Overtime Percent as a Capacity Signal, Not a Payroll Line
Why this matters
Overtime arrives on a payroll report, so it gets treated as a cost line and handed to whoever signs the checks. That routing is usually wrong. Most of the premium hours a small shop pays are not bought by working more in total, they are bought by working unevenly, and unevenness is a scheduling and assignment problem that payroll cannot touch. A shop that reads this number as a cost hires a body it does not need, or squeezes a crew that is already covering for a calendar nobody is smoothing.
Capacity is lumpy, across the calendar and across people
Hours paid at a premium over hours worked - and "hours worked" here has to be the FLSA figure rather than the on-site figure, because for a non-exempt field employee travel between jobsites during the workday is compensable under 29 CFR 785.38 while the ordinary home-to-first-site commute generally is not under 29 USC 254(a), so a shop that leaves job-to-job travel out of the denominator is reading a flattering ratio and carrying a wage exposure behind it. The assumption is that capacity is fungible: that any hour of demand can be met by any technician on any day, so a shop with enough total capacity will not pay a premium.
Real capacity is not fungible in either direction. It is lumpy across the calendar, because demand arrives in bursts, and it is lumpy across people, because one technician holds a licence, a customer relationship or a truck stock that the others do not. Both kinds of lumpiness produce premium hours at total volumes well inside the crew's aggregate capacity, and they look identical on the payroll report.
Before the cuts, one caution that has to sit in the same breath as any number here: premium thresholds are jurisdictional. The federal Fair Labor Standards Act sets one and one-half times the regular rate for a non-exempt employee's hours over 40 in a workweek, at 29 USC 207(a)(1), and several states add a daily threshold and sometimes a double-time tier on top of it. Which threshold applies to you changes which of the cuts below is diagnostic, so confirm yours before you act on any of this.
The first cut, because it uses the data you already have
Start with the floor. Given the total hours worked, how many premium hours were unavoidable even under a perfectly even distribution across every person and every week?
A 5-technician shop, a 13-week quarter, a nominal 40-hour week, working under a weekly threshold. Hours worked in the quarter: 2,704. Premium hours: 338. Overtime share: 338 / 2,704 = 12.5 percent of hours worked.
Spread perfectly evenly, 2,704 hours across 5 people and 13 weeks is 2,704 / 65 = 41.6 hours per person-week. Each person-week would run 1.6 hours over the threshold, and 65 person-weeks at 1.6 hours is 104 premium hours. That is the floor, and it is the part of the bill that total volume genuinely bought.
The actual figure is 338. The excess over the floor is 234 hours, which is 69.2 percent of the quarter's premium hours. Just under a third of the bill is volume, and just over two thirds of it was paid for unevenness. Both of those shares are of the same base, the 338 premium hours in that quarter.
That single subtraction reframes the whole conversation, and it needs nothing but total hours, headcount and weeks.
Four causes, and the cut that separates each
The four causes below produce the same headline number and separate cleanly once you slice it. They are ordered by the records you are most likely to already have, because each pass removes exactly one candidate either way, so there is no reason to start with the cut that needs data you have to go build.
| Cause | The cut | What confirms it |
|---|---|---|
| Volume: demand genuinely exceeds crew capacity | Total hours against the even-distribution floor | The floor accounts for most of the premium hours |
| One person absorbing everyone's overflow | Premium hours by person | One name carries far more than an even share |
| Uneven demand across the window | Premium hours by week | A minority of weeks carries a majority of premium hours |
| Jobs consistently underestimated | Actual hours against estimated, by job type | Actual runs systematically over estimate on repeated job types |
The worked quarter
Floor. Done above: 104 of 338 premium hours, 30.8 percent, explained by volume alone.
By person. The five technicians carried 82, 71, 68, 63 and 54 premium hours, summing to 338. An even share would be 67.6 hours each, so the heaviest carrier is at 1.21 times an even share. Nobody is absorbing the shop's overflow, and the assignment habit is not the problem. That candidate is out, and ruling it out matters because it is the one a shop is most tempted to fix with a conversation.
By week. Three of the 13 weeks carried 194 of the 338 premium hours, which is 57.4 percent of the quarter's premium hours in 23.1 percent of its weeks. That is the live cause. The shop's aggregate capacity was adequate and its calendar was not.
By estimate. Across jobs that carried an estimate, actual hours ran at 1.08 times estimated on the median job. Read that carefully: an 8 percent median overrun is not an 8-point contribution to a 12.5 percent premium share, because only the overrun hours that push a person-week past the threshold are paid at a premium, and in a week that started with slack none of them are. Underestimating is a real contributor here and a minor one, and the fix for it is a price-book correction rather than a scheduling change.
The Friday trap
The obvious cut is premium hours by day of the week, and under a weekly threshold it is worse than useless.
When the threshold is weekly, the hours after the 40th get tagged to whatever day they happened to fall on, and that is almost always late in the week by construction. A shop with completely flat demand across five days will still see its premium hours pile onto Thursday and Friday, and it will conclude it has a Friday problem. It does not. It has a weekly total, recorded in day order.
The day-of-week cut becomes diagnostic only where a daily threshold applies, in which case a long Tuesday genuinely costs a premium regardless of what the week totals. Confirm which regime you are in before you read anything into that chart, and if you are on a weekly threshold, cut by week instead.
Sizing the fix without hiring anyone
The floor of 104 premium hours is not a target. Real demand does not level perfectly, emergency calls do not wait for a light week, and a shop that chased the floor would be turning down the work that pays best.
Size the recoverable half instead. The excess over the floor is 234 hours. Halve it as an opening assumption and re-measure after one quarter rather than taking the halving as given, since no published recovery rate exists for this and the achievable share depends entirely on how deferrable your book is - on this shop's figures that leaves 104 + 117 = 221 premium hours against the same 2,704 hours worked, or 8.2 percent. That is a fall from 12.5 percent to 8.2 percent of hours worked, 4.3 points, with the same crew doing the same total work in the same quarter.
What moves in practice: maintenance visits with no customer-facing date, deferred repairs the customer has already agreed to, internal work, and installations where the schedule is yours to choose. What does not move: anything a customer has been given a date for, anything under a contracted response time, and emergency work. If the deferrable share of your book is thin, the recoverable excess is thin too, and this is where the answer is genuinely to add capacity.
Hiring against premium hours, and the comparison shops get wrong
The premium is the differential above straight time, not the whole hour: at one and one-half times the regular rate, 338 premium hours cost the shop the equivalent of about 169 extra straight-time hours over the quarter - but only where the regular rate is the base wage alone, because under 29 CFR 778.208 a non-exempt employee's regular rate must also include non-discretionary pay such as per-job spiffs, sales commissions, shift differentials and attendance or safety bonuses, so a shop paying any of those and computing overtime on base wage is both understating this figure and underpaying the hours behind it.
A sixth technician on a nominal 40-hour week for 13 weeks adds 520 straight-time hours of capacity, and 520 hours of straight-time cost. So hiring to erase this shop's overtime buys about 3 hours of paid capacity for every 1 hour of cost it removes, both sides of that ratio being straight-time hours in the same quarter.
That ratio is not an argument against hiring. It is the reason hiring is the right answer when you have demand for the 520 hours, and the wrong answer when you do not. In this quarter, 69.2 percent of the premium hours came from distribution rather than volume, so the sixth technician would spend a meaningful share of those 520 hours idle across the ten light weeks while the three heavy ones stayed heavy.
Premium hours are also the cheapest capacity a shop can buy in short bursts, carrying no recruiting, no vehicle, no tooling, no training period and no commitment past the week - and at the federal level the FLSA sets no cap on the hours a non-exempt employee aged 16 or over may be required to work, though a handful of states restrict mandatory overtime for specific occupations and a collective agreement can bar it outright, so confirm yours before you build a season around it. A shop that drives this number toward zero as a matter of policy ends up carrying staff through its own shoulder season, which costs more and shows up nowhere on a payroll report.
One more cost fact points the same way: overtime wages are subject to workers compensation premium at the regular straight-time rate in most states rather than at the 1.5x premium, but only where your payroll register breaks regular, overtime and bonus pay out separately, because an auditor who cannot see the split rates the full overtime gross. The audit-prep card in the references is where that breakout gets set up.
Who owns each fix, and when a high share is correct
| What the cuts found | Who owns the fix | What they actually change |
|---|---|---|
| Volume against the floor | Owner | Add capacity, decline work, or raise price until volume fits the crew |
| Concentration on one person | Whoever assigns work | Spread the overflow, and check whether that person holds a qualification nobody else does |
| Concentration in a minority of weeks | Whoever schedules | Move deferrable work out of peak weeks and hold peak capacity for work that cannot move |
| Systematic overrun against estimate | Whoever owns job templates and pricing | Correct the estimate on the job types that overrun, and do it in steps rather than closing the whole gap at once |
A high overtime share is the right answer in two situations, and both are worth naming because a shop in either of them should stop apologising for the number. The first is a business whose product is response: if customers pay a premium for same-day and after-hours work, the shop is reselling the premium it pays and the share should be high. The second is genuine seasonality, where the alternative to premium hours in the peak is idle payroll through the trough. The test that separates a real seasonal pattern from a chronic one is whether the annual figure looks like the peak quarter's. If it does, the pattern is not seasonal, it is your normal week, and the four cuts above apply to the whole year.
References
- 29 USC 207(a)(1), Fair Labor Standards Act - the federal weekly overtime threshold and the one-and-one-half multiplier used above; state daily thresholds and double-time tiers sit on top of it
- See related: Workers Compensation Annual Audit Preparation - where the regular/overtime payroll breakout gets built
- See related: Technician Utilization and What the Denominator Assumes
- See related: Jobs per Day per Technician: Why More Is Not Better
- See related: Average Travel Time and the Route Density It Implies