Payroll close is an autopsy. By the time the period is locked, every hour in it was already worked, already promised to somebody, and already spent. Reviewing overtime at close tells you what happened. It has never once prevented anything.
The decisions that create premium hours happen days earlier, usually mid-week, usually at a moment when nobody is thinking about payroll at all — a dispatcher with an open post and a list of phone numbers, making a reasonable choice with incomplete information. If you want to control overtime in a security company, that is the moment you have to reach. Not the report at the end.
The three numbers that make a forecast
You do not need a model. You need three numbers per officer, visible inside the open period, before it closes.
Hours already accrued. What the officer has actually clocked so far in the current period. Not what was scheduled for them — what the clock says. Those diverge constantly: the relief who arrived late, the hold-over nobody entered, the shift that ran long because the client’s night manager wanted a walkthrough.
Hours still scheduled. What is on the board between now and the end of the period, for that same officer.
The sum. Accrued plus scheduled is the projected close, and it is the only number that matters on Wednesday. An officer sitting at 32 accrued with 12 still on the board is not at 32. They are at 44, and every one of those last hours is already committed. The dispatcher who calls them at 2 a.m. Saturday thinking “she’s only worked 32 this week” is not making a mistake about the officer. They are making a mistake about the arithmetic.
Most branches can produce the first number a week late and the second one instantly, which is exactly backwards. The forecast requires both to be current at the same time, which in practice means clock data has to be live rather than collected — guard time and attendance that updates as officers clock in and out at the post, so the totals are true all week instead of being reconstructed from paper on Monday.
Reading the week forward instead of backward
Once you have the projected close, the weekly read takes about ten minutes and answers a different question than the payroll report does.
Sort the roster by projected hours, descending. The top of that list is your premium exposure for the period, and it is knowable on Wednesday. Then look at the forward schedule for the same window and mark every open post — every shift with no name on it, every shift with a name you know is not going to show, every slot held by an officer whose swap request has not been approved.
Now put the two lists next to each other. What you are looking for is the collision: an open post on Saturday night and a call list where the three officers trained on that site are all projected past the threshold. That collision is not a surprise on Saturday. It is visible on Wednesday, and on Wednesday you still have options that cost nothing.
The options shrink by the hour:
- Wednesday: move a shift. Swap a projected-46 officer off a Friday and onto a Monday in the next period. Bring the relief officer forward. Ask the client whether the Saturday walkthrough can move. All of these are free.
- Friday: cover it with a straight-time officer who has to be asked nicely, or split it. Cheap but not free.
- Saturday, 2 a.m.: hold over the officer who is already on post, or pay premium to whoever answers. Expensive, and you did not choose it — the clock chose it for you.
The coverage gap you can see coming is a scheduling problem. The same gap discovered at shift change is a payroll problem. Nothing about the gap changed. Only the number of people who can solve it.
What a fill actually costs
The instinct at 2 a.m. is to fill the post with whoever answers. It is the right instinct — an open post is worse than an expensive post, always — but “whoever answers” should not be a random draw. The call order should be built before the call.
Consider the same open 12-hour shift filled three ways:
The officer at 46 projected hours. Every hour of that shift is premium, and it is premium on top of premium — you are buying twelve hours at the highest rate on your roster and pushing one person deep into a stretch where the next week’s fatigue call-off becomes more likely, not less. You have often bought two problems.
The officer at 28 projected hours. Twelve straight-time hours. Same coverage, same post familiarity if they are trained on the site, materially different cost. The only reason this officer does not get the call is that nobody knew they were at 28.
The officer at 28 who lives ninety minutes away. Straight time on paper. In practice you are asking someone to spend three hours of unpaid windshield time to work a shift they did not plan for, and the second time you ask, they stop answering. Post proximity belongs in the call order alongside hours.
None of this requires a policy change. It requires the dispatcher to see hours and site clearance on the same screen as the open post, at the moment of the call. If they have to check a spreadsheet from last week to find out who is where, they will not check it, and they are right not to — the post has to be covered.
Why the Sunday shift costs double
A shift that falls on a Sunday is the most expensive shift in the week for reasons that have nothing to do with weekend rates.
By Sunday, the period is nearly closed. Every officer who worked a normal week is sitting near their threshold, so the pool of people who can take the shift at straight time has shrunk to almost nobody. Simultaneously, the pool of people willing to answer the phone has also shrunk — Sunday is the day officers have plans. So you are bidding for a scarce resource at the exact moment your budget for it is gone.
Then there is the spread. A Sunday hold-over does not just cost Sunday. It pushes the officer into the next period tired, which raises the odds of a Monday or Tuesday call-off, which opens another post at the start of a fresh period where you now start from behind. One uncovered Sunday can seed two weeks of fills.
This is why the weekend read has to happen on Wednesday and not Friday. By Friday afternoon the cheap moves are gone.
Patterns that forecast well and patterns that do not
Some of your exposure is not a dispatch problem at all — it is baked into the rotation. Four crews on 12-hour shifts average 42 hours a week by arithmetic: 168 hours of post divided by four people. Two premium hours per officer per week, every week, before anybody calls off. That is not a forecast, it is a constant, and it should be in the bid.
What varies is how those hours land inside a calendar week, which is entirely a function of the cycle and of where your workweek boundary sits. A 14-day cycle lets you choose. An 8-day cycle does not — it drifts against the calendar and produces 36-hour and 48-hour weeks in an order that never repeats. The comparison across the common rotations is in security guard shift patterns compared, and it is worth doing before you spend another quarter forecasting around a pattern that structurally cannot close.
Two other structural sources are worth separating from the weekly noise. Swaps arranged between officers and never entered move hours between people invisibly and wreck every forecast built on the published board — a written shift swap policy exists mostly to keep the schedule true. And the definition of your workweek, the regular rate and any state rules that go beyond the federal baseline decide what a projected 44 actually costs you; FLSA overtime and security companies is the orientation, and your counsel is the answer.
Making the forecast something people act on
A forecast that lives in a report nobody opens is worth nothing. Three things make it operational:
Put it where the decision happens. Projected hours belong next to the open post in the dispatcher’s view, not in a weekly email to the branch manager.
Give it a threshold with a name attached. A defined hour count above which the fill needs a second signature, and a named person who gives it. This is not about slowing the fill down — the post gets covered either way. It is about the fill being a decision somebody made rather than a thing that happened.
Record the fill when it happens. A fill agreed over the phone and entered on Monday is invisible to everyone forecasting on Wednesday, and it is usually the same fill that never makes it onto an invoice. When the fill is entered in the schedule at the moment it is agreed, the forecast stays true and the pay side and the billing side start from the same set of hours.
The levers for reducing the overtime you find this way — schedule the post rather than the person, surface open posts early, fix the roster before you fix the pattern — are covered in controlling overtime in a security guard company. This article is only about seeing it coming. But seeing it coming is most of the job: almost every expensive hour in contract security was avoidable at some earlier point, at a lower price, by somebody who did not have the number in front of them.
Wage and hour rules, workweek definitions and classification vary by jurisdiction and change over time. Everything here is operational guidance, not legal or payroll advice.
If you want the projected-hours view sitting next to your open posts, explore CGuardPro or get in touch.