Pull your last 50 callbacks, sort each into one of five causes (diagnosis, workmanship, parts, rushed job, or customer expectation), and fix the cause with the most entries first. Most companies find one or two causes account for most callbacks, and the fix is usually a process change, not a lecture.
The callback rate glossary entry covers how to measure it. This page is about what to do once you have the number. The mistake is treating every callback as a tech problem. Plenty of them are booking problems.
Step 1: sort the last 50
Take the last 50 callbacks (or the last three months if you have fewer). For each, write one cause:
| Cause | What it looks like | Usual fix |
|---|---|---|
| Diagnosis | Fixed the symptom, missed the real fault | Checklists, ride-alongs, a second opinion on repeat calls |
| Workmanship | Loose connection, leak at a fitting, bad seal | Final-test checklist with photos |
| Parts | Wrong part, failed new part, part not on the truck | Truck stock by job type, supplier returns |
| Rushed job | Tech left early to make the next stop | Longer slots for that job type |
| Expectation | Customer thought more was included | Clear scope in writing before work starts |
Count them. If 20 of 50 are "rushed job," a training program for technicians aims at the wrong target.
Step 2: work out what a callback costs you
A callback is usually unbilled, so the whole visit comes out of margin. And it doesn't just cost its own time. It takes a slot you could have sold.
Two points off the callback rate is worth about $30,000 a year in this example. Run it with your own numbers: the truck roll cost calculator gives you L.
Step 3: fix the biggest cause first
If it's rushed jobs
This is the one most owners miss. If a job type is booked at 60 minutes and routinely needs 80, the tech has two choices: run late for every customer after it, or cut corners. Many cut corners. Compare booked time to actual time on site for that job type, and set the slot from what really happens. The job duration estimate entry covers how.
If you use CrewLink, time on site is recorded when each job is marked done, and future bookings use it, so a job type that keeps running long stops being booked short.
If it's diagnosis
Pair the techs with the most diagnosis callbacks with your best diagnostician for a few days. Add a short checklist for the job types that come back most. Flag any address with a second visit in 30 days for a senior tech, not the original one.
If it's workmanship
Make a final test non-optional: run the system, check for leaks, photograph the finished work. Photos also settle the "you never fixed it" argument when it isn't true.
If it's parts
Match truck stock to the job types each truck actually runs, and send the right tech for the right job. Skills-based routing is mostly about this: a tech who can't fix it on the first visit is a guaranteed second visit.
If it's expectation
Write down what the job includes and doesn't, and get the customer to approve it before work starts. "Tune-up does not include repairs" said out loud and written on the estimate saves a lot of callbacks that were never really failures.
Step 4: measure the right way
Count callbacks per job type and per tech, using a fixed window (30 days is common), and attribute each one to the original visit. Watch the trend over a quarter, not week to week; the counts are small enough that one bad week looks like a crisis. Pair it with first-time fix rate, which catches the jobs that needed a second visit for parts even when nothing failed.
Don't punish techs for callbacks on job types that are booked too short. They will stop reporting them, and the number will get better while the problem gets worse.
What to take away
- Sort your last 50 callbacks into five causes before you pick a fix.
- Yearly cost is C × (labor + parts + the slot it displaced); the displaced slot is the part people forget.
- "Rushed job" callbacks are a booking problem: set slot lengths from actual time on site.
- Fix diagnosis with pairing and checklists, workmanship with a photographed final test.
- Attribute callbacks to the original visit and watch the trend by job type over a quarter.
Common questions
What is a good callback rate?
It depends on the trade and the job mix. Installs and complex diagnostics run higher than maintenance. Rather than chase an outside benchmark, track your own rate by job type and work on the highest one.
Should technicians be paid for callbacks?
Many companies don't pay commission or spiff on callback visits for the tech's own work, to keep the incentive pointed at fixing it the first time. If you do that, make sure slot lengths are realistic first, or you are penalizing techs for your schedule.
Are callbacks always the technician's fault?
No. Jobs booked too short, the wrong tech sent, missing parts and unclear scope all cause callbacks that no amount of tech training will fix. Sort them by cause before you assign blame.
How do I track callbacks if my software doesn't?
Add a simple flag on any job at an address that had a visit in the last 30 days for the same equipment. Review the flagged list weekly and write a cause on each one.