A single headline number tells you which way the gap runs. Two numbers tell you why.
UK gender pay gap reporting requires employers to publish both a mean and a median. Most commentary picks one — usually the mean — and stops there. That is a mistake, because the relationship between the two is itself informative.
Two people in the same role on different salaries do not show up here; what shows up is where women and men sit across the whole pay range.
The mean divides total pay across all employees by headcount. The median finds the midpoint of the distribution: half earn above, half below. When the mean gap is larger than the median gap, the distribution is skewed by high earners. In most large British technology employers, that is exactly what the data shows — a cluster of very highly paid roles occupied predominantly by men pulling the mean upward. When the two figures sit close together, the imbalance is more evenly spread across the pay range rather than concentrated at the top.
- Mean pay gap
- total pay divided by headcount, then compared across genders; sensitive to outliers at the top
- Median pay gap
- midpoint of the pay distribution compared across genders; reflects spread across the whole range
- Mean > median
- gap concentrated at the upper end of the pay scale
- Mean ≈ median
- gap more evenly distributed across pay levels
- Bonus mean gap
- variable pay compared by mean; often larger than the salary gap in tech employers
- Bonus median gap
- midpoint comparison for bonus; separates senior-end skew from a broader bonus distribution problem
This matters for diagnosis. A fifteen-percentage-point mean gap alongside a six-point median gap points to a structural problem at the senior end: the pipeline, promotion rates, or the grading of specialist technical roles relative to managerial ones. The same fifteen-point mean alongside a twelve-point median points to something different — a more pervasive pattern that runs through the organisation at every level, not just above a particular threshold. The remedies are not the same, and a single headline number cannot distinguish between them.
What pay gap reporting measures is a distribution comparison, not an equal-pay test. Two people in the same role on different salaries do not show up here; what shows up is where women and men sit across the whole pay range. That framing matters when reading the pair. A company can close its mean gap substantially by promoting a small number of women into senior roles — and leave the median almost unchanged, because the midpoint of the distribution has not moved. Conversely, if part-time and lower-graded roles are disproportionately held by women, the median will widen even in an organisation with strong senior representation. The pair of numbers surfaces which of these is occurring.
There is one further check worth applying: the bonus gap, which the regulations also require. Bonus distributions in tech are heavily weighted toward senior and commercial roles, and bonus figures are reported as both mean and median on the same basis as pay. A large bonus mean gap alongside a modest pay mean gap signals that the pay scale itself is not the only lever, and that variable pay is amplifying the overall difference. Reading all four figures — mean pay, median pay, mean bonus, median bonus — takes under a minute and yields considerably more than the single number most headlines carry.