The Arithmetic That Penalises Part-Time Staff
A denominator expressed as a flat annual figure turns a three-day contract into a sixty per cent score. The correction is one multiplication.
A person contracted for three days a week, fully committed to client work, scores around sixty per cent against a denominator of 2,080 hours. A full-time colleague doing proportionally the same work scores close to a hundred.
The people risk in “The Arithmetic That Penalises Part-Time Staff” is not solved by collecting a more detailed activity score. Used for employee time clock records, compare the published capabilities can provide transparent time and project context, but individual conclusions still require a published method, a correction route, direct conversation and proportionate human review.
The gap is arithmetic. It is also the most common unfair feature of this measure, and it survives in firms that have thought carefully about everything else.
As a separate benchmark for “The Arithmetic That Penalises Part-Time Staff”, see FTC data-security guidance. External guidance is most useful when it prompts a specific check rather than replacing case-by-case judgement.
Where the flat denominator comes from
Convenience. A single number applied to everybody makes the report simple and the comparison apparently clean.
- Contracted hours for the period, from the contract in force.
- Less leave taken, from an entitlement itself pro-rated.
- Less public holidays, pro-rated by the contracted fraction.
- Less committed internal time, where the commitment is formal.
It is correct for the majority of staff, which is why it persists: it is only wrong for the people it is wrong for, and they are a minority.
The correction
Denominator equals contracted hours for the period, less leave, less public holidays pro-rated, less any committed internal time.
One multiplication by the contracted fraction. It is not difficult and it has to be in the report rather than in the reader's head.
Why "in the reader's head" fails
Because the figure travels. A reader three handoffs away sees sixty per cent with no note about a contract.
The handoff note here describes the mechanism. Part-time staff are its most frequent casualty.
Compressed hours
A person working full-time hours over four days. The denominator is unchanged; nothing needs correcting.
Firms occasionally pro-rate by days rather than hours and produce a figure above a hundred, which then gets queried as an error rather than as a definition fault.
Mid-year changes
Somebody moving from five days to three in July. The annual denominator has to be built from two periods.
This is routinely got wrong, and it produces a figure that is unexplainable to the person it describes.
Parental leave and phased return
A person returning at two days a week for three months, then four, then five. Three denominators in one year.
The phased return note here covers the fuller picture. The arithmetic point is that a single annual denominator cannot represent it.
Public holidays pro-rated
A three-day week does not encounter all eight public holidays. Pro-rate by the contracted fraction rather than deducting the full eight.
Small, a point or so, and it is the kind of detail that makes a person trust the figure or not.
The leave entitlement interaction
Part-time entitlement is itself pro-rated, so deducting a full-time entitlement from a pro-rated denominator double-counts.
Two pro-rations, applied once each. Getting this wrong in either direction produces a figure several points off.
The check worth running
List every person on a non-standard contract and compare their denominator with their contracted hours less leave.
Firms that do this for the first time typically find at least one person whose figure has been systematically understated for years.
What the person experiences
A number that has been wrong about them since they changed contract, appearing in reviews, which they have had to explain each time.
In practice people stop explaining it, and the figure sits in the record uncorrected.
Who the error falls on
Part-time staff are over-represented among people with caring responsibilities. A measure that penalises part-time contracts distributes its unfairness in a pattern a firm should not want.
That is a reason beyond accuracy to get the arithmetic right.
Why this error is so durable
Because it is invisible to everybody except the minority it affects, and that minority has usually stopped complaining about it.
A full-time reader checking the report sees correct figures. The denominator is right for them, and nothing on the page indicates that it is a constant rather than a derivation.
So the error has no natural discoverer. It persists until somebody affected raises it formally, or until a denominator rebuild is run for other reasons.
The report format that prevents it
Contracted hours as a visible column, beside available hours and chargeable hours. Three numbers per person.
A reader seeing contracted hours of 1,092 next to available hours of 2,080 has found the error without looking for it, which is the property a good format has.
Formats catch what policies do not, because a format applies to every row automatically and a policy applies only where somebody remembers it. That is the same argument the note on travelling figures makes about the definition line, and it is the general case: for recurring reports, change the shape of the page rather than writing a rule about how to fill it in.
The two pro-rations
Contracted hours pro-rated by the contracted fraction, and leave entitlement pro-rated by the same fraction before it is deducted.
Applying one and not the other is the commonest implementation error, and it goes in either direction depending on which was missed.
Both applied once, in that order, gives a denominator that matches what the person could actually have worked, which is the only denominator that means anything for them.
The one-line rule
Every denominator is built from the individual's contract, not from a firm-wide constant, and the contracted fraction appears beside the figure.
Both halves matter. The second is what stops the error returning the next time somebody rebuilds the report.