Two weeks in August. The fortnight around Christmas. The February school break. Whatever the peak is where you are, the requests arrive together, and in an operation that cannot close you are going to say no to somebody.

The uncomfortable part is that most places have no method for this. They have a first-come-first-served inbox, which is a method - it is just one nobody chose, and it systematically favours the organised, the confident, and the people whose partner sorted the holiday in January.

The constraint is not headcount

The instinct is to set a cap: no more than three people off per week per department.

That is better than nothing and it is usually wrong, because it counts bodies and the rota needs capabilities. Three off is fine unless two of them are your only keyholders, or your only registered nurse for that shift, or the only two forklift tickets on nights.

So the cap has to be derived from cover, not from a percentage:

  • Which roles must be present on each shift, regardless
  • How many people can currently perform each of those roles
  • What that leaves as genuinely grantable, week by week

Do this once and the number falls out. It also, reliably, tells you something you did not want to know about how thin your coverage is - which is a separate problem and a more important one.

Choose a method, and say what it is

Every method is unfair in some direction. The point is to pick the unfairness deliberately and publish it, because an unstated rule is experienced as favouritism whether or not it is.

First come, first served. Simple, transparent, feels fair in the abstract. In practice it rewards whoever books earliest, which correlates with confidence, seniority and not having to coordinate with a partner's employer. It also produces a scramble the moment the window opens.

Rotation. Whoever missed out last year gets priority this year. The fairest over a career, and the only one that survives the "you had Christmas off last year" conversation intact. It requires you to have recorded the outcome, which is the reason most places do not use it.

Seniority. Predictable and easy to administer. Demotivating for new staff, and in a sector with high turnover it means your newest people - the ones most likely to leave - get the worst of it.

Ballot. Everyone submits by a deadline, contested weeks are drawn. Feels arbitrary but is genuinely impartial, and it removes the scramble entirely. Works better than people expect when combined with rotation as a tie-breaker.

Split the peak. Nobody gets the whole fortnight; everyone gets one week of it. Unpopular to announce, and often the only honest answer when demand is high and coverage is thin.

Most workable schemes are a combination: a window with a deadline, a cap derived from cover, and rotation as the tie-breaker.

The mechanics that make it work

Open a window and decide all at once. This single change removes most of the resentment. Requests are collected until a deadline, then decided together against the cap. Nobody is penalised for being at work when the email went out.

Publish the rules before the window opens, including the cap, the tie-breaker and the date you will answer by.

Answer everyone on the same day. Trickling out approvals means the people declined find out last, and find out by inference.

Record the outcome, not just the leave. Who asked, who was declined, and for which week. Without it, rotation is impossible next year and you will be relitigating last summer from memory.

Decide carry-over at the same time. If people cannot take leave in the peak, they will take it later - and if your rules make that hard, you get the December scramble where half the team has a week left and the rota cannot absorb it.

Sector realities

Hotels and logistics have the cruellest version, because demand peaks precisely when staff most want to be off. This is where splitting the peak and being explicit early matters most, and where seasonal or bank staff genuinely help - see below.

Health and social care does not peak in the same way, but it cannot drop below required cover for a single shift, so the constraint is harder rather than seasonal. Rotation is particularly well suited here because the teams are stable enough for it to be felt as fair.

Retail peaks at Christmas, which collides with the leave year-end in a lot of contracts. Worth checking whether your leave year is doing you a favour or creating the problem.

The one to watch for

The person who never requests peak leave. It reads as flexibility and it is usually one of three things: someone who has given up asking, someone who cannot afford to go anywhere, or someone whose absence would expose that nobody else can do their job.

All three are worth a conversation, and the third one is a risk you should be actively managing rather than quietly relying on.