Delivery Leak · The Untracked Redo
You Went Back and Redid That Job for Free. Nobody Wrote Down Why.
In This Article
Same truck. Same tools. Same address you were just at three weeks ago. Except this time there's no invoice, because the job's already paid for — you're just going back to fix it. You tell yourself it's one bad day. You get in, get it right, get out, and move on to the job you're actually getting paid for.
Here's what that "one bad day" actually is: a repeating cost with nobody's name on it. Not because you did anything wrong going back — that's usually the right call. Because nobody in your business writes down which job it was, what kind of work it was, or why it came back. So when it happens again next month, it feels like a new problem instead of the same one, twice.
The leak: what happens after "done"
Most of the ways a local business loses money happen before the sale — a call that goes unanswered, a quote that goes out too slow, a customer who goes quiet before day one. This leak is different. It happens after the invoice, after the "thank you," after the job was already marked closed. You did the work. You got paid. And then you had to do it again for free.
That's not the yes-to-day-one silence that kills a booking before it starts, and it's not the mid-job communication gap that costs you a referral. Those are real leaks too, and worth fixing — but this one only opens once a job is already finished and paid for, and somebody has to go back anyway.
Why it feels like bad luck (and isn't)
A callback doesn't feel like a pattern, because nothing captures it as one. The job gets marked "handled" a second time, and that's the end of the paper trail. No column says which tech, no column says what kind of job, no column says why. So three redo visits caused by the exact same thing — a bad batch of material, a step someone's skipping, a job type that's underscoped for how often it comes back — read as three unrelated bad days instead of one fixable cause repeating.
That's the whole mechanism. Not carelessness, not bad luck — an absence of the one piece of paper that would make the repeat obvious. You can't fix what you never wrote down.
What the redo actually costs
Nobody sends an invoice for a callback, which is exactly why it never gets counted as a real cost. But the hours are real, the drive time is real, and the materials are real — they just get quietly absorbed into overhead instead of landing on a line item anyone looks at.
As an illustration, not a claim about your numbers — plug in your own: say a redo visit runs your tech 2.5 hours between drive time and the fix, plus $75 to $100 in materials. Two of those a month, at a fully-loaded labor rate most trades already know off the top of their head, is real money — often a few thousand dollars a year — moving straight out of margin with nothing to show for it on paper.
Run your own versionHours per redo × your loaded labor rate, plus average materials, times how many free redos you can remember from the last 30 days. Whatever number you land on, that's the size of what "just eating it" has been quietly costing you — every month, on repeat.
The trust cost, not just the labor cost
The hours aren't even the full bill. A customer who needed a free fix doesn't remember the job as a success story — they remember it as the one that went wrong, that you happened to make right. That's a worse position than the job just going smoothly the first time, even though you spent extra time and money trying to land back at "fine."
That matters more than it looks like, because keeping a customer is the cheap side of the ledger compared to replacing one: losing a customer costs 5 to 25 times more to replace than keeping one costs — Harvard Business Review, citing research on customer retention economics. A callback that quietly erodes trust isn't just an afternoon of free labor. It's a small bet against ever having to acquire that customer again.
The 3-Field Callback Log
The fix isn't "stop doing redos" — sometimes the work just needs a second visit, and eating it is usually the right call for the relationship. The fix is a simple 3-field framework: writing down enough about each one that the pattern can find you, instead of you having to remember it.
Who — which tech did the original job
Not to keep score. If the same name shows up three times, that's a training conversation, not a talent problem — most techs redoing the same mistake just haven't been shown the fix yet.
What — the job type or system
Water heater, roof flashing, a specific service line — whatever the category is. If one job type keeps coming back across different techs, that's not a person problem at all. That's a material, a supplier, or a pricing problem: the job may be scoped too thin to do right the first time at the price you're charging for it.
Why — your best-guess reason
One line, written the day of the redo while it's fresh — a part failed, a step got rushed, the customer misused something you didn't explain clearly. You don't need certainty. You need thirty days of honest one-liners sitting next to each other where you can actually see them.
A notebook or one spreadsheet tab is enough — this doesn't need software. Log every free redo for a month, then read the Who, What, and Why columns straight down. Most owners find one repeatable cause sitting right there, hiding in plain sight, the first time they actually look.
This is the same instinct behind the guarantee that wins the job before the estimate — except that's the formal promise you make out loud, and this is the informal habit of quietly eating a redo whether or not you've ever written a warranty down. If the yes-to-day-one gap or the mid-job silence sound more like your leak than this one, the Delivery Leak and the Silent Job cover those two windows — this piece is strictly about what happens after the invoice, when you go back anyway.
Imagine the version where you've been running the log for a month. Three redos, and two of them turn out to be the same $40 part from a supplier you switched to in the spring. One phone call, one supplier swap, and most of that free labor disappears going forward — not because anyone worked harder, but because the pattern finally had somewhere to show up. The alternative is running the exact same leak a year from now, still calling it bad luck.
Frequently Asked Questions
Almost always because the same cause is repeating and nobody's written it down anywhere to notice. A callback for a bad valve in March and a callback for a bad valve in June look like two unrelated bad days if nothing connects them — but they're the same problem twice, whether that's a material, a step someone's skipping, or a job type that's underpriced for how often it comes back. Without a log, the pattern stays invisible even though it's sitting right there in your job history.
You don't need software for this — a notebook or a spreadsheet tab works. Every time you send someone back for free, write down three things: who did the original job, what kind of job or system it was, and your best guess at why it came back. Do that for 30 days and look for repeats in any one column. A real pattern almost always shows up before the month is out.
Nobody sends an invoice for a callback, so it never shows up as a cost on paper — but the labor hours, the drive time, and the materials are all real. As an example, not a measured average: a callback that runs a tech 2.5 hours plus $75 to $100 in materials, twice a month, is a few thousand dollars a year moving straight out of your margin with no line item to show for it. Run your own numbers with your real hours and rates — the shape of the math holds either way.
Log the reason for every redo for 30 days without judging it in the moment, then look at what repeats. Most patterns turn out to be a training gap, a material or supplier issue, or a job type that's priced or scoped too thin to do right the first time — not a person being careless. Treat the log as a way to find the one fixable cause, not a scoreboard on your techs, and it stays a process conversation instead of a blame conversation.
If the original job wasn't done right, most owners eat the redo to protect the relationship — and that's usually the right call in the moment. The mistake isn't eating the cost once; it's eating the same cost repeatedly without ever writing down why, so you never find the one fix that would stop most of them from happening again.
