There is a precise moment, around the eighth or ninth hour of a shift, when things start going wrong quietly. The change counted wrong. The price rung up twice. The curt reply to a customer who did not deserve it. The box cutter used in a hurry. None of these episodes ends up in a report, but anyone who runs a store knows them all, and knows the time of day when they cluster.
The temptation is to read them as a matter of character or personal tiredness. In reality they are arithmetic: human attention is a resource that runs out, and a rota that ignores it is scheduling the errors right alongside the coverage.
Fatigue has a cost, it just never appears on an invoice
Long shifts appeal to planners because they look efficient on paper: fewer changeovers, fewer overlaps, fewer people to coordinate. The cost sits elsewhere: till discrepancies clustering in the final hours, botched returns coming back as complaints, the small cutter or ladder incident that would not have happened on a fresh afternoon.
If you have the data, run a simple test: check what time of day till errors and minor incidents concentrate. In most stores the answer embarrasses the rota, not the people.
Not all hours weigh the same
Eight hours on a sleepy Tuesday and eight hours on a promo Saturday are not the same fatigue, even if the payslip treats them identically. A smart plan counts not just the hours but what is inside them. The practical rules that follow are few: after a heavy shift, avoid next morning's opening; do not pile the longest shifts onto the same "resilient" people; on peak days, where possible, split the day across more people rather than stretching it over the same ones.
Beware the volunteer myth too: there is always someone who says "it's no problem for me" and takes all the overtime. It holds for months, then presents the bill all at once, usually at the worst possible time.
The final hours should be designed, not endured
If the long shift is unavoidable (and sometimes it is), its final stretch deserves deliberate design. That means: no high-precision tasks in the last two hours if there is a choice. The complex till count, the department stocktake, the new hire's training all perform better mid-shift than at the end. The last hours are perfect for low-error-risk work: tidying, facing up, cleaning.
Breaks deserve the same logic: one real break two-thirds of the way through beats two short ones dropped at random. And a skipped break on a nine-hour shift is not efficiency. It is a till error taking a run-up.
The alarm signal is silence
Last point: chronic long shifts rarely erupt into protest. They show up as a quiet decline: fewer smiles at customers, more short sick days, less willingness to swap. By the time discontent reaches words, it is usually late. A monthly glance at actual hours per person, overtime included, is the thermometer that catches the problem early.
There is also the limit rotas breach most often without noticing, the gap between one shift ending and the next starting: the sums are in rest between shifts.
Spreading the load visibly, with limits on hours and heavy sequences, is exactly the kind of constraint that Sked Solve handles on its own when generating schedules: the site shows how.