Articles · Cleaning

Promised 20% off the cleaning payroll. Overtime is 8% of cleaners' pay

12 October 2026

Overtime piles up every month, payroll seems to be growing faster than revenue, and the cleaners, word has it, turn up late. The owner of a cleaning company concludes that a schedule for twenty-odd people cannot be built well by hand. The vendor agrees and shows a slide: a scheduling agent cuts payroll by 20%.

It is worth putting a second slide next to that one — built from the firm's own records. We will do it for Solfjell Rent. It is an invented Norwegian cleaning company from our sample reports: twenty-two cleaners, 301 customers, about 1,135 planned visits a month. The report on its records for the first quarter was written by our AI audit agent.

The vendor's slide was worked out for someone else's company

The headline number on the slide is 20% off payroll. The agent costs 6,900 NOK a month and 25,000 NOK to set up. The owner adds two hopes to the slide: the overtime will go, and the planner's job can become part-time.

It did not take long to find where the 20% came from: the figure is from a pitch slide about a much bigger company. It has nothing to do with Solfjell Rent's records. Our list of questions for a vendor starts in the same place: whose data was the figure worked out on.

The firm's slide, line one: overtime is 8% of what the cleaners are paid

The firm's overtime is real: about 173 hours and 57,342 NOK a month. But that is 8% of what the firm pays its cleaners. The rest is guaranteed hours, and they are paid whatever the schedule looks like — built by hand or flawless.

The promise: 20% off payroll. Overtime is 8% of what the cleaners are paid. The rest is paid whatever the schedule.

That sets the ceiling. Software that removed overtime down to the last hour would yield 57,342 NOK a month, and that is the most there is to gain, not a forecast. Just to cover the agent's monthly fee, overtime would have to fall by 12%. So far no fall at all is proven.

Line two: a long day sits next to a short one

Every cleaner has guaranteed hours. A short day is paid in full, and anything over the day's contract hours is overtime at the rate plus 40%. Under rules like these, what costs money is not the amount of work but its unevenness.

On average the planned day at the firm equals the contract day. But 319 cleaner-days in the quarter were planned at least an hour over the contract hours, and 337 at least an hour under. The firm pays a premium for the long day and the full rate for the short one: about 222 guaranteed hours a month fall on days planned shorter than the contract. And 60% of all overtime hours are matched by short-day hours of the same cleaner in the same week. The team lead has a plainer way of saying it: "one day I have ten hours and the girl next to me has five".

The week comes out that way because of how it is built. The planner copies last week: each customer keeps the cleaner they know, and the district comes second. With the priorities in that order, 38.8% of back-to-back visits by one cleaner are in different districts, and the plan leaves 60 minutes between them instead of 30. Overtime averages 0.24 hours on a day with no district change, 0.42 with one and 1.09 with two or more. That is a link in the figures, not a proven cause, and the report says as much.

From the inside none of this shows, because on average everything adds up. The average day equals the contract day. The payroll sheet shows overtime as a single sum, and the paid empty hours of a short day it does not show at all — they are ordinary wages. A long day and a short day only meet when one person's week is laid out in a row, and the firm has no such check before the schedule goes out.

Line three: the planner keeps 129 of her 160 hours

By her own count the schedule takes the planner 60 of her 160 hours a month. The other 100 go on same-day changes, complaints and cover for sick cleaners. With a scheduling program, she estimates, 40 of the 60 would go, and checking would add 8.67. About 129 hours remain. That does not make a part-time job, and freed time on a salary does not turn into money.

Neither slide shows the risk: one person knows the schedule

The schedule rests on one person. In the week of 9–13 February the sheet has one same-day change, against 7–23 in every other full week: customers' changes did not reach the cleaners. Keys, alarm codes, allergies and who cannot work with whom exist only in the planner's head. Closing this gap costs 8,000 NOK, once: the administrator is trained as backup planner and the routine is written down.

On lateness the records are silent. Times are written on paper sheets, and none of the 3,514 rows of the visit list has an arrival time. Anyone promising to cut lateness is promising to cut something nobody has measured.

Districts and one check now; the agent in two months

The first step is free. Each cleaner gets fixed districts for each weekday, and before the schedule goes out there is one check: no day far over or far under the contract hours. Offices are the place to start — the team lead says they do not mind who comes. Some customers will get a different cleaner, and that is the owner's decision, not the planner's.

At the same time, record the arrival time at every visit and ask the accountant for revenue and payroll by month. Without those, "payroll is growing faster than revenue" remains a feeling.

The report does not strike the agent out; it puts it off. It is worth coming back to after two months of working by the rule, if the rule is not enough. The agent's role is then more modest than on the slide: it drafts the week, the planner approves it, and the trial runs four weeks and is judged on one number — overtime hours a month. Complaints, same-day cover for a sick cleaner and the decision on which customers change cleaner stay with people.

Three numbers a cleaning company can pull today

1. Overtime as a share of everything paid to the cleaners. Last month's payroll run: the overtime total and the total of everything paid to the cleaners. That is the ceiling for any scheduling software. The warning sign is a promise bigger than that share — like 20% against 8%. Put the software's monthly fee next to it and work out what part of the overtime it would have to remove just to cover that fee. In the example, 12%.
2. Long and short days for one person in one week. A month's timesheet or schedule, and the contract hours. For each cleaner, mark the days at least an hour over the contract and the days at least an hour under. The warning sign is the two kinds side by side, for one person in one week. Then the overtime is not too much work but the way the work is laid out, and the first thing to do about it is a free check before the week goes out.
3. The share of mid-day moves between districts. In the schedule, take one cleaner's back-to-back visits and count what share of them are in different districts. In the example, 38.8%. Then compare overtime on days with no district change and on days with two or more: 0.24 and 1.09 hours in the example.

And one question without a number: who builds next week if the planner is off sick on Monday. If the answer is "nobody", that is where to start, before any software.

All articles