Overtime

Why is authorisation the first question?

Because without it there is no way to distinguish work the organisation asked for from time an employee chose to spend. Both may be genuine, and only one was a decision. A prior-approval requirement, with a named approver and a route for emergencies that cannot wait, gives the organisation a chance to say no while the cost is still avoidable. It also protects the employee, whose claim is far stronger when a manager agreed in advance than when it rests on a recollection months later. Approval after the fact should exist, but as an exception that somebody counts.

What is actually being measured?

The gap between scheduled hours and worked hours, which sounds simple and is not. It depends on what counts as the start of work, how breaks are treated, whether time spent travelling between sites is included, and whether a person on call is working. Each of those has an answer in law that varies by jurisdiction and by the kind of work, and each also has a practice that may quietly differ from the answer. Where the two diverge, the practice is what an inspection will find, so the definitions belong in writing and in the system capturing the time.

Why is persistent overtime a rostering symptom?

Because a schedule regularly requiring extra hours to deliver the plan was never adequate for that plan. Occasional peaks are real and are exactly what the mechanism exists for. A function where the same people work beyond their hours every week is running below the headcount its demand requires and paying for the shortfall at a premium instead of fixing it. The cost is not only financial: the people absorbing it tend to be the most capable and the least likely to complain, and they are the ones who eventually leave, taking with them the capability the roster was quietly relying on.

What does the law govern, and what is left to the employer?

Legislation typically addresses whether extra work must be paid at a premium, which employees are covered, what ceilings apply, and what registers must be maintained. Employer discretion sits above that: whether to permit extra hours at all, who may authorise them, how far in advance, and what internal budget controls apply. Confusing the two is common and expensive, since a policy generous where the law is silent creates a cost nobody chose, while one silent where the law is prescriptive creates a liability nobody priced.

None of the legal parameters are safe to assume. What must be paid, to whom, up to what limit and with what record-keeping differs by country, by state, by sector and by how the establishment is classified, and these provisions are amended. Any organisation setting or revising a policy here should have the current position confirmed for each location by a qualified advisor, and should treat figures found elsewhere as unreliable, because they are almost always specific to a jurisdiction the reader is not operating in.

How is it controlled without simply forbidding it?

By making it visible before it is incurred rather than afterwards. A prior-approval requirement with a budget attached to each team turns extra hours into a decision somebody owns, and the ownership matters more than the threshold: an approver who can see their own cumulative total behaves differently from one signing individual requests in isolation with no running picture.

Blanket prohibition fails for a predictable reason. The work still needs doing, so it gets done anyway and stops being recorded, which converts a manageable cost into an undocumented liability and a register that does not match reality. That is a worse position than the one the prohibition was meant to fix, and it is the position an inspection finds. Controlling authorisation while continuing to capture every hour worked, through [attendance management software](/attendance-management-software) that records time regardless of whether it was approved, keeps the data honest and confines the argument to the payment.

How does it reach payroll without error?

Through one route, with the definitions applied in one place. The common failure is a parallel path: hours captured in an attendance system, a separate spreadsheet of approved extra work maintained by a supervisor, and payroll reconciling the two under deadline pressure. Each cycle produces a handful of discrepancies, each is resolved by judgement, and none of it can be reproduced three months later when somebody asks how a particular figure was arrived at.

Feeding captured and approved hours directly into [payroll software](/payroll-software) removes the reconciliation rather than merely speeding it up. It also forces the definitional questions to be answered once, in configuration, instead of repeatedly by whoever happens to prepare the input. The remaining discipline is the cut-off: hours approved after the run closes belong in the next one, and the policy should say so, because the alternative is a supervisor requesting an exception every month and payroll granting some proportion of them.

What should be reported, and to whom?

The total is the least useful figure available. What tells anyone anything is the concentration: which teams, which individuals, which weeks, and against which activities. Extra hours spread thinly across a department describe a peak; the same quantity concentrated in a handful of names describes a dependency, and an organisation should know which of those two it has before deciding anything about budgets or headcount.

Send it to the managers who authorise it, not only to finance. A finance report produces a directive to reduce a number, which produces the unrecorded work described earlier. A report to the authorising manager showing their own team's pattern against the roster, and against the absence that drove it, produces a staffing conversation instead. Where the same names recur every period, the useful next question is what those people know that nobody else does, because that is the real constraint and budget pressure will not move it.