A callback does not show up as a cost. It shows up as a truck roll that already happened, parts already installed, and a morning that bills nothing new. Nothing about it looks like a loss. It looks like a normal Tuesday.
Published guidance puts an acceptable rate at 2% to 5% depending on who is writing, with top-quartile shops at 1–2% S-CB-01. The distance between those two is already real money. The distance between either of them and a rate nobody is counting is what this page is about.
What one callback costs
fully loaded cost of a single callback (service call to install), including the return trip and the technician time that bills nothing S-CB-01
That figure is not “an hour of wage”. It is what a truck roll costs once burden is loaded in, which is the same loaded rate every other number on this site is built from.
Multiplied out
Take a shop running 500 jobs a month at a 10% callback rate: 50 callbacks, at roughly $650 each, is about $32,500 a month S-CB-01, recurring, for as long as the rate stays where it is.
The same shop at a top-quartile 1.5% would spend roughly $4,900. The difference does not come from pricing, marketing or headcount. It sits inside execution that is already being paid for.
This is an illustration on published averages, not a projection about your business. Your own volume and your own rate are the only numbers that would make it yours.
Why it survives
A $32,500 loss printed on a P&L gets fixed in a week. The same loss spread across fifty invisible truck rolls does not, because measuring it requires tagging a job as “return visit, same issue” at the moment it happens, and almost no shop’s paperwork forces that tag.
The reserve question follows directly
A shop that has never measured its callback rate has, almost by construction, never sized a warranty reserve against it. The published recommendation is 2–3% of revenue set aside for rework S-CB-01. A reserve sized on an assumed 2% rate, when the real rate is 10%, is undersized by roughly the same multiple as the cost above. The gap surfaces the way the callback cost itself does: not as a tracked number, but as a cash-flow surprise in the month a run of bad installs comes due.
One tag, thirty jobs
For the last 30 to 60 completed jobs, mark whether each generated a return visit for the same issue inside a fixed window (30 or 90 days is standard). Divide return visits by total jobs. That is your rate. Multiply by the per-callback cost and the monthly figure stops being invisible.
“Some callbacks happen” is true and always will be: even top-quartile shops run 1–2%, not zero. The narrower and more useful statement is this one: without that tag, your rate could be 2% or 20% and you would have no way to tell which.