Every Review You Have Is From July: The HVAC Review Velocity Problem

Sort Your Reviews by Date

Open your Google Business Profile and sort your reviews oldest to newest. If you run an HVAC company anywhere with a real summer, you already know the shape you will see. A dense cluster in late June and July. A smaller bump after the first hard freeze. Then long flat stretches where nothing came in at all. Fourteen reviews in five weeks, then eleven weeks of silence. Most owners look at that and see a busy season. What they are actually looking at is a record of when their team had time to ask.

The Leak: Your Profile Goes Quiet Right When Buying Restarts

A homeowner shopping for a furnace tune-up in October is not counting your reviews. She is reading the top three and checking the dates. A profile whose most recent review is fourteen weeks old reads like a company that got slow, changed hands, or stopped caring. She has no way to know it only means your techs were busy.

She is not an unusual customer, either. Roughly 92% of consumers read reviews before buying. Your star rating survives the off-season fine. Your recency does not. The October jobs you never hear about are going to the shop whose last review posted on Tuesday.

Why It Happens: The Ask Is Attached to a Person, Not a Process

Your review count is not tracking how good you are. It is tracking job volume multiplied by a small, unstable number: the share of completed jobs where somebody actually asked.

For most shops that share sits in the low single digits, and it moves for reasons that have nothing to do with customer satisfaction. Run 200 calls in July at a 4% ask-and-convert rate and you get eight reviews. Run 60 calls in October at the same rate and you get two. Nothing got worse. The denominator shrank.

The second half is quieter. Whole categories of work never get an ask at all. Maintenance-plan tune-ups, warranty callbacks, second-opinion diagnostics, a filter swap on a Tuesday afternoon. Techs read those as "not a real job," so nothing goes out. Those are exactly the visits that fill your shoulder months. The result is a review profile that only reflects your emergencies, written during the eight weeks a year your team has the least time to do anything about it.

The Fix: Five Steps to a Review Floor

1. Count reviews per job, by month

Export the last twelve months of reviews from your Google Business Profile: date and rating, nothing else. Next to it, pull completed job counts by month out of your dispatch software. Two columns, twelve rows. Divide one by the other.

That number is your real ask-and-convert rate, and looking at it monthly tells you something an annual total hides completely. Most shops find the rate itself sags in peak months, when everyone is slammed. So the busy season is producing more reviews on worse execution, and the slow season is producing almost none on execution nobody is measuring at all. You cannot fix either one until you can see both.

2. Set a monthly floor, not an annual goal

"Get more reviews" is not a target anyone can act on. A floor is. Pick a number you can defend out loud — four a month is a fair starting point for a shop running sixty-plus calls in a slow month — and treat any month under the floor as a miss you talk about in the Monday meeting, the same way you would talk about a callback or a comeback.

Annual totals let a dead October hide behind a great July. Floors surface the gap while you can still do something about it, and they give whoever owns the process a number they can actually hit.

3. Widen what counts as a reviewable visit

Any visit where a customer's problem got smaller is reviewable. Maintenance-plan tune-ups. Warranty callbacks that went well. No-heat calls you turned around the same night. Second opinions where you told a homeowner she did not need the $9,000 system the other company quoted — those produce some of the best reviews you will ever get, and almost nobody asks for them, because the tech does not feel like he sold anything.

He did. He sold trust, and that is the thing the next homeowner is reading for. Write the list of reviewable visit types down and post it in the shop. "Use your judgment" reliably produces zero.

4. Fire the ask off the system, not the technician's memory

The ask should trigger when the job status flips to complete in your dispatch software, not when a tech remembers on the way to the next call. Text, not email. Send it inside thirty minutes, while the house is cooling down and the relief is still fresh.

Put the tech's first name in the message: "Was Dave able to get you taken care of today?" A named ask converts better than a branded one, because the customer is rating a person she just met rather than a company she barely knows.

Then put one question in front of the link. Did we get this right? A yes goes to the Google link. A no goes to your service manager's phone, and somebody calls that customer within the hour. This is not about burying bad reviews. It is about fixing a job before the customer writes about it — and shops that skip this step quietly ask less often, because they are afraid of what might come back.

5. Load the quiet months on purpose

Your October and March review counts get decided in August and February, when tune-ups are scheduled. Book maintenance visits into the slow months deliberately instead of letting them land wherever the customer suggests. Those visits become the engine that keeps your profile alive when the phone is not ringing on its own, and they cost you nothing extra — you were going to do the work anyway.

The Proof

Here is why a floor beats a push. Rating follows volume. Ask on 40% of your jobs instead of 4% and your average climbs, because satisfied customers are the ones who need prompting. The unhappy ones find the review form without any help from you.

A one-star improvement in average rating correlates with roughly a 12% lift in conversion.

For a shop closing sixty jobs a month at a $2,800 average ticket, that is not a branding win. That is another install or two every month, pulled out of work you have already done and already paid for.

What This Looks Like When It Runs Itself

Everything above works with a spreadsheet, a phone, and one person disciplined enough to run it every week for a year. The design is not the hard part. Month seven is the hard part — the heat wave hits, the person running it gets pulled onto dispatch, and the ask goes quiet through the rest of the summer.

That is the piece James, Tykon's AI sales agent, takes over. The ask fires on job completion whether or not anyone remembers. The routing question runs itself. Unhappy customers land on a manager's phone in minutes instead of on Google. And once a review posts, the same system asks that customer who else they know with a system limping into next season.

If you do one thing today, do step one. Export twelve months of review dates, put job counts beside them, and find your worst month. If you want a second set of eyes on that spreadsheet, send it over and we will read it with you.