# Why Filter Lateness by Branch, Employee and Position?

- Canonical: https://qrgate.az/en/blog/lateness-by-branch-employee-position
- Markdown: https://qrgate.az/en/blog/lateness-by-branch-employee-position.md
- Language: en
- Date: 2026-06-12

The monthly report contains a number: "84 late arrivals this month". The figure is accurate and has never once led to a decision. The reason is that a total shows neither the problem nor its address. **Lateness analysis** begins exactly there: reading the same data by site, by role and by person. In practice those three cuts usually converge on the same conclusion - most late arrivals come from a handful of specific conditions, and many of them are within the company's own control.

## Why a single total is not enough

Is 84 a lot or a little? The question cannot be answered, because there is no basis for comparison.

A total conceals three things: where the lateness happened, in whom it concentrates, and when it occurred. Without those, management is left with one instrument - a general announcement. And a general announcement never works.

Worse, when the total moves, nobody knows why. One month it is 84, the next 61, and no one can say what changed.

## The first cut: by site

Looking by site usually produces the fastest result, because the differences are large.

Three typical causes appear in practice: transport access, the entry process itself (parking, a long corridor, one narrow door) and the site's own shift structure.

The first and third sit within the company's control. That is why the site cut so often surfaces the problems that are quickest to fix.

## The second cut: role and working pattern

Differences by role usually relate to the schedule. Roles on an early shift always show more lateness - that is a matter of the clock, not of discipline.

The second typical case is a short rest carried over from a previous shift. Someone moving from a night shift to an early one has an objectively higher chance of arriving late.

Once those two surface, the conversation moves from "who is late" to "which schedule causes lateness".

## The third cut: individual history

The individual cut is the most sensitive, which is why it should come last.

The reason is practical: once the site and role cuts have removed the systemic causes, very few lines remain on the individual list. Those few are genuinely worth a conversation, and they are easy to work with.

In the opposite order - looking at individuals first - the team develops a sense of unfairness, because someone carrying a systemic cause appears on the same list as everyone else.

## The fourth cut: weekday and hour

This is one of the least used and most informative cuts.

| Pattern | Likely cause |
| --- | --- |
| A Monday peak | Switching out of weekend mode, late rota publication |
| A peak at one site | Transport or the entry process |
| A peak on the early shift | The gap between shifts |
| A peak at the start of the month | Schedule changes announced late |

Every cause in that table is a hypothesis - confirming it requires a reason field. But those hypotheses are what set the direction of the investigation.

## Combining the cuts

The most valuable finding rarely comes from one cut; it comes from the intersection of two.

A "site plus weekday" view can show that one store only has a problem on Mondays. A "role plus hour" view can show that only one role on the early shift is affected.

The practical rule is: look at one dimension, find the pattern, then deepen with a second. Opening three dimensions at once usually blurs the picture.

## Reading the result correctly

This is the most important section. Lateness data shows one thing: that an employee did not create a record at the expected time. It says nothing about quality of work, productivity or commitment.

That distinction has practical weight. When lateness statistics feed directly into performance reviews, two things happen: data reliability falls and the team starts treating the system as an adversary.

The correct use is to read lateness as a test of the *process*, not of the person. How to read those indicators is covered on the [analysing attendance indicators](https://qrgate.az/en/analysing-attendance-indicators) page.

## What to measure

Four indicators are enough in practice:

1. **Count of late arrivals** - by site and role, not as a total.
2. **Average lateness** - read together with the count.
3. **Repetition** - how many people show more than three instances a month.
4. **Reason distribution** - which category dominates.

The urge to add a fifth indicator is common and usually makes the report heavier and less readable.

## You need a comparison baseline

Numbers mean something only in comparison. Three baselines work: last month, the same month last year, and other sites.

The last is the most useful, because conditions are similar. On one condition: the sites must be measured against the same baseline. Ignore differing schedules and the comparison loses its meaning.

## Who should read the report

The value of the analysis depends on its audience. The site cut belongs with site managers, the role cut with operations leadership, and the individual cut only with a direct line manager.

Sending every cut to everyone is the most common mistake: the report goes unread while sensitive data circulates widely.

## When to run the analysis

Timing affects the outcome directly. Two rhythms work in practice, and they serve different questions.

A **weekly view** is short and operational: what changed at which site this week. The purpose is not to find a trend but to react quickly - for instance if shift coverage has slipped.

A **monthly view** is structured: cuts are opened, compared and turned into a decision. Running the monthly analysis every week is pointless, because the numbers have not yet formed a pattern.

## A practical example

At a six-site service company, monthly lateness moved between 70 and 90, and for two years management reached the same conclusion: discipline is weak.

The site cut changed the picture at first glance: roughly half of all late arrivals concentrated in two locations. The role cut narrowed it further - at both sites the problem existed only on the early shift.

The cause turned out to be simple: both sites sat on the edge of the city and the first public transport service did not reach them before the shift began. After the shift was moved by 30 minutes, lateness fell by around a third within two months.

What stands out is that the data had existed for two years; the cut had not. We describe how lateness records are formed on the [tracking employee lateness automatically](https://qrgate.az/en/tracking-employee-lateness-automatically) page.

## How to assess where you stand

Three questions: can you see lateness by site in your report; is a cut by role possible; can you name the most frequent reason from last month?

If all three answers are no, what you have is lateness *statistics* rather than lateness *analysis*.

## Next step

Group last month's lateness list by site. That single table usually produces a first hypothesis the same day. We break the report into its components on the [attendance report](https://qrgate.az/en/attendance-report) page.

QRGate shows lateness records by site, role and reason with weekly and monthly summaries. [See what QRGate can do](https://qrgate.az/en/advantages) or [calculate the price](https://qrgate.az/en).

## FAQ

### Why is a single total not enough?

Because it hides three things: where the lateness happened, in whom it concentrates and when it occurred. Without those, only a general announcement is possible.

### Which cut should you start with?

The site. Differences are largest there, and some causes - transport, the entry process, shift structure - sit within the company's own control.

### When should you look at individuals?

Last. Once site and role cuts have removed the systemic causes, very few lines remain on the individual list and those are easy to work with.

### What should be done with the result of the analysis?

An analysis only pays off when it turns into a concrete step: adjusting a schedule, shifting a shift start, revisiting transport arrangements or hiring criteria. A monthly breakdown with no consequence quickly becomes a formality nobody reads.

## Related pages

- Tracking lateness - [Markdown](https://qrgate.az/en/tracking-employee-lateness-automatically.md) | [HTML](https://qrgate.az/en/tracking-employee-lateness-automatically)
- Attendance analysis - [Markdown](https://qrgate.az/en/analysing-attendance-indicators.md) | [HTML](https://qrgate.az/en/analysing-attendance-indicators)
- Attendance report - [Markdown](https://qrgate.az/en/attendance-report.md) | [HTML](https://qrgate.az/en/attendance-report)
- Leave and time-off management - [Markdown](https://qrgate.az/en/leave-and-time-off-management.md) | [HTML](https://qrgate.az/en/leave-and-time-off-management)
- QR check-in system - [Markdown](https://qrgate.az/en/qr-code-check-in-check-out-system.md) | [HTML](https://qrgate.az/en/qr-code-check-in-check-out-system)
- Who QR attendance suits - [Markdown](https://qrgate.az/en/who-is-qr-attendance-suitable-for.md) | [HTML](https://qrgate.az/en/who-is-qr-attendance-suitable-for)