Callbacks are eating your margin. Here's how to see them clearly.

You finished the job, invoiced it, moved on. Two weeks later the customer calls back: it’s not fixed. Now you’re sending a truck out again, for free, while a paying job waits. Most shops know this hurts. Few know how much, because callbacks are nearly invisible in the way most shops keep their books.

What a callback actually costs

A callback means you pay twice to earn once. Run the math on a single one:

• The truck roll. Fuel, wear, insurance, the same cost as any dispatch, with no invoice at the end.
• The labor. Two or three hours of a tech’s day, paid out, billed to nobody.
• The parts. Whatever it takes to make it right this time comes off your shelf.
• The displaced job. That slot could have held a paying call. The revenue you didn’t earn is the biggest line, and it never shows up anywhere.
• The customer’s patience. They were happy once. Now they’re watching. A second miss and they’re calling someone else, and telling their neighbor why.

If your average job nets a few hundred dollars and a callback costs you that much again in expense and lost capacity, a handful of callbacks a month can quietly take a real bite out of your margin. You don’t need a study to see it. You need your own numbers.

Why callbacks hide

Here’s the trap: in Jobber or Housecall Pro, a return visit usually gets logged as a brand-new job. Nothing connects it to the original. The dispatcher is busy, the tech is busy, and “redo of the Hendersons’ compressor job” becomes just another line in this week’s schedule. Your job count goes up. Your software thinks business is good.

So the first fix costs nothing: tag your redo visits. Pick one convention, a job type called “Callback” or “Warranty,” or a tag like redo, and make it a habit at booking. Whoever takes the call asks one question: “Have we been out for this before?” If yes, it gets the tag and a note linking it to the original job. That’s the whole system. Without it, every number that follows is a guess.

How to read the numbers

Once you’re tagging, look at three things, in this order:

1. Rate, not raw count. Five callbacks in a slow month is a very different story than five in a month where you ran 200 jobs. Divide callbacks by completed jobs. Most shops land somewhere in the low single digits; what matters more than the level is whether it’s yours and whether it’s moving.
2. Trend over time. One bad month happens. Three months of a climbing rate is a pattern. Watch the line, not the data point.
3. Clusters. When the rate moves, slice it: by tech, by equipment type, by job type. A general rise tells you something is off. A cluster tells you where to look.

This is the same discipline behind the 6 numbers every shop should watch: a rate you can trust beats a feeling you can’t. The same rework math shows up everywhere it’s tracked, failed-inspection returns in electrical, free re-service visits baked into a pest-control guarantee, and the part-return-vs-callback line that defines appliance repair.

What a cluster actually means

Say one tech has three callbacks this month and everyone else has one or none. The tempting read is “he’s the problem.” Resist it. A cluster is a question, not a verdict.

The right first move is to pull those three jobs and look at what they have in common, and the tech is only one of several things they share:

• Same equipment? Maybe he drew three of the same finicky unit, or a brand with a known issue.
• Same job type? If they’re all reassemblies of a repair only he gets assigned, the cluster may belong to the job, not the man.
• Same parts batch? A bad run of capacitors or fittings will follow whoever installed them.
• Same week? Look at his schedule. If he was running seven calls a day in a heat wave, the schedule made those callbacks, and it would have made them for anyone.

Sometimes the answer is a training gap, and now you can fix it specifically, with a real conversation about a real pattern, not a vague talking-to. But if you skip the diagnosis and go straight to blame, you’ll demoralize a good tech and miss the bad parts batch that’s about to give you three more callbacks under someone else’s name.

The honest payoff

A falling callback rate shows up twice. First in margin: fewer free truck rolls, fewer eaten parts, more slots holding paying work. Second in referrals: the customer whose fix held is the one who hands your number to a neighbor. Neither happens overnight, and neither requires new software, just the tag, the rate, and the habit of diagnosing before deciding.

If you’d rather not build that spreadsheet, watching the rework rate is exactly what Guidepost does. It reads Jobber or Housecall Pro and QuickBooks, tracks callbacks as a rate, and when a cluster shows up it frames it the way you just read: here is what these jobs share, here is what to check before you talk to anyone. Every number links back to the jobs behind it, so a hard conversation starts from a real pattern, not a hunch.

In the sample shop, that flag reads “Dwight has 3 callbacks this month.” Read a full sample digest to see how it’s worded before it lands in your inbox.