Whether somebody has days remaining is a lookup. Whether the business functions while they are away is the question, and most leave systems never ask it.
Leave management gets automated in the least useful direction. Most systems are built to track balances, which was never the difficult part. The difficult part is whether the business functions while somebody is away.
What approval should actually be asking
Whether somebody has days remaining is a lookup. Whether anything only they can do falls in that window, and who covers it, is the question that decides whether the absence causes a problem.
Answering it at request time, while the dates are still movable, is the whole point. Discovering it the week before is how leave gets cancelled, and cancelled leave is why people stop asking properly.
If leave has to be cancelled because of a clash nobody spotted, the problem is not the leave. It is that one person was the only route to something.
How to build it
1. Make the request assemble its own context
When somebody enters dates, generate what is due from them in that window, who else is already away, and any recurring obligation they own. The approver should be deciding, not investigating.
2. Require a named coverage owner per conflicting item
Not a general reassurance that it is fine. A name against each thing that falls in the window. If nothing can be named, you have found a continuity problem that existed before the leave request.
3. Show overlapping absences before submission
Half of leave friction is two people wanting the same fortnight and finding out after both have booked flights. Show the calendar at request time, not at approval.
4. Update everything on approval, automatically
Shared calendar, out of office, capacity in your utilization reporting, and any client-facing note. Each of these is small and each is forgotten regularly.
5. Trigger the handover artefact
Approved leave should generate the handoff document described elsewhere in this series, due before the last working day. This is the step that converts an absence from a risk into a routine event.
6. Flag people who are not taking leave
An automation that reports who has taken nothing in six months. Not to police it, but because an assistant or operator who cannot be absent is a continuity risk before it is a wellbeing question.
Tools and what they cost
| Option | What it costs | Honest trade-off |
|---|---|---|
| HR platforms (BambooHR, Rippling, Personio) | Per employee monthly. | Balances, approvals and calendars handled properly. Usually no view of what work falls in the window, which is the part that matters. |
| Google Workspace calendar plus Apps Script | Free with Workspace. | Full control including the deliverable clash check and the handover trigger. You build the balance tracking, which is the easy half. |
| Leave tracking add-ons | Low tens per month. | Cheap and adequate for balances at small headcount. No coverage logic. |
| A shared calendar and a spreadsheet | Free. | Works up to a point. Nothing surfaces clashes until somebody looks, and somebody usually looks too late. |
What it is actually worth
The direct saving is small and I would not build a case on it. Approvals are not a large time cost in most businesses.
The value is in what the automation exposes. Running the coverage check reveals how many obligations have exactly one possible owner. That count is a continuity measure, and it is the same number the documentation and handoff automations elsewhere in this series are trying to reduce.
Your own number: for the next five leave requests, count how many produced at least one item with no available cover. If that number is high, you do not have a leave problem, you have a single-point-of-failure problem that leave requests happen to reveal.
I am not going to quote statistics on unused leave or productivity effects. The figures in circulation come from HR software vendors and vary enormously by country and employment context.
How it breaks
Coverage is named but never briefed. A name on a form is not a handover. This is why approval should trigger the handoff document rather than just recording a name.
The clash check produces noise. If it flags every minor task, approvers stop reading it. Flag only work that person alone can do, which requires knowing who else can do what.
It becomes a surveillance tool. Reporting who has not taken leave should go to the person and to nobody else in the first instance.
Nothing updates on approval. The calendar entry, the out of office and the capacity report are each small and each forgotten, and their absence is what makes an approved absence feel unplanned.
How to tell whether it worked
Leave requests where a clash was found at request time rather than in the final week, which is the direct measure. Items with no available cover, which is your continuity exposure and should fall. And leave actually taken against leave entitled, which tells you whether people believe absence is genuinely possible.