A utilisation rate is a fatigue policy
IT services companies set billable utilisation as if it were a financial ratio. For field engineers and on-call staff it is also a decision about how tired people will be when they drive, climb and change production systems at night.
Every IT services company I know sets a billable utilisation rate for its engineers. Finance usually sets it during the budget as a share of paid hours that can be charged to clients, and it travels down the organisation as a performance measure. For a consultant in an office the number mostly describes commercial health. For a field engineer driving between a hospital and a water treatment plant two hundred kilometres apart, or an infrastructure engineer on a week of night on-call, it also describes how tired that person will be when they next pick up a ladder, a steering wheel or a production change.
I did not fully understand that until March 2025. One of our field engineers, finishing a thirteen-hour day with two regional callouts and a late change window, rolled his vehicle on a Queensland highway while driving home. He survived with serious injuries and has since returned to work. Our investigation found no breach of procedure. He had done exactly what our roster, our dispatch rules and our utilisation rate asked of him, and that finding is the reason for this essay.
What eighty-eight per cent meant on the road
Torvelle planned its 780 field engineers at 88 per cent billable utilisation. The figure had been inherited from a period when most of our field work was in cities, and nobody had re-examined it as the business moved into regional Queensland, New South Wales and South Australia. An engineer on a 38-hour week at 88 per cent must charge about 33 hours to clients. In a city that is demanding but achievable. In the regions, where an engineer may drive three hours to a site, wait for a part, and drive two hours to the next, it is achievable only by working longer days.
The numbers showed it, if anyone had read them that way. Overtime averaged 31 hours per engineer a month. The median annual-leave balance had grown to seven weeks, because nobody could be spared. Attrition was 27 per cent a year, and the engineers who left were disproportionately the experienced ones who knew how long a regional day really took. Each point of utilisation was worth about A$1.4 million of field-services EBITDA, which is why the rate had survived every budget review I attended.
I had read those figures as a productivity problem. Overtime suggested we were short of people; leave balances suggested poor planning; attrition suggested a pay question. Each reading was partly true. What I had not done was put them together and ask what kind of day they described.
The hours a percentage hides
A utilisation rate is a ratio, and ratios hide their denominators. An engineer at 88 per cent of a 50-hour week has worked far longer than one at 88 per cent of a 38-hour week, yet both appear identical on a dashboard. Our reports showed the ratio by region and by engineer. They did not show total hours, consecutive long days or the number of nights an engineer had been woken by an on-call page. The measure we managed was the one least connected to fatigue.
On-call work makes this worse. An infrastructure engineer who takes three calls between midnight and five in the morning may record ninety chargeable minutes and lose a night's sleep. The next morning the roster shows them as available and lightly utilised. A change scheduled for that afternoon will be made by someone whose judgement is impaired in ways no utilisation report records. Service desks have a version of the same problem: a major incident bridge that runs through the night is staffed by people expected back at their desks the next day.
Work health and safety laws in Australia require a business to manage risks to its workers so far as is reasonably practicable, and fatigue from long hours and driving is a recognised risk. They also place a due-diligence duty on officers, which includes directors. A board that approves a budget built on a utilisation rate is therefore making a decision that sits inside that duty, whether or not the paper in front of it says so. In my experience the paper rarely does.
We changed four things in 2025. We lowered the planning rate for field engineers to 78 per cent and hired 60 more of them. We capped combined working and driving time at twelve hours, with a ten-hour break before the next shift. We limited on-call to four consecutive nights. And we asked clients to pay for regional callouts at a rate that reflected the travel, which two large clients accepted and one smaller one did not. Field-services EBITDA margin dropped from 13.4 per cent for 2024–25 to 10.9 per cent in the following half-year, then climbed back to 12.2 per cent in the six months to June 2026. Attrition fell to 14 per cent, overtime to 12 hours a month, and the lost-time injury frequency rate from 6.1 to 2.4.
What I now ask a board to read beside margin
The board pack now shows utilisation with its denominator, so directors can see hours as well as the ratio. Beside it sit four measures. The first is hours worked per engineer, shown as a median and as the ninetieth percentile, because the risk sits in the tail. The second is the number of engineers who worked more than four consecutive on-call nights in the quarter. The third is the count of people carrying more than eight weeks of untaken leave. The fourth is near-miss reports per hundred engineers, which I want to rise rather than fall, because a team that reports near misses is a team that still believes reports change things.
There is an honest objection to all this. Competitors plan field work at 85 to 90 per cent, and clients compare prices. A supplier that carries more people will lose some work on price, and we have. My answer is partly arithmetic: replacing an experienced engineer costs us about A$38,000 in recruitment, training and clearance, and lower attrition saved around A$3.6 million in 2025–26. That does not cover the margin we gave up. The rest of the answer is that a business which can only make its numbers by tiring its people is not as profitable as its accounts suggest; it is borrowing from their health and from its own future reliability.
I do not think utilisation should disappear from IT services management. It is a useful measure of whether the work sold matches the people employed. But it should be set by people who know what a regional day involves, reviewed against hours and not only against margin, and owned by a board that understands it is approving a pattern of work as well as a budget. A utilisation rate is set in a spreadsheet. It is lived on a highway at night.