Teams searching for how to calculate gender pay gap usually need two answers at once. They need the formal formula that produces a clean number, and they need an operating method that survives real-world data. The formula itself is simple. The difficulty comes from deciding which employees belong together, which elements of pay to include, and how to interpret results without overreacting to noisy populations. If you skip that groundwork, your pay gap number may be mathematically correct but operationally useless.
This guide breaks the process into the same steps strong employers use internally: define the population, clean the pay fields, calculate mean and median gaps, review results by category, and separate unadjusted results from the adjusted analysis that helps explain root causes. That sequence keeps the reporting number and the remediation strategy connected.
Start with the basic formula
The most common unadjusted gender pay gap formula compares average male pay with average female pay and expresses the difference as a percentage of average male pay. This is the version many employers use in high-level reporting because it creates a consistent baseline across populations.
If average male pay is EUR 60,000 and average female pay is EUR 54,000, the gap is 10%. The calculation is `(60,000 - 54,000) / 60,000 x 100 = 10`. That tells you women are earning 10% less than men on average in the population you selected. It does not yet tell you why. That distinction matters. The gap metric is a signal, not a verdict.
Mean vs median gap
The mean pay gap uses the arithmetic average. It is useful because it reflects total payroll distribution, including the effect of outliers and senior leadership concentration. If a small number of highly paid men sit in the top of the organization, the mean gap will often widen sharply. That is not a flaw. It tells you something real about how rewards are distributed.
The median pay gap uses the midpoint value in each population. It is more resistant to outliers and often gives a clearer view of the typical employee experience. If the mean gap is much larger than the median gap, your issue may be concentrated in senior or specialist roles. If both are large, the imbalance may be more systemic.
Good reporting includes both because they answer different questions. Mean shows the overall payroll effect. Median shows the central tendency. Neither replaces the other.
Adjusted vs unadjusted gap
The unadjusted gap is the first number most authorities and executives want to see because it describes the result without controlling for explanatory variables. But employers then need an adjusted view that tests whether the gap persists after accounting for legitimate factors such as grade, role, geography, working time, or tenure. The adjusted analysis is where you investigate whether the same or equivalent work is being paid differently.
In practice, the unadjusted gap answers “what is happening?” and the adjusted gap answers “what may be driving it?” The two should never be confused. A low adjusted gap does not erase a high unadjusted gap if women are still concentrated in lower-paid parts of the organization. Likewise, a modest overall gap can hide serious category-level issues that need direct remediation.
Calculate by relevant category, not just company-wide
A single company-wide number is useful for disclosure and board reporting, but it is too blunt for action. The most defensible approach is to calculate the gap within categories of workers performing the same work or work of equal value. That is also closer to how the EU pay transparency regime expects employers to analyze issues. When you calculate by category, you can see whether the problem sits in engineering, sales, operations, customer support, or another part of the workforce.
Example: suppose an employer has an overall gap of 7%, but the company-wide number is pulled mainly by two categories. Operations has a 3% gap and is broadly stable. Customer Success has a 12% gap because women are underrepresented in senior account roles. Commercial Leadership has a 14% gap because variable pay and market adjustments were not consistently governed. Without category analysis, those root causes remain hidden.
Worked example: mean gap calculation
Imagine a category with 20 men and 24 women. Men have average base salary of EUR 58,000 and women have average base salary of EUR 54,520. The category mean gap is `(58,000 - 54,520) / 58,000 x 100 = 6`. On its own, that tells you the category is above a 5% working threshold and needs investigation. Next, check whether the category definition is sound. If it is, review seniority mix, performance-based pay, recent promotions, and any discretionary increases.
If the difference cannot be explained with objective, gender-neutral criteria, the organization should model the cost to close the gap. In this example, reducing the gap to 5% would require women’s average pay to rise to EUR 55,100. That implies an average increase of EUR 580 for the underpaid population, or roughly EUR 13,920 annually across 24 employees. This is why finance and reward teams need category-level remediation math, not just a dashboard score.
Common calculation mistakes
The most common error is mixing incomparable populations. If you calculate a pay gap across a category that contains entry-level coordinators, senior managers, and country leads, the number may be technically true but analytically weak. Another frequent problem is mixing pay elements inconsistently, such as comparing base salary for one group and total cash compensation for another. Missing or mislabeled gender data can also distort results, especially in small samples.
A subtler mistake is over-trusting a single result. One snapshot can be affected by timing, hiring cycles, or a one-off variable pay event. Strong teams review trends over time, compare mean and median signals, and combine the gap number with job architecture and salary band data before deciding what to remediate.
See it in the product
How PayComply automates the workflow
PayComply normalizes compensation data, groups roles into comparable categories, calculates the gap by category, flags the populations above threshold, and shows remediation scenarios before you export the report. That cuts out the usual spreadsheet handoffs between HR, reward, and finance.
A repeatable calculation process
Step one is define scope: legal entity, country, and reporting population. Step two is clean the data and align the pay fields you want to compare. Step three is build categories that reflect the same work or work of equal value. Step four is calculate mean and median gaps overall and by category. Step five is review outliers and perform the adjusted analysis. Step six is convert the findings into action: documentation, pay adjustments, structural fixes, or manager controls.
The important point is that calculation is only the middle of the workflow. Most organizations do not fail because they cannot divide one number by another. They fail because the data model is weak, the categories are unstable, or the remediation step never gets translated into an approved compensation plan. A good gender pay gap process links measurement to action quickly enough that the next reporting cycle actually improves.