Scheduling staff across multiple locations

Why is scheduling harder across several locations?

One schedule is a puzzle. Five schedules are five puzzles that share pieces, because the same people, the same skills and the same overtime thresholds move between them.

Each manager also optimises for their own unit. That is the right instinct locally and the wrong one for the group, since a person sitting idle at one unit is a shift another unit is paying overtime to fill.

The constraint that makes it genuinely hard is availability. It changes weekly, it lives in texts and conversations, and it is the input most likely to be out of date when the schedule gets built.

Where does labor cost actually drift?

Rarely in the obvious place. Most drift is overstaffing during quiet hours that nobody notices, because no shift is dramatically wrong and the total only looks bad at period close.

The second source is overtime that was predictable on Wednesday and only visible on Sunday. By then the hours exist and the cost is fixed.

The third is the gap between the schedule and the clock. Early clock-ins, late clock-outs and unrecorded breaks are small individually and structural once they repeat across every unit, every week.

Who should build the schedule?

The manager, working inside constraints you set. They know who works well together and who is about to quit, and no system knows either.

What you own is the frame: target labor percentage per daypart, overtime rules, who is qualified for which station, and the deadline for publishing.

Give them a draft rather than a blank grid. Reviewing and overriding a proposed schedule takes a fraction of the time, and the constraints are applied before the manager starts, not after.

How do you cover callouts without overtime?

Order matters more than speed. Broadcasting an open shift to everyone gets it filled by whoever is fastest, which is often the person already closest to overtime.

Ask in sequence instead: qualified, under their hours, and not scheduled the next morning. The shift still gets covered, at straight time.

Cross-training is the underlying fix. A unit where only two people can run a station has no cover, and the callout problem is really a coverage problem showing up late.