The Callback That Is Not Your Fault and Still Yours to Answer

Why this matters

A customer calls back three weeks after you were there. Something is wrong. It may or may not be connected to what you did. The shop that treats "is this my fault?" and "am I going to answer this call?" as the same question makes one of two expensive mistakes: it argues about cause on the doorstep and loses the customer, or it absorbs every return visit forever and never learns what its own work actually costs it. These are two separate promises with two separate triggers, and keeping them separate is the whole skill.

Two duties, and only one of them is conditional

The duty to respond is unconditional and fast. Someone bought work from you and something in that vicinity is now wrong. You go, or you route them somewhere specific. This duty triggers on the phone call, not on the evidence.

The duty to absorb the cost is conditional. It triggers only when your work created the condition, or when your written warranty or guarantee says it triggers regardless of cause. That determination requires evidence you do not have while the customer is still talking.

Conflating them is what produces the doorstep argument, because a shop that thinks answering means accepting starts protecting itself before it has looked at anything. It also produces the opposite failure, the shop that quietly eats every return because sorting them out feels petty. Both are the same mistake wearing different clothes.

The gate

One rule decides cost attribution, and it is narrower than most shops write it:

Did your work, or an omission in your work, create the condition the customer is calling about?

Not "is the fault in a system I touched." Not "did I have my hands in that cabinet." Created, or failed to catch when catching it was inside the scope you sold. That gate is the only thing separating the two cases below, and they resolve in opposite directions.

Case A: same system, not your condition

A shop replaces a failed circulator pump on a hydronic heating system in March. In November the customer calls with no heat. The tech arrives inside the shop's stated no-heat response window, spends 0.6 hours on diagnosis, and finds the circulator running normally. The fault is a failed thermostat the shop never touched, on a control circuit that was not part of the March scope.

Run the gate. The shop's work did not create the thermostat failure and catching a future thermostat failure was not inside a pump replacement. Outcome: the visit is billable and the repair is billable. The customer is told this before the tech opens his hand on the repair, not after.

Case B: different symptom, your condition

The same week, a customer from a September install calls about a knocking noise. The tech finds air in the loop. The install scope included filling and purging the system, and it was not fully purged. Time on site to purge, re-bleed and confirm quiet operation is 1.4 hours.

Run the same gate. The condition traces to an omission inside the sold scope. Outcome: no charge for the visit and no charge for the correction, whether or not the workmanship term has words in it that would technically let you argue the customer bled a valve.

Same gate, same week, same response standard, opposite results. That is what makes it a gate rather than a mood.

Responding is not an admission

Say this out loud once so a nervous tech has the words: "I do not know yet whether this is connected to our work. I am coming out either way, and I will tell you what I find before I touch anything."

That sentence does three jobs. It commits to the response, which is the thing the customer actually wants in the first sixty seconds. It refuses the cause question honestly instead of guessing. And it puts the pricing conversation after the diagnosis, where it belongs, rather than in a phone call where neither party has any facts.

What you must not do is guess in the customer's favour on the phone to calm them down. "That sounds like something we did, we will take care of it" costs you the gate permanently. You cannot un-say it after the diagnosis, and you have taught that customer that the way to get free work is to sound upset early.

If the callback report includes a gas odor, a burning smell, a shock, or water at an energized panel, the cost conversation stops entirely. Tell them to get everyone out of the building now, touch no switches, turn no lights on or off, use no phone inside, and call the gas utility emergency line or the power utility from outside once they are clear. That instruction goes first, before you check the schedule and before anyone mentions who pays. A life-safety report is not a commercial negotiation and an article that treats it as one has already failed the customer.

Worked example: what conflating the two duties costs

A shop with three field techs pulls a year of return visits: 34 return visits, averaging 1.1 hours on site each, so 37.4 hours of response time total. They run every one of those 34 through the gate after the fact:

Twelve plus fourteen plus eight is 34, and 12/34 is 35%, 14/34 is 41%, 8/34 is 24%. The three shares sum to 100%.

Under the gate as written, 12 visits are the shop's cost. At the 1.1 hour average that is 13.2 unbilled hours across the year, all of it technician time the shop absorbs.

Under the habit of eating everything to avoid the argument, all 34 are the shop's cost: 37.4 unbilled hours. The difference is 24.2 hours a year, and the ratio of absorbed hours under the habit to absorbed hours under the gate is 37.4 to 13.2, or about 2.8 to 1.

Both figures are the same currency, unbilled technician hours on site, so the ratio is honest. Neither is a profit number: the 24.2 hours you stop absorbing do not become 24.2 hours of margin, they become 22 visits you can decide about individually, some of which you will still waive on purpose because the customer is worth it. That decision is worth making one at a time. The point of the gate is not to collect on all 22, it is to know that 22 exist.

Now read the 12 back. Those 12 are the only visits that tell you anything about your own quality. Mixed into 34, a shop looking at "our callback rate" is reading a number that is 65% other people's failures, and any conclusion it draws about crew performance from that blend is drawn from the wrong denominator. See related: How to Cost Callbacks and Rework Honestly, and Callback Cause: Installer vs Product vs Customer for the attribution method itself.

What the record has to carry

The gate only works if something written can answer it a year later, when nobody remembers. Three fields, captured at the callback, not reconstructed:

• The condition as found, in the tech's own words, before any correction. Photograph the state, not just the fix.
• What the original scope did and did not include, pulled from the original job record rather than from memory. This is the field that decides most disputes, and it is the one nobody writes down.
• The attribution call and who made it, so a supervisor reviewing a pattern later knows whether the classification came from the tech who did the original work.

That last one matters more than it looks. A tech grading his own work is not a neutral read, and a shop whose attribution is self-reported will see its own-work share drift downward over time for reasons that have nothing to do with quality.

When the gate does not decide it

Three conditions override the attribution result, and each one has a specific reason:

Your written terms are broader than the gate. If you sold a satisfaction guarantee or a stated period of no-charge return visits regardless of cause, that document governs and the gate is irrelevant. Read what you actually wrote, not what you meant. Warranty and guarantee obligations, including what a written promise can and cannot disclaim, vary by state and several rules turn on whether the customer is a consumer or a business, so have your own attorney review the terms you issue rather than resolving it from a checklist.

The finding is a safety condition, whoever caused it. A cracked flue, a scorched conductor, an unsupported gas line, a missing bonding jumper: you write it down and you say it plainly, in writing, whether or not the customer is currently arguing about a bill. That duty does not trade against the commercial conversation and it is not softened because you are annoyed about the call.

You cannot tell. A genuine "the evidence does not resolve this" is a real outcome and it is not a tie you break in your own favour. Say what is unresolved, say what you did, and pick a split you can defend consistently across customers rather than one that reflects how the phone call went.

How to verify you got this right

Pull your last twenty return visits and check four things.

1. Did every one get a response inside your stated standard, including the ones that turned out to be billable? If billable calls are systematically slower, your dispatch is quietly pre-judging attribution before anyone has looked.
2. Does each record name the original scope, or does it only name the fault? A record that only names the fault cannot answer the gate.
3. Do the attribution shares hold up if a second person re-reads the ten oldest? Disagreement above roughly one in five says your gate is being applied as a feeling.
4. Did anyone promise a cost outcome on the phone? If your intake notes contain "told customer no charge" before a diagnosis exists, that is the leak, and it is a training fix rather than a policy fix.

References

• See related: How to Cost Callbacks and Rework Honestly
• See related: Deciding Whether a Callback Is Warranty or Billable
• See related: Callback Cause - Installer vs Product vs Customer Decision Tree
• See related: The Warranty Language You Put in Writing and What It Commits You To