them. It calls the addresses where the equipment is old and offers an assessment, which is a much easier conversation than the one that starts with a failure.

None of this is new as an idea. Every good HVAC company knows it should be happening. It does not happen because it depends on a person with time, and in this trade nobody has time in the season and nobody remembers out of it.

It is fair to be sceptical about an AI on the phone with your customers, so here is what the conversation is made of rather than an assurance that it is good.

On an inbound call it establishes the problem in plain terms. Is there any heating or cooling at all, or is it running badly. What is the thermostat set to and what is the house actually reading. Is the outdoor unit running, and is it making a noise it did not make before. How old is the system, roughly, and did you have it put in or was it here when you moved in. Is anyone in the house elderly, very young or unwell.

That last question is what separates a call that needs a tech tonight from one that can wait until Tuesday, and it is the question a script without judgement gets wrong.

Then it books. It reads your real availability rather than promising a window you cannot serve, confirms by text and writes the job with the answers attached, so the tech arrives already knowing the age of the unit and what the homeowner described.

On an outbound call it is a different job and a shorter one. It has a reason to be calling, a renewal coming up or a system at replacement age, it says so, and it is trying to get a time in the diary rather than a sale on the phone.

Every call is recorded and transcribed, and a coach scores it against how you want your company to sound. You can read a bad call, hear what happened and change what the agent says next time. That is the part worth insisting on with any vendor: you should be able to inspect the conversations, not just the outcome.

An HVAC CRM that becomes a second place to look is worse than no CRM, because now the truth about a job lives in two systems and neither is complete. If your dispatch and invoicing already run in ServiceTitan, Housecall Pro or Jobber, the test is not whether the new thing has its own calendar. The test is whether a booked job appears in the system your techs already open in the morning without anyone retyping it.

That is the question to ask a vendor first, and to ask concretely. Does the appointment land in my dispatch board. Does the customer record match the one that is already there or does it create a second one. When the tech closes the job, does the outcome come back. If the answers are vague, what you are being sold is a parallel system and your office will end up doing double entry to keep it honest.

The same applies to the phone number. Moving numbers is the thing that makes contractors put a decision off for a year, reasonably, because a dead number in season is a catastrophe. It should connect to the number you already publish rather than asking your customers to learn a new one.

What you should end up with is one place where a customer's history is complete and no place where your team is typing the same thing twice. Next to that, the rest of a feature list is decoration.

This is the section most vendors leave out, so here is the honest version, including the parts that do not flatter us.

Ask what happens in your busiest week, not your average one. Software that keeps up with forty calls a day may behave differently at two hundred. Ask specifically what happens when calls arrive at the same moment, because that is the situation that costs you jobs.

Ask to hear real calls, not a demo recording. Every voice product on the market sounds capable when the demo is scripted. What you want is a call where the customer interrupted, changed their mind or was annoyed, and to hear how it went.

Ask how long until it is doing anything useful. Most products in this category take a couple of months of configuration before they earn a cent, and the demo never mentions it. Get a go-live date, in writing, before the season turns.

Ask what it costs when you use it hard. Per-minute and per-call pricing looks cheap on an average month and lands badly in a heat wave, which is exactly the month you needed it. Work out the bill for your worst week, not your normal one.

Ask what happens to your data if you leave. You should be able to take your customer list, your call recordings and your history out without a negotiation.

And be honest with yourself about whether the problem is software. If you are missing calls because you are one van short, a CRM will show you the missed calls in more detail and will not fix them. The software is worth buying when the work is arriving and falling through gaps, which in this trade it usually is, but not always.

HVAC leads arrive in bursts. A heat wave or a cold snap sends the phone off the hook for three days, then it goes quiet, and the quiet is when the business gets decided.

It helps to separate the sources. Emergency calls come from people whose system has just failed. They call two or three companies and book whoever answers. Replacement enquiries come from people with an ageing unit who are comparing quotes and can wait a week. Maintenance leads come from existing customers who are due a visit and will say yes if someone asks. Social media and local search feed the first two. Your own customer list feeds the third, and it is usually the cheapest.

Each source needs a different response. An emergency lead needs an answer in minutes, day or night, and a human on the way. A replacement lead needs a call back, a clear quote and a follow-up after the quote, because that is where most of them go quiet. A maintenance lead needs a reminder at the right time of year.

The agent covers the gaps you cannot cover by hand. It answers when your team is on a roof, takes the address and the problem, passes true emergencies straight to a person and books the rest. It then chases the quotes that never got an answer.

The point of tracking leads this way is simple. When you can see which source produces booked jobs, you stop paying for the ones that only produce calls.

An HVAC business can feel busy and still not know where its jobs come from. The fix is to count a few things every week and keep the habit.

Count the calls that came in and how many were answered. A missed call in peak season is usually a lost job. Count the quotes you gave and how many turned into work, because the gap between those two is where most of the money sits. Count booked jobs by source, such as search, referrals, your own customer list and social media, so you know which to feed.

The same simple tracking shows when to act. If quotes are going out but not closing, the follow-up is the problem and not the price. If calls are going unanswered, the cure is coverage, not more ads. If the maintenance list is not being contacted, the next-year revenue is sitting in a spreadsheet.

A CRM does this counting for you. Every call, quote and booking is recorded against the customer, and the agent follows up on what is outstanding. The result is a clear picture of your leads without anyone filling in a report.