Attendance and Absenteeism Metrics: Every Formula You Need
Quick definition: Attendance metrics turn punch records into a small number of figures you can track over time and compare between teams. There are five worth calculating. Everything else is a slice of one of them.
The reason to compute these is not reporting for its own sake. It is that attendance problems are almost always structural, and a structural problem is invisible at the individual level. One person late three times looks like one person. A whole shift late three times a week is a transport, handover or rota problem, and only a rate calculated by team will show you that.
Each metric below has the formula, a worked example, and a note on what it hides, because every one of them can be read the wrong way.
1. Attendance rate
Formula: (Scheduled days attended divided by total scheduled days) times 100. Exclude approved leave and non-working days from the denominator.
Worked example. A team of 10, each scheduled 22 days in a month, so 220 scheduled days. Approved leave accounted for 14 of those days, leaving 206 expected. Two days were unexplained absence and three were certified sick. Attended days are 206 minus 5, which is 201. Attendance rate is 201 divided by 206 times 100, which is 97.6 per cent.
What it hides: everything about lateness. A team where everyone arrives 40 minutes late every day scores 100 per cent. Never report attendance rate without tardiness rate beside it.
2. Absenteeism rate
Formula: (Days absent divided by total scheduled days) times 100.
The critical decision is what counts as absent. Two versions, and you must label which you are using:
- Unplanned absence rate. Unexplained absence plus uncertified short absence. This is the operational number, because unplanned absence is what you cannot staff around.
- Total absence rate. Adds certified sick leave. Useful for workforce planning and for spotting a health or safety problem.
Worked example, same team. Unplanned absence rate is 2 divided by 206 times 100, which is 1.0 per cent. Total absence rate including the three certified sick days is 5 divided by 206 times 100, which is 2.4 per cent.
What it hides: frequency. Ten separate single days and one two-week absence can produce the same rate, and they are completely different management situations. That is what the Bradford Factor is for.
3. Bradford Factor
Formula: S squared times D, where S is the number of separate absence instances and D is the total days absent, over a rolling period, usually 12 months.
It weights frequent short absences far more heavily than one long one, on the basis that frequent unpredictable gaps are more disruptive to cover.
| Pattern | Instances (S) | Total days (D) | Score |
|---|---|---|---|
| One two-week absence | 1 | 10 | 10 |
| Two absences of 5 days | 2 | 10 | 40 |
| Five absences of 2 days | 5 | 10 | 250 |
| Ten single days | 10 | 10 | 1,000 |
What it hides, and this matters: the reason. The formula cannot distinguish a chronic condition producing frequent short absences from someone taking Mondays off, and it will score the first far higher than a single long absence. Used as an automatic trigger it discriminates against people with fluctuating health conditions and disabilities. If you use it at all, use it to prompt a conversation, never to trigger a consequence, and exclude protected absence from the calculation. Attendance point systems covers the exclusions in detail.
4. Tardiness rate
Formula: (Shifts started after the grace period divided by shifts attended) times 100.
Track total minutes late alongside it, because the two answer different questions. A 20 per cent tardiness rate at four minutes each is a grace-period conversation. A 5 per cent rate at 50 minutes each is a staffing problem.
Worked example. 201 shifts attended, 18 started beyond a ten minute grace period, totalling 240 minutes late. Tardiness rate is 18 divided by 201 times 100, which is 9 per cent, and average lateness is 240 divided by 18, which is 13.3 minutes.
What it hides: whether the schedule is realistic. Always segment by shift start time. Lateness concentrated on one start time is telling you about transport, handover or the rota, not about character.
5. Overtime ratio
Formula: (Overtime hours divided by regular hours) times 100.
Worked example. 1,608 regular hours and 96 overtime hours gives 96 divided by 1,608 times 100, which is 6 per cent.
What it hides: distribution, which is the whole point. A 6 per cent ratio spread evenly is capacity planning. The same 6 per cent concentrated in three people is a burnout and turnover risk, and often a sign that one skill exists in too few heads. Always report this per person as well as per team. Burnout in small teams covers the early signals.
How to read them together
No single number means anything on its own. The combinations are where the diagnosis is.
| Pattern | Likely cause | First thing to check |
|---|---|---|
| High attendance, high tardiness | Schedule or commute mismatch | Lateness grouped by shift start time |
| Low absence, high overtime | Understaffed, not over-committed | Overtime concentration per person |
| High absence in one team only | Local cause: manager, site, shift pattern | Absence by team and by supervisor |
| High Bradford, low total absence | Frequent short absences | Whether a health condition is involved, before anything else |
| Everything worsening after a date | A change you made | Rota changes, site moves, a new system |
| Absence concentrated in new starters | Onboarding or role mismatch | First 90 day attendance versus tenured staff |
Segments that matter more than the totals
A company-wide absenteeism rate is close to useless. The same number split these ways is actionable:
- By team and by supervisor. The most revealing cut, and the most uncomfortable.
- By site or branch. Isolates location-specific causes such as transport or facilities.
- By shift start time. The fastest way to find an unrealistic rota.
- By tenure band. First 90 days against everyone else.
- By day of week. Monday and Friday concentration is a well known pattern worth measuring before assuming.
- Month on month. A single month is noise. Three months is a trend.
What you need in place to calculate any of this
Every formula above depends on distinctions your records have to make. If they cannot, the numbers will be confidently wrong:
- Scheduled days per person, per team, in the right time zone. Without a schedule there is no denominator and no definition of late.
- Absence types kept separate. Approved leave, certified sick, uncertified, unexplained, protected. Present-or-absent records cannot produce these metrics.
- Punch-level times, not daily totals, or tardiness and overtime cannot be computed at all.
- Absence instances as well as days, for the Bradford Factor's S value.
- A stable team and site structure, or your segments shift underneath you month to month.
If you are tracking this in a spreadsheet, the Excel attendance tracker guide has the formulas. If you are choosing a system, the complete guide to attendance management systems covers what to look for, and HR metrics every small business should track puts these alongside the rest.
Key takeaways
- Five metrics cover it: attendance rate, absenteeism rate, Bradford Factor, tardiness rate, overtime ratio.
- Exclude approved leave and non-working days from denominators, and label whether your absence rate includes certified sickness.
- Attendance rate hides lateness entirely. Never publish it without tardiness rate.
- Absenteeism rate hides frequency. Bradford Factor exposes it but cannot see the reason, so never automate a consequence from it.
- Overtime ratio hides concentration. Report it per person, not only per team.
- Segment by team, supervisor, site, shift start time and tenure. Company-wide totals hide every actionable cause.
- One month is noise. Compare three.
Frequently Asked Questions
How do I calculate attendance percentage?
Divide scheduled days actually attended by total scheduled days, then multiply by 100. Remove approved leave and non-working days from both sides first, otherwise a team that took a lot of booked holiday looks like a team with an attendance problem.
What is a good absenteeism rate?
It varies so widely by industry, shift pattern and season that an external benchmark is rarely useful. The number that matters is your own trend and the gap between your teams. A team running at twice the rate of a comparable team is a finding regardless of what any industry average says.
What is the Bradford Factor and should I use it?
Instances squared times total days absent, over a rolling year, which weights frequent short absences heavily. Use it to decide who to have a supportive conversation with. Do not use it as an automatic disciplinary trigger, because it systematically penalises fluctuating health conditions and disabilities.
Should certified sick leave count in absenteeism?
Report both versions and label them. Unplanned absence excluding certified sickness is the operational figure for staffing. Total absence including it is the planning and wellbeing figure. Using the combined number for disciplinary purposes is where businesses get into difficulty.
How do I measure lateness fairly across different shifts?
Measure against each team's own scheduled start in its own time zone, with a published grace period, and treat a shift crossing midnight as one shift. A single company-wide start time will record night and remote staff as late every day and make the metric worthless.
What is a healthy overtime ratio?
Under roughly 10 per cent is comfortable for most businesses, but concentration matters more than the average. Check whether a handful of people are carrying most of it, which usually signals a skill held by too few staff rather than a general capacity shortfall.
How often should I review these numbers?
Monthly for the rates, with a quarterly look at trends and segments. Reviewing weekly invites reaction to noise. Reviewing annually means you find out about a rota problem eleven months after it started.
TracefyHR keeps punch-level detail with scheduled hours per team and time zone, separates approved leave, certified sick and unexplained absence, and reports by team, site and person, so the figures above come out of the records rather than out of a spreadsheet rebuild. See how attendance and reporting work or start a 30 day trial with no card.