The HVAC Owner Who Couldn’t Find the Leak

Mark had built a serious business. Thirty million dollars in revenue, 85 technicians across three states, over 1,200 commercial clients. By every visible measure, things were working.

But the margins weren’t moving. Revenue was up year on year — and the profit stayed flat. He was working harder, managing more complexity, adding headcount to keep up, and somehow not getting further ahead. The kind of feeling that’s worse than a clear problem: no single thing you can point to. Just the accumulating sense that something is off.

He’d already tried to fix it. A route optimisation tool that looked great in the demo and sat unused after six weeks. A new scheduling software that the dispatchers worked around rather than with. A consultant who delivered a report and left. Every attempt followed the same arc — impressive at the start, invisible in the operation three months later.

And in the background, quietly, competitors were starting to market something he didn’t have language for yet. AI-backed maintenance guarantees. Predictive service contracts. He didn’t know exactly what they were doing, but he could feel the gap opening.

Previous tools had failed because they required his operation to change to fit the software. Previous consultants had diagnosed from a distance and handed over a document. What was different here was that the analysis came with an implementation: dispatchers trained in individual conversations, calibrated to what each person needed to hear to actually change how they worked. And an engagement structure with a success fee that only paid out if Mark’s accountant confirmed the numbers had moved.

The change management wasn’t a Phase 2. It was built into Day 1.

The problem wasn’t that Mark hadn’t tried. The problem was that he’d never had a number. Not ‘drive time is probably an issue’ but ‘your drive time is running at 34% of each technician’s day, comparable operations run at 22–24%, and the annual cost of that gap is $680,000.’ Without the number, he couldn’t decide what to fix first. Without the benchmark, he couldn’t tell whether his operation was poor or merely average.

The CoterieLabs assessment ran against three years of Mark’s actual dispatch data and compared it against benchmarks from comparable Australian HVAC businesses. It found three specific gaps — and quantified each one. The highest-ROI opportunity wasn’t the one he’d expected: it was predictive maintenance. 60% of his emergency callouts could be anticipated and prevented from equipment sensor data his systems were already collecting. He just wasn’t using it.

The pilot ran on 50 buildings for 60 days. A model trained on five years of service records and sensor data correctly predicted 12 of 14 compressor failures seven to fourteen days in advance. Twelve emergency callouts became scheduled preventive visits — at higher margins, without the emergency labour premium, without the Friday afternoon client call.

The first time the model named a specific compressor, in a specific building, eleven days before it failed — and a technician arrived to confirm it before any client had called — that was the shift. Not ‘the model is working’ as an abstraction. The practical reality that a Friday afternoon emergency had become a Tuesday morning scheduled visit. Mark’s operations manager put it plainly: ‘We’re finally running the business instead of being run by it.’

His Monday meetings now run on real operational data. He knows which equipment is likely to fail this month. His service guarantee is in writing — and he can prove it.