Lateness: what to measure, what to ignore and when it becomes a problem
Echipa HR 365 · reviewed 2026-09-11 · 5 min read
Lateness matters to the extent that it affects somebody else. Five minutes at a desk job, where nobody is waiting, is noise; five minutes at a shop opening or a shift handover is a real problem. The same number, two completely different situations — which is why a single company-wide threshold does not work.
What you actually measure
| What you count | What it says | When it is useful |
|---|---|---|
| Cumulative minutes per month | The total volume | Almost never; it mixes different situations |
| Number of days with a late arrival | Frequency | Useful; frequency matters more than duration |
| Late arrivals past an operational threshold | How many affected somebody | The most useful indicator |
| The pattern across weekdays | Whether it is systematic or random | Diagnosis, not measurement |
The operational threshold
It is set per role, not per company, and it answers a single question: from how many minutes does somebody start waiting, or something stop working? In a shop with an opening time, it is zero. In a desk role with no fixed morning interactions, it might be fifteen.
Once set, the threshold is communicated. An unwritten threshold inevitably becomes one applied inconsistently — lenient with some, strict with others — and that produces more resentment than the lateness itself.
The pattern tells you the cause
| The pattern | Likely cause | What you do |
|---|---|---|
| The same weekday | A fixed constraint: school, another job, transport | It can be solved permanently, with a schedule adjustment |
| Only on the morning shift | Incompatible schedule or transport | Check whether the rotation is balanced |
| After a recent change | Something changed in their life | A conversation, not a sanction |
| Random, rare | Ordinary life | Nothing |
| Constant, increasing | Disengagement | A conversation about something other than the clock |
The first row is the most frequent and the easiest to solve: someone always late on Mondays almost certainly has a real constraint they never mentioned because nobody asked. A fifteen-minute schedule adjustment permanently solves something that otherwise produces monthly arguments.
The conversation, when it is warranted
How it goes
- Manager — starts from the fact and the effect: “you arrived after opening four times this month and Ana was left on her own”
- Manager — asks what is going on — often the answer is a solvable constraint
- Both — agree what changes: either on their side, or in the schedule, if that is possible
- Manager — checks back in a month; without the check, the same conversation repeats in three
The second step is what changes the outcome. A conversation that begins and ends with “you need to be on time” produces nothing, because the person already knows. One that finds out why can solve the cause — or, if it cannot be solved, at least reach an understanding both sides can hold to.
What if the lateness comes from unreliable public transport?
That is a real constraint and you can only solve it through the schedule, not through conversations. A fifteen-minute adjustment to the start time, where the role allows it, removes the problem permanently and costs nothing.
Should I let people make up lateness at the end of the day?
Where the work allows it, that is the simplest solution and it avoids the whole conflict. It does not work where the post has to be covered at a fixed hour — and there, making up the time is not the answer, adjusting the schedule is.
How do I handle lateness in remote work?
Usually you do not: if there is no hour at which somebody depends on them, the start time is not the right indicator. What matters is availability in the agreed windows and delivery.
Where to start
Set the operational threshold for each type of role and write it down. Without it, the measurement produces a number that everyone interprets differently.
Then look at the weekday pattern, not the total. If lateness clusters on a particular day, you have a scheduling solution, not a discipline problem.
What not to do
- Do not post lists of lateness by name. It produces humiliation, not punctuality, and it also lands on the people who are never late.
- Do not apply the same threshold everywhere. Five minutes does not mean the same at a till and at a desk.
- Do not ignore it silently and then raise it at the review. If it was not said at the time, it cannot be used six months later.
- Do not measure what you will not use. If you do not intend to do anything with the number, collecting it only produces distrust.
The third point is the most often broken and the most unfair. A manager who says nothing for four months and then brings the list to the review is communicating that they kept a silent record — and that is remembered far longer than the observation itself.
The other half: leaving early
It is measured far more rarely and is, in many teams, more relevant. Someone who arrives on time and leaves twenty minutes early, every day, produces the same effect as someone arriving late — except that nobody looks at the leaving time.
If you measure one and not the other, the indicator is incomplete and feels arbitrary. And if you measure both, most punctuality conversations turn, correctly, into conversations about schedule coverage.
Attendance recorded with the actual time, compared against the planned schedule
You see the pattern by day and by shift, not just a monthly total — so the conversation starts from the cause rather than from a cumulative figure.
Free account, every module for 7 days, no card required.