team
Somebody has asked for a 12 hour rotating shift schedule, and the question underneath it is usually narrower than it looks. Either the current eight hour rota cannot produce the coverage the operation needs from the headcount available, or the rota already runs on twelves and the same three people keep absorbing every gap. Both problems get solved in the same order. Work out what the arithmetic permits, then pick a pattern, then decide how the rotation moves between days and nights. Picking a named pattern first is how a rota ends up looking fair on paper and collapsing the first week somebody takes leave.
Continuous coverage of one post, 24 hours a day and seven days a week, needs 168 staffed hours every week. On twelve hour shifts that is fourteen shifts a week, seven days and seven nights. Divide those fourteen shifts among the crews available and the average working week falls out before any pattern is named.
| Crews | Shifts per crew per week | Average hours per week | What that costs |
|---|---|---|---|
| 3 | 4.67 | 56 | Sixteen hours a week above a forty hour baseline, every week, forever |
| 4 | 3.5 | 42 | Two hours a week above forty, which is the usual landing point |
| 5 | 2.8 | 33.6 | Slack for leave and training, paid for with a fifth payroll |
Four crews is the floor for continuous twelve hour coverage, and that is an arithmetic result rather than a preference. Three crews can be made to work, but only by treating a 56 hour week as normal or by buying back the difference in overtime, and a rota that depends on overtime being available is a rota that fails when somebody declines it.
The forty two hour average is worth sitting with, because it is the same number for every four crew pattern described below. DuPont, Pitman and four on four off all produce it, since all three divide the same fourteen weekly shifts among the same four crews. Whether those two hours above forty attract a premium depends on the jurisdiction and the contract, and that is a question for an employment adviser rather than a scheduling article. What the patterns actually differ on is the shape of the time off, not its quantity.
One more number matters before the patterns. A twelve hour shift is not twelve hours away from home. Add a handover at each end and a commute and it is closer to thirteen and a half. That figure, not the twelve, is what makes four consecutive shifts hard.
All three of the following are rotating in the sense that each crew takes both days and nights. They differ in cycle length, in how long the longest run of shifts is, and in what happens to weekends.
| Pattern | Cycle | Longest run | Longest break | Weekends |
|---|---|---|---|---|
| DuPont | 28 days | 4 shifts | 7 consecutive days off | Varies through the cycle |
| Pitman (2-2-3) | 14 days | 3 shifts | 3 consecutive days off | Every other weekend off, in full |
| 4 on 4 off | 8 or 16 days | 4 shifts | 4 consecutive days off | Drifts across the calendar |
Each crew works four nights, takes three days off, works three days, takes one day off, works three nights, takes three days off, works four days, then takes seven days off. That is fourteen shifts and fourteen days off inside 28 days. The seven day break is the reason people ask for it by name, and it is genuine. The cost is the single day off in the middle, which sits between a block of day shifts and a block of nights and is the least restful day off on any common rota.
A fourteen day cycle of two on, two off, three on, two off, two on, three off. Seven shifts a fortnight. Nobody works more than three shifts in a row, and each crew gets a full three day weekend every second week, on the same weekends every time. Predictability is what this pattern sells, and for teams whose people have fixed commitments outside work it usually beats a longer break they can only take once a month. The variant written 2-3-2 is the same arithmetic with the blocks reordered.
Four shifts, four days off, repeating. In the rotating form it runs as four days, four off, four nights, four off, which is a sixteen day cycle. Because eight days does not divide into a seven day week, the pattern walks across the calendar and takes eight weeks to return to the same weekday alignment. For an operation that does not care which day of the week it is, this is the simplest of the three to administer and the easiest to explain. For anyone with a Saturday commitment it is the worst of the three, because no weekend is reliably theirs.
Two choices hide inside the word rotating, and they are often made by accident.
The first is speed. Pitman rotates fast, switching after two or three shifts. DuPont and four on four off rotate more slowly, in blocks of three and four. Fast rotation means nobody stays on nights long enough to adapt to them, which sounds bad and has a defence: without adaptation there is no readaptation to pay for on the days off. Slow rotation gets closer to genuine adjustment, then spends the first two days off undoing it.
The second is direction. A forward rotation moves days to evenings to nights, which delays the body clock. A backward rotation moves nights to evenings to days, which advances it. Delaying is the easier direction for most people, which is the practical argument for forward rotation where the pattern allows a choice. On a two shift twelve hour system the distinction is thinner than on a three shift eight hour system, but the block boundary still exists and can still be arranged either way.
Fixed nights sit outside this discussion and deserve a mention, because they are sometimes the honest answer. A crew that works nights permanently adapts better than any rotating crew. The costs are that the night crew becomes invisible to everyone who works days, that promotion and training tend to happen in daylight, and that the arrangement depends on volunteers who may not stay. Any rota that relies on one person never wanting to change shifts has a single point of failure with a name and a family.
Not on the arithmetic. The arithmetic is the easy part.
A four crew rota has no spare crew, so every absence is a twelve hour hole. There is no overlap to absorb it and no fifth crew to draw from. The hole gets filled by extending somebody who has already worked twelve hours or by calling somebody in on a rest day, and both of those decisions get made at short notice by whoever is on duty. That is the mechanism by which a fair rota turns into three people carrying it.
Handovers are the second failure point, and they cut both ways. Twelves mean two handovers a day rather than three, which is a real reduction in the number of times information can be dropped. Each handover now carries twelve hours of context instead of eight, which raises what gets lost each time one goes badly. Written handover notes matter more on twelves than on eights, and they are the first thing to disappear when a shift ends late.
Third, training and team meetings have nowhere to live. An eight hour rota with overlap has slack in it. A twelve hour rota does not. Anything the team needs to learn together either happens inside a shift, which means the operation is thinner while it happens, or on somebody's rest day, which means paying for it. Rotas that quietly assume the second without budgeting for it stop producing trained people after about a year.
Fourth, and most often missed, is the qualification chokepoint. A rota can balance hours perfectly and still be impossible, because only two people hold the certificate the night shift legally requires. Count qualifications per crew before counting hours per crew.
Rest periods, consultation duties and the treatment of hours above the weekly threshold are set by local law and by whatever agreement covers the workforce. These vary enough between jurisdictions that guessing is worse than not answering, so the point to confirm with a labour regulator or an employment adviser is the minimum rest between shifts and whether a change of this kind requires consultation before it is announced.
Then set the trial honestly. One cycle is not a trial. DuPont takes 28 days to complete once, and the seven day break that sells it arrives at the end, so a four week pilot tests only the hard half. Three cycles is a reasonable minimum for anything longer than a fortnight, which puts a DuPont trial at roughly three months.
Decide in advance what would count as failure. Useful figures are unplanned absence, total overtime hours, how many shifts were filled by somebody on a scheduled rest day, and how many handovers lost something that had to be redone. The last of those is the one nobody records and the one that decides whether twelves were the right call.
Most rotas fail socially before they fail mathematically. A spreadsheet emailed once a month produces a week where one person is working from version three while version five is the truth, and the argument that follows is about the tool rather than the schedule.
What removes that failure is not a better spreadsheet but a single place where the rota, the swap requests and the conversation about them all live. Four questions need answers anyone can find without asking: who is on tonight, who agreed to which swap, which requests are still pending, and who is owed the next unpopular block. A grid of days against time slots handles the first, a view of people against dates handles the fourth, and the feature list for this kind of tool is worth reading with those four questions in hand rather than as a list of boards. Teams already running a rota inside a general task tool will find the tool comparisons more useful than a feature list, because the difference that matters is whether the schedule and the discussion about it sit on the same screen.
One rule is worth setting on the first day, whatever tool holds the rota. A swap is not agreed until it is recorded where the next person to look will see it. Verbal swaps between two people who both remember correctly work until one of them is ill.
Count crews before anything else. At three crews the pattern is not the problem and no rearrangement of DuPont or Pitman will fix it, so the decision is a fourth crew or a documented reliance on overtime. At four crews, pick between a long break every 28 days and a predictable weekend every fourteen, ask the people who will work it which one they want, and then put the rota somewhere everyone reads the same version. Pinateca is free for up to five people and ten boards, which covers a single crew rota before any decision about paying for a tool.
The arithmetic says each crew would average 56 hours a week, since fourteen weekly shifts split three ways. Some operations run this way and cover the difference with overtime or a part time relief crew. It is sustainable only while overtime remains voluntary and available, and it leaves nothing spare for leave or training.
DuPont, which ends its 28 day cycle with seven consecutive days off. The trade-off is a single isolated rest day in the middle of the cycle, positioned between day shifts and night shifts, and less predictability about which weekends are free.
Forward rotation, meaning days to nights rather than nights to days, delays the body clock rather than advancing it, and delaying is easier for most people. On a two shift twelve hour system the effect is smaller than on a three shift system, but where the pattern allows either order, forward is the safer default.
At least three full cycles. A DuPont trial of four weeks covers the demanding part of the cycle and only reaches the seven day break at the very end, which produces a verdict on the worst half of the pattern. Three cycles puts a DuPont trial near three months and a Pitman trial at six weeks.
No slack for absence. Four crews cover the schedule exactly, so one person calling in sick creates a twelve hour hole that gets filled by extending someone or calling in someone on a rest day. Without an agreed rule for who fills those holes and in what order, the same few people end up filling all of them.