HVAC callback queue: Why promised callbacks never happen
Two things called a callback
In HVAC, "callback" usually means a technician going back to a job that was not fixed the first time. ACCA benchmarks that rate at 3 to 8% of jobs, and each service manager tracks it. This post is about the other callback, the one a CSR promises a homeowner: "Let me check with dispatch and call you right back." Nobody tracks that one, and in most shops nobody owns it.
Call it the customer callback to keep the two apart. The technician callback costs a truck. The customer callback costs the job.
The third bucket
Missed-call math covers the calls nobody answered. Booking-rate math covers the calls answered but not booked. There is a third bucket neither of them names: Answered, promised a callback, never called. It hides from both numbers because the call was answered and the booking is still, on paper, pending. On a report, a pending callback looks like a lead in progress. To the homeowner, it looks like silence.
For an operator taking 8,000 calls a month, if one call in ten ends with a promise and a third of those promises are never kept, that is more than 250 homeowners a month who reached you, were told to wait, and heard nothing.
A task with no alarm against a phone with one
Callbacks do not fail because CSRs are careless. They fail because of what a promised callback is: A task written on a sticky note or a CRM field, with no alarm, competing for the same person's attention as a phone that rings. The phone wins each time, because the phone is a homeowner right now and the sticky note is a homeowner from 40 minutes ago.
By 3 PM the note is under four others. By 5 PM the CSR goes home and the note becomes tomorrow's problem, and tomorrow has its own phone. A promise with no timer is a promise with no owner, and a promise with no owner is a lead you answered and then abandoned.
The window is shorter than the queue
Harvard Business Review's 2011 study of 2,241 companies found that 23% never responded to a lead at all, and that those who did took an average of 42 hours. Hathaway, Emadi, and Deshpande, writing in Management Science in 2021, found that customers value a callback only when it arrives inside a short window; past that window it is worth little more than no call at all.
Homeowners will take the callback if it is real. Nextiva's Customer Patience Benchmark, from May 2026 with 400 respondents, found 75% prefer a callback to waiting on hold. A TCN survey found 48% expect that callback within 30 minutes. Homeowners are not asking for an answer on the spot. They are asking for the promise to be kept inside half an hour.
Two numbers that make the queue visible
Metric | Definition | What it takes to measure |
|---|---|---|
Promise rate | Answered calls that end with a promised callback, as a share of all answered calls | A disposition code for "callback promised" on each call |
Callback completion rate | Promised callbacks that were made, as a share of all promised | The promise logged as a task the moment it is made, closed by the outbound call record |
Callback SLA | Minutes from promise to completed callback, at the median and the 90th percentile | Timestamps on the task and on the outbound call |
Most operators cannot produce the second or third number today, because the promise was never recorded as anything a report can count. That is the first fix: A disposition code and a task with a timer, created on the call, not after it. Once the numbers exist, the targets follow. Completion rate should approach 100%, because a promise is a promise. The SLA target is the TCN figure, 30 minutes, measured at the 90th percentile so a good median does not hide the callbacks that took a day.
Most callbacks should not exist
The second fix is upstream of the queue. Listen to 20 calls that ended in a promise and sort them by what was promised. Most are promises made in place of a decision the CSR could have made on the call.
What the CSR said | What should have happened |
|---|---|
"Let me check with dispatch on availability" | Book the first open window from the live board |
"I need to ask a tech about that noise" | Book the diagnostic; the tech answers on site |
"Someone will call you with pricing" | Quote the diagnostic fee and book; equipment pricing belongs on a booked comfort advisor visit |
"The person who handles that is out today" | Take the details, set a specific callback time, and log it with a timer |
What remains after that sort is the real queue: Warranty lookups, commercial quotes, a homeowner who asked for a specific person. That queue needs an owner who is not on the inbound line. A traditional answering service makes the problem worse, because a message taken is a callback manufactured. An AI intake agent can book the first three rows on the call and hold the fourth as a timed task it works itself or hands to a CSR with the context attached.
Count the promises before you count the bookings. The bookings you are missing are hiding in the promises you never kept.
Why do HVAC companies never call back?
Rarely out of indifference. A promised callback is a task with no alarm, written on a note or a CRM field, competing for the same CSR's attention as a phone that keeps ringing with new homeowners. The live call wins each time, the note sinks under others, and by closing time it belongs to tomorrow, which has its own phone. Without a timer, an owner, and a completion metric, promised callbacks fail by default.
How long should a business take to return a call?
Inside 30 minutes. A TCN survey found 48% of customers expect a callback within that window, and research published in Management Science in 2021 by Hathaway, Emadi, and Deshpande found that callbacks are valued only when they arrive quickly; a late callback is worth little more than none. Set the target at 30 minutes and measure it at the 90th percentile, so a fast median does not hide the calls that took a day.
What is a callback completion rate?
The share of promised callbacks that were made, measured over a period. If CSRs promised 200 callbacks last month and 140 outbound calls closed those promises, the completion rate is 70%. Measuring it requires logging the promise at the moment it is made, usually as a disposition code plus a task with a timer, and closing the task from the outbound call record. The target should be close to 100%.
Do customers prefer a callback or waiting on hold?
A callback, by a wide margin, as long as it arrives. Nextiva's Customer Patience Benchmark from May 2026, surveying 400 people, found 75% prefer a callback to waiting on hold. The preference comes with a condition: A TCN survey found 48% expect the callback within 30 minutes. Offering a callback and then missing that window is worse than the hold, because the customer stopped waiting on the strength of a promise.
Why do callbacks never get made?
Because they are recorded as notes rather than as tasks with a deadline, and because the person who promised them is also the person answering new calls. Most promised callbacks are also unnecessary: The CSR could have booked the appointment or quoted the diagnostic fee on the call instead of deferring. Fix the two together: Book on the call whenever possible, and give the remaining promises a timer, an owner off the inbound line, and a completion metric.
About Revin
Stop losing HVAC jobs in the callback queue
Revin answers the call, books the job live, and runs the follow-up calls your CSRs promised, into ServiceTitan or your CRM.
