Leave Policy

What has to be settled before the document can be drafted?

Four questions, in order. Which categories the organisation will operate, since every additional category adds an approval path and a balance to track. Whether each category is granted at the start of a cycle or accrues through it, because that single choice determines what a joiner and a leaver are owed. Who approves what, and what happens when that person is unavailable. And how an absence is priced once the balance runs out, which is where the document meets [payroll software](/payroll-software) and where most of the ambiguity lives. Teams that begin drafting before answering these end up describing categories rather than deciding rules, which is why so many policies read clearly and settle nothing.

Why does an undefined situation become a dispute?

Because somebody has to decide it anyway, and whoever decides sets a precedent nobody agreed to. A request that spans a public holiday, an absence that turns out to be medical after it was approved as personal, a resignation with an outstanding balance, a manager who approves after the payroll cut-off has passed: each of these will happen, and if the document is silent the answer comes from whichever manager is asked first. The second employee in the same situation gets a different answer, hears about the first, and now there is a grievance about consistency rather than about leave.

How often should the document be revised?

On a fixed review cycle rather than in response to the last argument. Reactive edits accumulate into a document that contradicts itself, because a clause added to close one loophole rarely gets checked against clauses written years earlier. A scheduled review also forces someone to read the whole thing, which is the only dependable way to find the contradiction. Keep a version history with the date each version took effect, since a dispute about an absence taken last year has to be judged against the rules that applied then rather than the ones in force now.

Which decisions does the document actually contain?

Strip away the preamble and a leave policy is a set of rules on a small number of axes. Eligibility: who can use each category, and from when. Accrual or grant: whether the balance arrives in instalments or in full. Notice: how far ahead a request must be made, and which categories are exempt from that. Approval: who signs, and what the escalation is when they are away. Carry-forward: what survives the end of a cycle and what does not. Interaction with pay: at what point an unpaid day is created and who authorises the conversion. Every one of these is a decision the organisation can make differently, and every one of them will eventually be tested by a real case.

Note what is absent from that list: the entitlement quantities themselves. How much leave of each kind an employee must receive is imposed by legislation in most places, and what the legislation requires depends on where the person works and how the workplace is classified. Those requirements also move. Company policy sits on top of that floor and can be more generous but never less, so the quantities belong in a dated schedule checked against the current legal position by a qualified advisor for every location the organisation employs in, rather than being embedded in prose nobody revisits.

How does the policy connect to payroll and the attendance record?

Through two joins, and both are where policies leak. The first is the balance: an approved absence must reduce a balance somewhere, and if the approval lives in email while the balance lives in a spreadsheet, the two diverge within a quarter. The second is the cut-off: payroll closes on a date, and an absence approved after that date lands in the following run regardless of when it was taken. A document that never names the cut-off leaves employees discovering it through a payslip. Running approvals through the same [leave management software](/leave-management-software) that holds the balance removes the first gap; naming the cut-off in the text removes the second.

The third connection is quieter. When a balance is exhausted the absence does not stop being an absence, it changes into an unpaid day, and somebody has to decide whether that conversion is automatic or needs a separate approval. If it is automatic, the employee should be able to see the balance before requesting, which is the argument for exposing it through an [employee self-service portal](/employee-self-service-portal) rather than through a query to human resources. If it needs approval, the policy has to name who gives it and say what happens when nobody does before the run closes.

What are the recurring failure modes in the document?

The frequent one is a category that exists in the text and not in the system, or the reverse. Somebody adds a category in a revision, nobody configures it, and requests for it arrive as email that never touches a balance. The mirror image is a withdrawn category still configured long afterwards, quietly accruing for a handful of long-serving employees. Both are found the same way: list the categories the system holds, list the categories the document names, and compare them line by line. The exercise takes an afternoon and gets skipped for years.

A second failure is a document written for the head office population and applied to everyone. Field staff, shift workers and employees at other sites often work under different rules, sometimes different statutory rules, and a single text that ignores this either under-serves them or promises what the local operation cannot deliver. A third is drafting in the passive voice. A clause saying leave should be applied for in advance names nobody; a clause saying the employee applies to their reporting manager, who responds before the payroll cut-off, can be followed and can be shown to have been broken.

How is a change introduced without generating a grievance?

By separating the change from its effective date and from the balances people already hold. A revision that takes effect immediately penalises employees who planned around the old rules, which is the fastest way to make a reasonable amendment look arbitrary. An effective date far enough ahead that anyone can adjust, with existing balances either explicitly protected or explicitly bought out, removes the objection that carries the most weight in any consultation. It also gives managers time to plan around whatever behaviour the change is meant to produce, which is usually the part that determines whether it works.

The other half is explaining what problem the change solves. A carry-forward limit introduced with no reasoning reads as a takeaway; the same limit explained as a response to people accumulating balances they never use, alongside a commitment that managers must approve leave rather than defer it, reads as a trade. Publish the revised document in full rather than a summary, record which version each employee acknowledged, and keep the superseded version accessible. Anyone judging an old absence needs the old text, and reconstructing it from memory is not a defence.