The EU Directive requires seven pay gap metrics, not one. Most employers are preparing for the wrong calculation.

When European employers discuss pay gap reporting, the conversation almost always begins with a single number: "our gender pay gap is X%." This framing is understandable. It is also dangerously incomplete.

The EU Pay Transparency Directive (2023/970) does not ask for one number. Article 9 requires employers to report seven specific metrics, each designed to expose a different dimension of pay inequality. Some reveal overall organisational imbalance. Others isolate variable pay disparities. One breaks the workforce into quartiles. And the most consequential metric, the per-category gap under Article 9(1)(g), is the one that triggers mandatory remediation under Article 10.

This guide sets out the complete calculation methodology that EU employers need to follow, from defining comparable work groups to interpreting the 5% threshold. It is written for HR directors and compensation teams at companies with 250 or more employees who will need to submit their first report by 7 June 2027.

The seven Article 9 metrics

Article 9(1) of the Directive lists the information that employers must provide to their designated national monitoring body. The seven metrics form a structured diagnostic, not a single headline figure.

Metrics (a) through (f) are organisation-wide calculations. They require no grouping logic. Metric (g) is fundamentally different: it requires the employer to first define categories of workers performing "equal work or work of equal value" under Article 4, then calculate the gap within each category. This is where the analytical complexity sits, and where most compliance failures will occur.

Step 1: Define your comparable work groups

Before any calculation begins, you need to answer a structural question: which employees are being compared to which? Article 4(4) of the Directive provides the criteria. Work is of equal value when it is comparable on four dimensions: skills, effort, responsibility, and working conditions. These are the same four criteria used in the EIGE/EC gender-neutral job evaluation guidelines published in March 2026.

The practical implication is that comparable work groups must be formed using a structured job evaluation methodology, not by job title alone. Two roles with different titles but similar evaluation scores belong in the same group. Two roles with the same title but materially different responsibilities may need to be separated.

The number of groups is not prescribed by the Directive. It depends on the complexity of the organisation's role structure. A company with 300 employees might have 4 to 8 groups. A company with 1,500 might have 8 to 15. The key constraint is that each group must contain enough employees of both genders to produce a statistically meaningful comparison. Groups with fewer than six employees of either gender should be flagged for suppression or aggregation.

Step 2: Collect and validate your pay data

Article 3 of the Directive defines "pay" broadly. It includes not only basic salary but also complementary and variable components: bonuses, overtime pay, travel allowances, housing allowances, training compensation, severance payments, and statutory sick pay. Employer pension contributions are included where they constitute pay under national law.

This is wider than most payroll exports. The most common data quality failure is incomplete variable pay capture. If your bonus data sits in a different system from your base salary data, you need to merge them before any calculation begins. Missing variable components will systematically understate the gap if men receive more variable pay on average, which is the case in the majority of European organisations.

All pay figures should be normalised to full-time equivalent (FTE) values. Part-time employees must be included in the calculation on an FTE-adjusted basis. Excluding part-time workers, a disproportionate share of whom are women in most EU labour markets, would undermine the validity of every metric.

Step 3: Run the calculations

Once work groups are defined and pay data validated, the calculations themselves are formulaic. The mean gap is the difference in average pay between male and female employees, expressed as a percentage of average male pay. The median gap uses the midpoint of each gender's pay distribution instead of the average.

For the organisation-wide metrics (a) through (d), you aggregate all employees regardless of work group. For the per-category metric (g), you run the same formula within each comparable work group separately. This produces a set of gap figures, one per group, rather than a single number.

The quartile metric (f) requires a different approach: rank every employee in the organisation by total pay, divide the ranked list into four equal groups, and report the percentage of women and men in each quartile. This is an organisation-wide calculation with no grouping logic.

The variable pay proportion metric (e) is the simplest: for each gender, count the number of employees who received any form of variable pay during the reporting period, divided by the total headcount of that gender.

Step 4: Interpret the 5% threshold

Article 10 of the Directive is triggered when the per-category pay gap (metric g) exceeds 5% in any comparable work group, and the employer cannot justify the difference using objective, gender-neutral criteria. This is not an organisation-wide threshold. A company with a 2% overall gap can still be required to conduct a joint pay assessment if a single work group shows a 7% gap.

The joint pay assessment under Article 10 must be conducted in cooperation with workers' representatives (in Germany, the Betriebsrat). It requires a detailed analysis of the causes of the gap, the remediation measures taken or planned, and the effectiveness of previous measures. The employer has six months from the reporting date to complete this assessment if no objective justification can be provided.

The adjusted gap: what it means and when you need it

The seven Article 9 metrics are all unadjusted. They report the raw difference in pay between male and female employees without controlling for factors such as experience, education, or performance.

The adjusted gap becomes relevant at the justification stage. If a comparable work group shows a gap above 5%, the employer must demonstrate that the difference is attributable to objective, gender-neutral factors. This typically requires a regression analysis that isolates the impact of legitimate pay determinants (tenure, qualifications, location, shift patterns) from the residual gap that cannot be explained. The residual is the adjusted gap.

A critical distinction: the unadjusted gap tells you about the pay inequality in your organisation. The adjusted gap tells you about the pay discrimination. Both matter. The Directive requires you to report the first and justify the second. Employers who focus exclusively on the adjusted figure (because it is typically smaller) will find themselves unprepared for the public scrutiny that comes with mandatory reporting of the unadjusted numbers.

Practical sequencing: what to do first

For employers who have not yet begun their methodology work, the following sequence reflects the dependencies in the process. Each step builds on the one before it.

1. Complete a gender-neutral job evaluation. Use the EIGE/EC four-criteria framework (skills, effort, responsibility, working conditions) or the German EG-Check GleichWertCheck. This produces the comparable work groups that every subsequent calculation depends on. Without this, you cannot calculate the per-category metric (g) or assess Article 10 compliance.
2. Audit your pay data completeness. Map every pay component against the Article 3 definition. Identify where variable pay data sits, whether it can be linked to individual employees, and whether FTE adjustments are reliable. Flag any components that are stored in separate systems from base salary.
3. Run the seven calculations. Produce the organisation-wide metrics (a) through (f) and the per-category metric (g) for each comparable work group. Document the methodology in sufficient detail for Works Council review.
4. Assess the 5% threshold per group. For any group exceeding 5%, prepare a justification analysis using objective, gender-neutral criteria. Where the gap cannot be justified, begin scoping the joint pay assessment with worker representatives.
5. Lock the methodology before the first report. The first report is due 7 June 2027. Working backwards, the methodology should be finalised by Q4 2026 to allow time for a dry run, data validation, and Works Council consultation before the live report is submitted.

The calculation is the easy part

The formulas for calculating a gender pay gap are arithmetic. Any competent payroll team can divide averages. What separates a defensible report from a fragile one is the structural work that precedes the calculation: defining comparable work groups on gender-neutral criteria, capturing the full scope of pay components, and understanding that the 5% threshold applies per group, not per organisation. Employers who treat this as a spreadsheet exercise will produce a number. Employers who treat it as a compliance programme will produce a report that withstands scrutiny from monitoring bodies, workers' representatives, and, under Article 18 of the Directive, the courts.