Short answer: the Bradford Factor scores absence as S² × D — the number of separate absence spells squared, times total days absent. Squaring the spells makes ten one-day absences score 1,000 while a single ten-day illness scores just 10, deliberately flagging the frequent-short-absence pattern that disrupts teams most and correlates strongest with disengagement. Used with judgment and clear trigger bands, it’s a genuinely useful early-warning signal. Used as an automatic disciplinary tripwire, it’s a lawsuit and a morale problem. This post covers both halves.
The formula, and why spells are squared
Bradford Factor = S² × D
S = number of separate absence spells in the period (usually a rolling 52 weeks)
D = total days absent across those spells
A “spell” is one continuous absence, whatever its length: out Monday–Wednesday with flu is one spell, three days. The squaring is the whole point of the metric — it encodes the operational reality that frequency hurts more than duration. One planned two-week absence can be covered; ten unpredictable Mondays cannot. Cover is arranged eleven times, work is re-planned eleven times, and the team absorbs the uncertainty every week.
Three employees, same total days, wildly different scores
All three missed 10 working days in the last 12 months:
| Employee | Pattern | S | D | Score |
|---|---|---|---|---|
| Asha | One surgery, 10 consecutive days | 1 | 10 | 10 |
| Ben | Two flu bouts, 5 days each | 2 | 10 | 40 |
| Chirag | Ten separate single days, mostly Mondays | 10 | 10 | 1,000 |
Same absenteeism rate — a plain rate calculation treats them identically (which is why you track both; the Absenteeism Rate Calculator computes rate and Bradford Factor side by side). The Bradford lens says: Asha needs a get-well card and a phased return; Ben is unremarkable; Chirag’s pattern needs a conversation — this month, not at the annual review.
Standard trigger bands
Most organisations that use the Bradford Factor adopt bands like these (calibrate to your own workforce before adopting):
| Score | Typical response |
|---|---|
| 0–49 | No action — normal life happens |
| 50–199 | Informal, supportive check-in: “I noticed a pattern — is everything OK?” |
| 200–399 | Formal review meeting; attendance expectations documented |
| 400–499 | Written warning territory; occupational health referral considered |
| 500+ | Final warning / disciplinary process |
Two calibration notes. First, run the numbers on your current workforce before setting bands — in a customer-facing shift environment a 50 threshold might flag a third of your staff; in a small office it might flag no one for years. Second, use a rolling 52-week window, not the calendar year: a fixed January reset lets a December pattern vanish, and invites exactly the wrong kind of gaming.
Where the Bradford Factor goes wrong
The metric has well-documented failure modes, and every one of them comes from removing human judgment:
- Disability and chronic illness. Conditions like asthma, migraine, epilepsy, or chemotherapy cycles produce exactly the frequent-short-spell pattern the formula punishes. In the UK, disciplining on an unadjusted score can constitute disability discrimination under the Equality Act — the standard adjustment is discounting disability-related absences from the count or raising the individual’s trigger. Wherever you operate, the principle holds: medical patterns need accommodation, not arithmetic.
- Pregnancy-related absence must be excluded outright in most jurisdictions.
- Presenteeism. A hard trigger teaches people to come in sick — infecting the team and converting short absences into long ones. If scores drop but your absenteeism rate later spikes, that’s the mechanism.
- It measures pattern, not cause. A rising score under one specific manager, or clustered in one team, is telling you about the manager or the workload — the same diagnostic logic as turnover clustering. Score first, ask why second.
A sane small-business policy in five lines
- Track S and D per employee over a rolling 52 weeks; compute the score monthly.
- Use the 50 threshold as a conversation trigger, never an automatic sanction.
- Discount disability-, pregnancy-, and statutory-leave-related absences before scoring.
- Pair the score with the plain absenteeism rate — one catches pattern, the other catches volume.
- Publish the policy. A scoring system employees discover after being disciplined by it destroys trust permanently.
FAQ
Is a half-day a spell? Count it consistently — most policies treat any absence within one day as one spell with D = 0.5 or 1. Consistency matters more than the convention.
Does approved leave count? No. Planned holiday, parental leave, jury duty, and approved medical leave are excluded. The Bradford Factor is for unplanned absence only.
Two spells separated by a weekend? If the employee was absent Friday and again Monday for the same illness, most policies count one continuous spell. Different causes, two spells.
What’s a “good” average score? Below ∼30 as a workforce average is unremarkable. But the metric is designed for individual patterns, not averages — a team mean hides the one 900 it exists to find.
Related: the Absenteeism Rate Calculator computes both the absenteeism rate and Bradford Factor from the same inputs, with the productivity-cost impact. Policy thresholds here are common practice, not legal advice — check employment law in your jurisdiction before disciplining on any absence score.
