Your booking rate is worst on the hottest afternoon of the year
The hours you most want to convert are the hours you convert least, and almost nobody has looked at it by hour.
Pull the call log for last summer and break it down by hour and day of week. Count answered, abandoned and sent to voicemail. The pattern is usually the same narrow window across every hot week, and usually nobody in the building has seen it.
The fix is rarely a website change. It is staffing that window, or an overflow service with a real script and access to book, or an automatic text reply that goes out within seconds of a missed call and keeps the person from dialing the next result.
Set a callback standard for missed calls and hold someone to it. Minutes, not hours. A homeowner with no cooling in July is calling three companies and stopping at the first live voice.
Then test it yourself. Call your own number at four on a hot Friday and at nine on a Saturday night. Whatever happens is what happens to every paid click you bought that week.
Dispatch by drive time, and teach the phone to say no to Arlington
Booking a job you cannot reach on time costs more than declining it politely.
The metro spans Dallas, Tarrant, Collin, Denton, Rockwall and Ellis counties, and Fort Worth is roughly an hour from Dallas without traffic. A job accepted at the outer edge tends to arrive late, run short and end in a review that follows you around.
Define the boundary by drive time from where trucks stage, not by county lines, and enforce it at the phone. Give whoever answers a short script: a polite decline, a referral to somebody you trust, and a note in the log.
Log every decline with its ZIP code. After a season that list tells you exactly where a second truck or a satellite location would pay for itself, and it is better evidence than anybody's hunch.
Where you do accept edge jobs, price the drive honestly and set the arrival expectation wide enough to keep. A window you meet is worth more than a window that sounded better on the phone.
A changeout is a three bid decision, so the follow-up is the sale
Most replacement quotes are lost after the visit, not during it.
A homeowner replacing a system is spending an unplanned sum and almost always getting other numbers. The comfort advisor who leaves without a written proposal in the customer's hands is relying on memory to compete against paper.
Send the proposal the same day with clear options and financing terms attached. Call the next morning. Make one more contact about a week later that speaks to the specific thing they hesitated over, whether that was ductwork, capacity or timing.
Set a defined stopping point so follow-up does not become nagging, and log the reason for every loss. Price, timing, chose a competitor, decided to repair instead. Six months of those notes is the most useful document in the business.
Make the person who gave the quote the person who follows up. Handing it to an administrator drops the relationship and the technical detail at the same time.
Five numbers, from answered to collected, and most shops track one
Lead counts hide the leak, so measure every step of the ladder instead.
Track five: inquiries answered or returned within your standard, jobs booked, jobs actually run without a cancellation, jobs sold, and invoices collected. Compute the ratio between each pair.
Break those ratios by lead source and by county. A source that produces plenty of booked calls and almost no sold replacements is not the same asset as one producing fewer, better inquiries, and a lead total treats them identically.
The worst ratio in the ladder is your project for the quarter. Usually it sits in one of two places: booked jobs that cancel before the truck arrives, or quotes that went out and were never chased.
Review it monthly, and compare against the same month last year rather than against last month. Demand here swings hard between a heat wave and a mild week, and month-over-month comparisons mostly measure the weather.
Listen to every call where somebody asked about a rebate
The incentive question is where an untrained phone answer quietly costs you credibility.
Retail electricity here is deregulated. The household picked a retail electric provider, while the wires, the meter and outage response belong to the transmission and distribution utility. So there is no single utility whose rebate you can confirm on the phone.
A confident wrong answer creates a hard conversation later, usually after the equipment is installed. Script it instead: acknowledge that programs exist, explain the split in one sentence, say that eligibility depends on their own provider and is worth confirming with them, then move to scheduling.
Pull thirty to fifty recorded calls a quarter and listen with a notepad. The rebate answer is one finding. The price question is usually the other, and most shops discover their team handles it worse than anyone assumed.
Do the same for storm calls. After a hail event people ask questions about damage and insurance. Keep your team to what they can inspect and repair, and route claim questions to the homeowner's own insurer, since Texas restricts what a contractor may say about a claim and the boundaries are worth confirming with counsel.
Seasonality ruins split tests, so change big things and compare like weeks
One metro and one trade rarely produce the volume for small-effect testing, and the weather moves the baseline underneath you.
A test that hopes to detect a small lift in form completions needs traffic most single-market HVAC companies will never have. Running it anyway produces a number that looks like a result and is mostly noise.
Change things large enough to see. Whether a human answers after six. Whether a missed call triggers a text within a minute. Whether financing appears in the proposal itself rather than on request. Whether the form asks for a photo of the data plate.
Compare against the same weeks a year earlier, and against a second market if you run one. Fort Worth and Dallas can serve as rough controls for each other when the change is rolled out to one at a time.
Keep a dated log of every change. When July looks strange, the log is what separates a real improvement from a heat wave, and it stops the same experiment from being run twice.
Questions we actually get
- Is an after-hours answering service worth paying for?
- Usually, if the service can actually book into your schedule and follows a script you wrote. A service that only takes a message competes poorly against a competitor who answers live. Measure it directly: count the after-hours calls it converted into run jobs, not the calls it handled.
- What conversion rate should we expect from our website?
- There is no honest benchmark to hand you, because the number depends on your traffic mix, your ad targeting and how much of your volume is repeat business. Measure your own baseline over a full year, then improve against it. Anyone quoting an industry figure for your business is guessing.
- How fast do we need to respond to a form submission?
- Faster during a heat wave than during a mild March. Set a standard you can hold, keep it in minutes rather than hours for service inquiries, and measure compliance rather than intention. A homeowner without cooling has already contacted somebody else by the time a next-day reply arrives.
- Should we ask for reviews, and when?
- Yes, at the moment the customer is most satisfied, which is usually when the technician is packing up and the house is cooling down. Make it one link, sent by text, from the person who did the work. Never offer anything in exchange, and never write the review for them.
- How do we tell which marketing channel is actually working?
- Tag every inquiry with its source at the first touch and carry the tag through to the invoice. Then compare sold revenue by source, not lead counts. Most shops that do this once find one channel they were underfunding and one they had been paying for out of habit.
