Key takeaways
"How many technicians do we need?" is usually answered with a number somebody heard at a conference. It is answerable from your own data in about an afternoon, and the arithmetic is not hard. What makes it uncomfortable is that it exposes whether the maintenance plan was ever costed.
Everything the maintenance function is expected to do, in hours per year. Five buckets:
If your work order records are thin, the corrective bucket is where they hurt most. Estimating it is legitimate for a first pass, but the estimate should carry a health warning, and fixing the record keeping should go on the list.
Start from paid hours and subtract reality.
| Line | Hours per year |
| Contracted hours (40 per week) | 2,080 |
| Annual leave | minus 200 |
| Sickness and absence | minus 60 |
| Training and certification | minus 40 |
| Meetings, handover, administration | minus 80 |
| Attended hours | 1,700 |
| Hands on tool share, 50 percent in this example | times 0.50 |
| Productive hours per technician | 850 |
That last multiplier is the one that decides the answer, so do not borrow it. Wrench time is the share of attended hours actually spent working on equipment rather than travelling, waiting for parts, looking for a drawing or waiting for a machine to be released. Measure yours by sampling. Most teams find it lower than they expected, and the gap between the number they assumed and the number they measured is usually worth more than a new hire.
A worked example. A mid sized plant totals its demand:
| Planned maintenance | 3,400 h |
| Corrective work, last 12 months | 4,100 h |
| Backlog to clear this year | 900 h |
| Improvement and projects | 600 h |
| Total demand | 9,000 h |
At 850 productive hours per technician, 9,000 divided by 850 gives 10.6, so 11 technicians on workload alone.
Now the second constraint. Suppose the plant runs three shifts, seven days a week, and one technician must be on site at all times. One position covered continuously needs roughly 4.6 people once leave, sickness and training are accounted for, so five people are consumed by presence before any planned work is done.
The correct answer is the larger of the two numbers, evaluated together, not the sum. Eleven technicians can cover the shift pattern and still deliver the plan, provided the roster is built so that the people on nights are not all standing idle while the day shift carries the whole PM load. If the workload number had come out at three, the coverage requirement of five would have won, and the honest conclusion would have been that this plant is paying for availability rather than for work.
Notice what the example says about the plant: 4,100 corrective hours against 3,400 planned hours. Nearly half the workload is unplanned. Hiring eleven technicians to serve that ratio locks it in for years.
Before signing off headcount, ask what the demand would be if the plant were run properly. Reactive work is expensive per hour: it arrives at the worst moment, it takes longer than the same job scheduled, and it consumes the planned work that would have prevented the next failure. Cutting the corrective bucket by a quarter, through better PM quality and real root cause analysis on repeat offenders, removes about 1,000 hours of demand, which is more than one full technician.
So run the calculation twice. Once for the plant you have, once for the plant you are trying to build. The gap between the two answers is your improvement case, expressed in headcount, which is the unit finance understands.
Total headcount says nothing about whether the right people are on shift. A team of eleven where only one person can fault find on the packaging controls is understaffed in the only way that matters at 03:00 on a Sunday. Run a skills matrix alongside the staffing calculation and check coverage per critical task, not just bodies per shift.
It also says nothing about the split between internal staff and contractors. Specialist, seasonal or high peak work is often cheaper to buy than to employ. The rule of thumb that holds up: keep the work that touches your most critical assets in house, because that is where response time and accumulated knowledge pay for themselves.
Every number in this calculation comes out of maintenance records: PM task durations, completed corrective hours, backlog age, who did what and how long it took. Plants that keep those records in a CMMS can produce a defensible staffing case in an afternoon. Plants that keep them in a shared drive spend three weeks and still argue about the corrective figure. Adding OEE data on top answers the question finance will ask next, which is not how many technicians you want but how much production the current shortfall is costing. Watch the trend in MTTR as you add or remove capacity: if repairs are getting slower while the fleet is unchanged, you are seeing the shortage before the backlog shows it.
Book a demo to see planned and corrective labour hours by asset for a real plant.
Divide total annual demand hours, meaning planned maintenance plus corrective work plus backlog and projects, by the productive hours one technician delivers, which is attended hours multiplied by measured wrench time. Then compare that figure with the number required to cover your shift pattern and take the larger of the two.
Start from contracted hours, subtract annual leave, sickness, training and administration to get attended hours, then multiply by your measured hands-on-tool share. In the worked example above, 2,080 contracted hours become 1,700 attended hours and 850 productive hours at a 50 percent share.
No. Published ratios vary enormously because they hide differences in asset age, criticality, shift pattern and outsourcing. Use them only as a rough cross check after you have calculated your own number from your own hours.
Treat that ratio as a finding, not as an input. Staffing to a heavy reactive load makes it permanent. Model the demand again assuming a realistic reduction in repeat failures, and present the difference as the value of the improvement programme.
A backlog that is stable in size means capacity roughly matches demand. A backlog that grows month after month means demand exceeds capacity, and the growth rate in hours per month, annualised, tells you approximately how many technicians are missing.
Login
Blog