Conversion optimization for HVAC companies, mostly away from the website

Leads is the single number most likely to hide the actual problem

A count of inquiries tells you nothing about whether the business converted any of them, and it moves for reasons that have nothing to do with performance.

Between an inquiry and money there are several distinct conversions: answered, qualified, scheduled, arrived, quoted, sold, installed, collected. Each one has its own rate, and a business can be excellent at some and poor at others while the lead total looks fine.

The failure this hides is common enough to be worth naming. Marketing increases inquiries, the office cannot absorb them, answer rate drops, and total booked jobs stay flat or fall. The report says leads are up. The board says nothing changed. Both are true.

Writing the chain down is usually the whole first project. Not modeling it, just counting it honestly for a month: how many rang, how many were answered, how many got on the schedule, how many got a quote, how many said yes.

The step with the worst rate is rarely the one anybody expected, and it is usually not on the website. It is often a stretch of the day when nobody is on the phone, or a category of job the office quietly declines, or a quote that never got a second call.

Once the chain is counted, most decisions get easier. You stop debating a headline when you can see that a quarter of calls in the busiest hours are not being answered.

You almost certainly cannot run a valid test, so change the process instead

Split testing needs volume that a contractor does not have, and pretending otherwise produces confident conclusions from noise.

Detecting a small improvement reliably takes far more conversions than most shops produce in a month. Run a test at contractor volume for two weeks and you will get a result, but it will mostly be describing the weather.

That is not an argument against improving anything. It is an argument against a particular method, and for the ones that work at low volume: reading what people actually said, watching how the process runs, and changing things large enough that you do not need statistics to see the difference.

Big changes are readable. Adding after-hours answering, changing what the quote looks like, introducing a follow-up sequence that did not exist. When something moves by half, you do not need a significance calculation.

Small changes are not readable at this volume, which is a reason to stop spending time on them. The color of a button is not going to be the thing.

There is one more honest caveat. Because the trade is seasonal, any before and after comparison is confounded by the calendar. Comparing this February to last February is imperfect and still far better than comparing February to July.

Financing is a conversion step, and hardly anyone instruments it

For replacement work the finance application is often where the sale is actually won or lost, and most shops have no visibility into it at all.

A meaningful share of system replacements involve financing. That introduces steps most companies never look at: whether the option was presented at all, whether the customer applied, whether they were approved, and whether an approval turned into a signed job.

Each of those can fail independently. A technician who is uncomfortable raising financing skips it. A customer intimidated by an application does not start one. An approval comes back at terms that kill the deal. None of it appears anywhere in a marketing report.

Presentation moves this more than most people expect. Whether the option is offered to everybody or only to people who flinch at the price, whether it appears on the quote as a line or as an afterthought, and whether the person presenting it can answer a question about it.

Watch the gap between presented and applied especially. A large drop there is usually about how it was raised rather than about the customer's finances, and it is fixable in a way that credit decisions are not.

Whatever you promote about a finance offer carries disclosure obligations. Worth getting the presentation right and the advertising checked separately.

Your estimate gets read by somebody who never met you

The document leaves your hands and then argues on its own against two competitors, usually at a kitchen table that evening.

A replacement quote is rarely decided in the room. It is decided later, by two people, at least one of whom was not there, reading three documents side by side. Whichever one is most legible has an advantage that has nothing to do with price.

Most contractor quotes are not built for that. A model number, a total and a signature line requires the reader to already understand what they are buying, and the person who was not home does not.

What travels well is plain: what is being replaced and why, what is included down to the parts people forget about, what the work will involve in their house, how long it takes, and what happens if something goes wrong afterward.

The comparison problem cuts both ways. If your quote includes work a cheaper one leaves out, an unexplained higher number reads as a higher number. Saying what is in yours is the only way the difference is visible.

This is one of the highest leverage documents in the business and it is usually generated by software nobody has looked at critically in years. Reading three of your own recent quotes as though you were the spouse who missed the appointment is a short and uncomfortable exercise.

Reviews do their converting before your website ever loads

For a large share of prospects the decision narrows in the search results, and the site only meets people who already got through that filter.

A homeowner comparing contractors is usually reading ratings and recent reviews before clicking anything. By the time somebody arrives on your site, a selection has already happened that no page can undo.

That makes review generation a conversion activity rather than a reputation activity, and it responds to process. Asking at the right moment, which is generally while the technician is still there and the relief is fresh, works far better than an email three days later.

Recency carries weight with readers in a way that a lifetime average does not. A strong overall rating whose newest review is from two years ago reads as a company that used to be good.

Responding matters more than most owners think, and it is read by prospects rather than by the reviewer. A measured reply to a hard review does more for the person scrolling than a perfect record would.

Do not buy reviews, do not gate them by asking only happy customers, and be aware that incentivizing them has drawn regulatory attention. The downside is not a policy warning, it is the day a competitor points it out.

The technician standing in the basement is the one closing the work

The most important conversion conversation in an HVAC company happens with no marketing person present, and it can be equipped.

The moment a system is condemned or a repair turns out to cost more than the equipment is worth, a service call becomes a sale. The person holding that conversation is a technician who was hired to fix things.

Equipping that is practical work. A tablet with the options already priced, photographs of what was found, a written summary the homeowner can keep, and a clear handoff to whoever schedules the estimate if the technician does not do it themselves.

Photographs of the actual failure do more than any brochure. A homeowner who has seen a cracked component or a rusted-through pan understands the decision. One who has only been told about it is deciding whether to trust a stranger.

What usually goes wrong is not persuasion, it is documentation. A technician explains everything clearly, leaves, and the customer cannot reconstruct any of it two hours later when the spouse asks what he said.

None of this needs to turn a service department into a sales floor. It needs the findings to survive the trip from the basement to the kitchen table.

When there is no data yet, fix the things that cannot be wrong

A short list of changes is worth making before anybody has measured anything, because the downside is close to zero.

Answer the phone during the hours you advertise, and know what happens to a call you miss. Somebody with no heat does not leave a voicemail, they dial the next result.

Text back missed calls. It costs almost nothing, it recovers people who have already moved on, and it works because the customer's phone is in their hand either way.

Put the diagnostic fee, the trip charge and the after-hours rate somewhere findable. Withholding them does not prevent the question, it moves it to a phone call where somebody has to say the number out loud to a person who feels ambushed.

Follow up on quotes more than once. A single call two days later is the most common follow-up sequence in this industry and it is roughly half of one.

None of these need proof to justify. They are the floor, and until they are in place, anything measured is measuring the floor rather than the marketing.

Answer times, dispatch zones and scripts belong to your market

The chain above is the same everywhere. The numbers in it, and what the phone should say, are not.

What counts as a reasonable arrival window depends on traffic. What the office should say about tonight depends on how full the board is in your season. Which inquiries to decline depends on where your trucks can actually get.

We work through those a city at a time, because an arrival promise written for one market is wrong in the next one.

If we publish for your metro, read that next. If not, the chain here is still the right place to start, and we are happy to walk it with you against your own numbers.