The Bradford Factor is a small piece of arithmetic that turns someone's absence record into a single number:

B = S² × D

where S is the number of separate absences and D is the total days lost.

The squaring is the whole idea. Ten separate one-day absences score 10² × 10 = 1,000. One ten-day absence scores 1² × 10 = 10. Same days lost, a hundred-fold difference in score.

The part that is right

The premise is operationally sound, and anyone who builds a rota knows it instantly.

One person off for two weeks is a planning problem. You know on Monday, you arrange cover, you get on with it. Ten single days scattered across a quarter, each announced at 6am, is ten emergencies. Each one costs a manager's morning, a phone round, a favour, and often overtime at short notice - and it is the person who came in who pays for it, in a shift they did not choose.

So a metric that weights frequency over duration is measuring something real. Total days lost, the number most organisations actually report, treats those two cases as identical and they are nothing alike.

The part that goes wrong

The formula has no idea what it is counting. Ten single days could be:

  • Ten separate minor illnesses
  • A chronic condition that flares
  • A disability
  • Migraines
  • Fertility treatment
  • A child who keeps getting sent home from school
  • Domestic abuse
  • Ten hangovers

The score is identical in every case. The correct management response is different in every case, and in several of them a sanction triggered by a threshold would be both wrong and, in most European jurisdictions, legally risky - absence connected to disability or pregnancy is treated differently, and applying a neutral-looking score without adjusting for it is a well-known way to end up with a discrimination claim. Take advice in your own jurisdiction before any policy of yours has an automatic consequence attached to a number.

There is also a quieter failure. A score that people understand is a score people manage. Once staff know that frequency is what counts, the rational response to feeling ill on a Tuesday is to take the rest of the week rather than one day - which raises total days lost, harms the person, and improves nobody's score-based view of them. Or they come in ill, which in care, food handling and any customer-facing role is considerably worse than the absence you prevented.

How to use it without doing damage

As a trigger for a conversation, never as a trigger for a consequence. The number's only legitimate job is to tell you where to look. What you find when you look is the actual information.

Exclude what must be excluded, before it scores. Absence related to disability, pregnancy, industrial injury, bereavement and any statutory leave should not be in the calculation at all. If your system cannot categorise absence, it cannot do this, and it should not be producing scores.

Publish the thresholds and what happens at them - and make what happens at the first one supportive rather than punitive. A trigger that means "your manager will ask how you are" behaves very differently from one that means "stage one warning", including in what it does to reporting honesty.

Look at the team score, not just individuals. A department where several people are triggering is telling you something about the department - rota pattern, workload, a manager, a physical task - and no individual conversation will find it.

The point almost everyone misses

Ask why short-notice absence is expensive and the answer is not really the absent hour. It is that you could not cover it.

Two operations with identical absence rates can have completely different costs depending on whether there is anyone trained, available and contactable to fill the gap. One pays a premium to an agency at 7am; the other moves a shift and carries on.

Which means a meaningful share of what gets managed as an absence problem is actually a coverage problem, and coverage responds to things you control:

  • How many people can do each role, rather than the minimum that works on a good day
  • Whether you know who is available before you start ringing round
  • Whether swaps and extra shifts can be offered to the whole eligible group at once instead of in sequence
  • Whether the gap is visible early enough to be filled cheaply

None of that appears in a Bradford score, and all of it moves the number that matters.

If you keep it

It is not a bad tool. It is a bad policy. Keep the arithmetic as one signal among several, put it next to the return-to-work conversation rather than in place of it, and remember that it was designed to describe disruption to an operation - not to characterise a person.