How to Handle Shift Swaps Fairly (A Simple Approval Rule That Works)

In our experience swaps fail in four ways, and three of them are invisible on the day. The first is that nobody wrote it down, so the rota says Beth is on Saturday, Beth thinks Dev is on Saturday, and Dev has forgotten. The second is that the swap was fine in isolation but broke the gap between shifts either side: a person picks up a Saturday early having worked the Friday late, and arrives with eight hours between shifts instead of eleven. The third is that the shift changed hands but the skill did not; the one person on Saturday with medication training has just swapped out. The fourth is slower and does more damage: the same two or three people, who are quick, sociable and good at asking, end up with every weekend off they wanted, and everyone else notices.

The fix for all four is the same. Decide in advance what counts as an acceptable swap, make every swap pass through one place, and look at the totals once a quarter.

A swap policy does not need to be long. It needs to answer six questions clearly enough that a new starter can follow it without asking. Who can swap with whom. How a swap is requested and where it is recorded. Which swaps are approved automatically and which need a manager. How far in advance a request must be made. What happens if the person who took the shift does not turn up. And how often the log is reviewed for fairness. Here is the version we have seen work, condensed into a table you can adapt.

The rule that does most of the work is this. A swap is like-for-like when the two shifts are the same type (an early for an early, a night for a night), the same length, and the two people are in the same role band. A like-for-like swap that passes the rest-gap check and the skill-mix check is approved the moment both people confirm it in the log; nobody has to wait for a manager. Everything else is manager-checked: a night for an early, an 8-hour for a 12-hour, a senior for a junior, or any one-way pick-up where a shift changes hands without one coming back.

This splits swaps into the majority, which are dull and safe and should not take anyone's time, and the minority, which change hours, pay, skill mix or sleep and deserve a look. On the rotas we have watched, roughly four in five swaps are like-for-like, and the fifth is the one that would have caused trouble. Managers who insist on approving every swap end up rubber-stamping the dull ones and missing the odd one out.

The rest-gap check looks at the day before and the day after the swapped shift for both people. Under the Working Time Regulations an adult needs 11 hours between shifts, and a like-for-like swap can break that just as easily as any other: taking a Saturday early is fine unless you already have the Friday late. Check the gap from the end of the previous shift to the start of the new one, and from the end of the new one to the start of the next. If either is under 11 hours the swap fails, whatever type it is. On 12-hour patterns such as 4-on-4-off it is the day-to-night turn that fails most often.

The skill-mix check asks whether the shift still has what it needs after the swap. Write down, per shift, the minimum: one first-aider, one person who can do the till close, one medication-trained member of staff, one driver. If the person leaving the shift was the only one with a required skill, the swap fails unless the person arriving has it too. This is the check that a text-message swap never does, because neither of the two people swapping can see the whole shift.

The log is the whole system. It records who asked, what was offered, who took it, which type it was, whether the two checks passed, and who approved it. It does not need to be software; a shared sheet with these columns works, as long as it is the only place swaps live. Here are three entries from one week, which we work through at the end of the guide.

Once the log exists, fairness is a matter of counting. Two numbers are worth tracking per person per quarter: how many swaps they initiated, and how many weekend or bank-holiday shifts they gave away net of those they took on. A person with six outbound requests and a net gain of four weekends is not breaking any rule, but the rest of the team will have noticed long before you do. We suggest a soft cap, for example four initiated swaps a quarter before the next one becomes manager-checked regardless of type, and a hard rule that a weekend shift can only be given away to someone who has fewer weekend shifts that quarter than the person giving it up. Publish both numbers. Nothing dampens the quiet resentment about swaps like a list everyone can see.

The same logic applies to the shifts nobody wants for other reasons: Christmas Day, the night before a stocktake, the Sunday of a bank-holiday weekend. Our guide on covering annual leave in small teams uses the same counting approach for leave, and the two logs are worth reading together.

A deadline exists so that a failed check can be fixed before the shift, not after it. Forty-eight hours before the earlier of the two shifts is enough for most day shifts; for nights we suggest 72, because a night that fails a rest-gap check on a Friday evening is very hard to re-cover. Requests inside the deadline are not banned, but they always go to a manager, and the person asking should come with a proposed cover rather than a problem. The deadline should be measured from the request appearing in the log, not from the conversation in the car park.

When a swapped shift is missed, the first thing anyone says is "it was not my shift". The policy has to have answered that already. Our position, and the one that has held up on every rota we have seen adopt it, is that the moment a swap is approved, the shift belongs entirely to the person who took it. Their absence is recorded against them, their sickness is their sickness, and the person who originally held the shift has no obligation to come in. Anything softer than that and the original holder stops treating the swap as real, and the person who took it stops treating it as theirs. Write it in the policy, and say it out loud when the first one happens.

Five care assistants cover one early (E, 06:00 to 14:00), one late (L, 14:00 to 22:00) and one night (N, 22:00 to 06:00) every day. Here is the week as published. Every column has one E, one L and one N.

Chloe offers her Wednesday night and Ella offers her Sunday night in return (S-041). Both nights, both the same role, and the gaps hold: Chloe finishes Tuesday's night at 06:00 Wednesday and next works Sunday at 22:00; Ella was off on Tuesday and now starts four nights on Wednesday. Approved automatically. Beth then asks to give up Saturday's early for a wedding and offers Dev's Sunday early in exchange (S-042). On paper it is like-for-like, but Dev works the Friday late until 22:00, so a Saturday 06:00 start gives him 8 hours. Declined. Beth's second attempt is a one-way pick-up: Chloe, who is on only three shifts this week, takes the Saturday early with nothing coming back (S-043). That is manager-checked because it changes paid hours and moves Chloe from nights to a day. Chloe's last night, now Tuesday's, ends at 06:00 Wednesday, 72 hours before the early, and the early ends at 14:00 Saturday, 32 hours before her Sunday night. Saturday's early still has a medication-trained person on it. Approved.

Read the columns again: still one E, one L and one N on every day, 21 shifts in total, and the hours have moved from 5-5-3-4-4 to 5-4-4-4-4, which is fairer than the original. Two swaps approved, one declined, all three in the log, and nobody had to remember a conversation.

If you want a pre-publish test to run on the whole rota after a batch of swaps, the shift coverage checklist is the one we use.