The approval flow that does not jam when the manager is on leave themselves
Echipa HR 365 · reviewed 2026-09-02 · 5 min read
An approval flow needs three things most people leave out: a deadline on each step, a stand-in declared in advance for every approver, and a rule for what happens when the deadline passes. Without them, the flow works perfectly for nine months of the year and fails in exactly July.
Why the flow fails precisely when it is needed
Leave requests cluster in the same periods when the approvers are away too. It is a structural coincidence, not bad luck: summer and holidays simultaneously produce the peak volume of requests and the minimum availability of those who approve them. A flow designed for normal conditions is designed for the period in which it is not tested.
The practical consequence: people start asking verbally, managers approve by message, and the records drift apart — which is exactly the problem from the record-keeping article, produced by an over-rigid flow.
How many approval steps, really
The simple rule: each step has to be able to refuse for different reasons. If the second step cannot say “no” for a reason the first did not have, it is a pointless check that adds only delay.
| Step | What it checks | Can refuse for |
|---|---|---|
| Direct manager | Team coverage in that period | Overlap with colleagues, a peak in activity |
| HR | The available balance and the correctness of the record | Insufficient days, the wrong leave type |
| Leadership | Only above a threshold (long periods, a critical season) | Business reasons, not team ones |
Delegation: declared in advance, not improvised
Every approver has a named stand-in, agreed at the start of the year, not on the day they leave. The rule that matters: the stand-in activates automatically, for the period of the absence, without anybody having to do anything. A delegation that requires an action from the person who is away never happens — because they are away.
How you set the chain, once a year
- HR — lists every approver and asks each for a stand-in, by name, in January
- Manager — confirms the stand-in genuinely knows which team they cover and which periods are critical
- HR — checks there are no loops: A covers B and B covers A, and both are away the same week
- HR — declares the company’s critical periods — not as a ban, but as an escalation threshold
The loop check in step three looks pedantic until it happens. Two colleagues who cover for each other are, statistically, away at roughly the same time: they are in the same team, with the same activity peak and the same school holidays.
Deadlines and escalation
Each step has a deadline. Without one, “pending” is a state that can last indefinitely and appears in no report as a problem. Reasonable deadlines: 2 working days for the manager, 1 day for HR, 3 days for the leadership step.
- When the deadline passes, the request escalates automatically to the stand-in. It is not cancelled and not approved by default — it moves to somebody else who can decide.
- The employee can see where the request is and since when. That transparency removes half the questions to HR.
- Requests submitted fewer than x days before the start go down a route marked urgent, with a shorter deadline. Otherwise exactly the ones that cannot wait get lost.
- A monthly report shows how many requests exceeded the deadline and at which step. It is the only way you learn that a particular manager is the bottleneck.
Nobody approves their own request
It looks obvious and happens constantly, especially in small companies where the same person is both manager and HR lead. The rule: nobody is the final step for their own request. If the structure does not allow otherwise, the request goes up to leadership, even if that step is not used in any other case.
How you communicate a refusal without damaging things
The flow handles approvals; refusals handle the relationship with the person. A request refused by a status change in the system, with not a word, produces more resentment than the period refused.
Three things that must accompany any refusal
- Manager — the concrete reason: who else is away in that period or what activity falls then — not “it is not possible right now”
- Manager — a proposed alternative, even an approximate one: what nearby period would work
- System or manager — the reason stays written on the request, so that in three months nothing has to be reconstructed from memory
The second part makes the practical difference. A refusal with an alternative turns, in most cases, into a new request the same week. A refusal with no alternative turns into a conversation with HR and, sometimes, into a day of absence in that period anyway.
One rule about critical periods too: they are declared at the start of the year, not two weeks ahead. A period announced in January as busy is information people use when making their plans; the same period announced in May is a restriction imposed on plans already made, and it feels like one.
How many steps are too many?
More than two on ordinary requests. Each step adds, on average, a day and a half of waiting, and the third rarely refuses anything the first two would have accepted.
What about last-minute requests?
You keep them possible, but flagged. A rule banning them outright gets circumvented verbally and takes the requests out of the records — exactly what you were trying to avoid.
An approval chain with a stand-in, deadlines and automatic escalation
Requests escalate to the stand-in on their own when the deadline passes, and the employee can see at any moment who has their request and since when.
You can create an account in a few minutes and use every module for 7 days, no card required.