The Callback Rate Rose When the Tech Was Hurting

Why this matters

When a good tech's callbacks climb, the usual explanations are carelessness, a bad batch of parts, or a tech with one foot out the door. Sometimes the explanation is that he is hurt and has not told anyone, and the callbacks are the procedure bending around the pain. A shop that treats that as a quality problem coaches, writes up, and eventually loses a tech who needed a different route for a month. This case was reconstructed afterwards from the shop's own records, and the record contained the answer three months before anyone asked.

The number that started it

A 14-person HVAC shop reviews callbacks per tech each quarter. One tech, Luis, eight years in and one of the shop's best diagnosticians, came back from the second quarter like this:

• First quarter: 140 calls, 4 callbacks, 2.9 percent.
• Second quarter: 150 calls, 17 callbacks, 11.3 percent.
• Rest of the crew, second quarter: 3.5 percent combined.

Roughly a fourfold rise for one tech, against a crew that did not move. The owner's first theory was a new condensate pump the shop had started stocking in April. The second was that Luis was interviewing elsewhere and coasting. Before talking to him, the owner pulled the records to see which theory they supported. Neither, it turned out.

Reading the record backwards

Split by job type. The 150 second-quarter calls split into 60 that put Luis in an attic, a crawlspace, or on his knees at low equipment, and 90 that did not. Of the 17 callbacks, 13 were on the first group and 4 on the second.

The everything-else rate moved from 2.4 to 4.4 percent, close to the crew's 3.5. The attic, crawl and low-work rate moved from 3.6 to 21.7 percent, about six times higher. Whatever changed, it changed in one kind of position, not across his work.

Check the theories against the split. The new condensate pump went into ground-level and attic installs alike, and the crew's callbacks on the same pump were normal. A tech coasting before a move would be expected to slip everywhere, not in one posture. Both theories predicted a spread the record did not show.

Rule out the job type itself. A third possibility was that the attic work had got harder: a new subdivision with tight attics, or a run of older air handlers. The owner checked the other techs' attic and crawlspace work over the same quarter. Dan, who had picked up several of Luis's attic calls since March, had 2 callbacks on 41 attic, crawl and low-work jobs, about 4.9 percent, in line with his own ground-level work. The jobs were the same jobs. Only the tech doing them had changed, which pointed at the tech rather than the work, and that is a reason to ask a question, not to reach a conclusion.

Read what the callbacks were. Of the 13 position-related callbacks, 7 were condensate overflows where the trap had not been primed or cleared, 3 were panels or service doors left with screws missing so the safety switch did not seat, and 3 were evaporator coils noted as clean that the next tech found dirty. Every one of the 13 is a step done at the bottom or back of a unit: kneeling at the trap, reaching to the lower panel, getting low enough to see the coil face.

Read the time on site. Luis's median time on site for attic jobs dropped from 74 minutes in the first quarter to 51 in the second, a cut of 23 minutes. His times on ground-level jobs were flat.

Read the documentation. The shop requires a photo set on every attic air handler: the trap, the coil, the closed panel. Luis completed it on 92 percent of first-quarter attic jobs and 48 percent of second-quarter ones. The written notes on attic tickets got about half as long.

Find the start. The owner went looking for a date and found it outside the ticket system:

TIMELINE REBUILT FROM RECORDS, TECH: LUIS
Mar 14  Job 4471 attic air handler. Team chat:
        "slipped off a joist lol, all good"
Mar 15  Swaps 2 attic calls with Dan (dispatch)
Q1->Q2  Attic median on site 74 -> 51 min
Q1->Q2  Attic photo set complete 92% -> 48%
Apr 22  First callback: overflow, trap not primed
May     Attic ticket notes about half as long
Jun 30  Q2 closes: 17 callbacks of 150, 11.3%
Jul 3   Owner asks. Knee and low back since Mar 14
Jul 3   Injury report opened, 111 days after

Nothing in the ticket system recorded an injury, because nobody reported one. The only trace was a joking chat message and a call swap the next morning. That gap is part of the finding.

What pain does to a procedure

The callbacks were not random errors. They were the specific steps a body in pain negotiates away first, and the pattern is predictable enough to look for in any trade.

• The lowest or furthest step goes. Kneeling at a condensate trap on a hurt knee, or reaching to the back of a crawlspace with a hurt back, is the step that gets "probably fine". Here that was the trap, 7 of the 13.
• The second trip goes. Going back down the attic ladder for the right screws or a coil cleaner costs a climb, so the panel goes back with what was in the pocket. That was the 3 missing-screw callbacks.
• The look that needs a bad posture goes. Inspecting a coil face means getting low with a light. Writing "coil clean" from a standing glance is the shortcut, and it was the 3 coil callbacks.
• The documentation goes. Photos of the trap and coil require the same posture as the work. A falling photo rate is often the first visible sign, because it costs nothing to skip until a callback.
• Time compresses. A tech in pain in a hot attic wants out, and 23 fewer minutes on site is the procedure being shortened, not sped up.

This is the cumulative-strain story seen from the quality side: the injury was a single slip, but what turned it into three months of rework was working around it without anyone knowing. See related: How Cumulative Strain Injuries Actually Develop.

The conversation

The owner had objective evidence about the work, an overall callback rate more than three times the crew's (11.3 against 3.5 percent) concentrated in one kind of position, and a date. That made it reasonable to ask about the work directly, without guessing at a diagnosis. See related: Fit for Duty: When and How to Ask.

The owner opened with the attic split and the March 14 date, and asked what was going on with attic jobs. Luis said his right knee and low back had hurt since the slip, that he had expected it to clear up, and that he had not said anything because he did not want to be "the hurt guy" on a crew that jokes about it, or to lose attic work and the install spiffs that come with it. Both reasons are common, and both are the shop's to fix, not the tech's. See related: The Language That Keeps People Quiet About Pain.

The fix

The report, late but made. The owner opened the injury process that day, attached the chat message and the dispatch swap, and called the workers' compensation carrier. States set different deadlines for an employee to give notice of a work injury and treat late notice differently, often excusing it when the employer knew or was not harmed by the delay; whether a chat message the owner had read counts as notice is the carrier's question and, if it is disputed, an attorney's. The owner's job was to report what was known, with dates, and not to discourage a late report. See related: The Tech Who Would Not Report the Back Injury.

Clinician first, route second. Luis saw a clinician with the role's physical demands description. The note: no attic or crawlspace entry and no sustained kneeling for four weeks, then review.

Modified duty. For those four weeks, Luis took ground-level diagnostics, rooftop units reached from a ladder he could climb under the note, and quoting replacements, while the attic work went to two other techs. His spiff-eligible work dropped, so the owner paid his average spiff for the four weeks rather than let the injury cost him money for reporting it. See related: Returning to Work After an Injury: The Modified Duty Plan.

Back in, gradually. After the review, attic calls came back at two a week, then normal volume, with knee pads and a padded attic board on his van.

How they knew it worked

Eight weeks after the fix, Luis had run 96 calls with 3 callbacks, 3.1 percent, back under the crew's 3.5. His attic photo set was complete on 22 of 24 attic jobs, 92 percent, the same as his first quarter. Median attic time on site was back to 72 minutes, within a few minutes of the 74 before the slip.

The coaching conversation the owner had planned, about attention to detail, would have addressed none of this. The detail was fine. The knee was not.

What the record should have had

The case was solvable only because the shop happened to keep chat history and dispatch swaps. Three changes came out of it.

• An early discomfort report. A same-day log entry for aches and pains with no specific event, and a direct route to the injury process for anything with one. A "slipped off a joist" message would now prompt a lead to log it the same day. See related: The Early Discomfort Report SOP.
• Callbacks split by position, not just by tech. The per-tech rate hid the signal for a quarter. Split by job type, it is visible in the first month: 3 callbacks out of 20 attic jobs stands out where 3 out of 50 calls of every kind does not.
• A falling photo rate treated as a question. Photo compliance dropped before the callbacks rose. A drop like that is a reason to ride along or ask how the attic jobs are going, not a reason for a reminder email.

References

• See related: How Cumulative Strain Injuries Actually Develop
• See related: The Early Discomfort Report SOP
• See related: Fit for Duty: When and How to Ask
• See related: Returning to Work After an Injury: The Modified Duty Plan
• Your state workers' compensation agency's rules on employee notice of injury