Is Your Company — and Your Tool — Ready for Directive 2023/970?

“Directive-compliant reports” is now a standard claim on every vendor’s website. The problem is that the claim alone proves nothing. Directive 2023/970 imposes more than a dozen separate, very specific obligations. Only by mapping each of them to a tool’s actual features can you see whether a product will carry a company through the compliance process or simply calculate a single number.

Below we map every requirement of the directive to a concrete feature of the Levelly.ai platform. At the end, we answer the questions HR teams ask most often during conversations and demos.

What Directive 2023/970 actually introduces

Directive (EU) 2023/970 of the European Parliament and of the Council on pay transparency was adopted on 10 May 2023. Member States were required to transpose it into national law by 7 June 2026. In Poland, work on the implementing act is underway.

The obligation to report the pay gap periodically depends on the size of the organisation:

• 250 or more employees — first report by 7 June 2027, and annually thereafter.
• 150–249 employees — first report by 7 June 2027, then every 3 years.
• 100–149 employees — first report by 7 June 2031, then every 3 years.
• Under 100 employees — no reporting obligation, but the remaining rights (pay transparency in recruitment, the right to information) apply to every employer.

The directive is not only about reports, however. It introduces pay-range transparency as early as the recruitment stage and bans asking candidates about their previous pay. It gives employees the right to information on average pay and removes confidentiality clauses on pay. The employer bears the burden of proof. And when the gap reaches 5%, it triggers a joint pay assessment.

The compliance matrix: directive requirement vs Levelly.ai feature

Status legend: Available — the feature works in the platform. To confirm in a demo — scope depends on data and configuration; we show it live. Process support — the tool provides data and a signal, while the decision and the process remain with the organisation.

A. Reporting of metrics (Art. 9)

• Average pay gap — the directive: difference in the mean pay of women and men (basic pay). Levelly.ai: gap dashboard — mean, basic pay and basic pay with bonuses. Status: available.
• Median pay gap — the directive: difference in the median pay of women and men. Levelly.ai: dashboard — unadjusted gap as mean and median. Status: available.
• Gap in variable components — the directive: difference in bonuses and complementary components. Levelly.ai: metrics with and without bonuses. Status: available.
• Share of women and men receiving bonuses — the directive: proportion of women and men receiving variable components. Levelly.ai: compliance view — % of employees with a bonus, women and men. Status: available.
• Distribution across quartiles — the directive: share of women and men in each pay quartile. Levelly.ai: compliance view — pay distribution by quartile. Status: available.
• Gap by categories of workers — the directive: gap by category of work, in basic pay and in variable components. Levelly.ai: analysis by job levels (L1–L5) and adjusted gap. Status: available.

B. Pay transparency in recruitment (Art. 5)

• Pay ranges before employment — the directive: candidates know the pay or pay range before the interview. Levelly.ai: pay band data per level as a basis for setting pay ranges. Status: process support.
• Ban on asking about pay history — the directive: the employer does not ask about previous pay. Levelly.ai: outside the tool’s scope — a matter of the recruitment procedure. Status: process support.

C. Employee’s right to information (Art. 7)

• Information on average pay — the directive: an employee may request average pay by sex within their category. Levelly.ai: individual view — position relative to the female and male median and the band for the level. Status: available.
• Ban on confidentiality clauses — the directive: employees may not be barred from disclosing their own pay. Levelly.ai: outside the tool’s scope — a matter of policies and contracts. Status: process support.

D. Work of equal value and criteria (Art. 4)

• Categories of work of equal value — the directive: comparison within the same work or work of equal value, based on objective, gender-neutral criteria. Levelly.ai: grouping by levels L1–L5 and mapping the job structure to categories. Status: to confirm in a demo.
• Gender-neutral criteria — the directive: skills, effort, responsibility, working conditions. Levelly.ai: documentation of the criteria for assigning a level. Status: to confirm in a demo.

E. The 5% threshold and joint pay assessment (Art. 10)

• Signalling the 5% threshold — the directive: a gap of at least 5% in a category, not objectively justified, triggers an obligation to act. Levelly.ai: compliance view — gap by level with flagging of breaches. Status: available.
• Joint pay assessment — the directive: assessment with worker representatives, justifying differences. Levelly.ai: data and report history as the basis for the assessment; the process itself remains with the organisation. Status: process support.

F. Documentation and burden of proof (Art. 18)

• Reversed burden of proof — the directive: the employer proves the absence of discrimination. Levelly.ai: report history, monitoring, and documentation of results and recommendations. Status: available.
• Justification of differences — the directive: documenting objective causes of the gap. Levelly.ai: AI recommendations and the breakdown of the gap into explained and unexplained parts. Status: to confirm in a demo.

A methodological note: the part of the gap a model treats as “explained” does not automatically become legally justified, and the “unexplained” part is not automatically proof of discrimination. The tool points to areas for analysis. The justification and the decision are made by the organisation.

FAQ: the questions clients ask most

Does the directive apply to my company if we have fewer than 100 employees?

The obligation to report the gap periodically starts at 100 employees. The other rights, however, apply to every employer regardless of size: pay-range transparency in recruitment, the ban on asking about pay history, the employee’s right to information on average pay, and the ban on confidentiality clauses.

When does it take effect? Are we waiting for the Polish act?

The deadline to transpose the directive into national law passed on 7 June 2026. In Poland, work on the act is underway. The first reports for the largest companies fall on 7 June 2027, so now is the time to prepare data and processes. Collecting and organising pay data usually takes longer than expected.

What is the difference between the raw and adjusted gap, and which one must be reported?

The raw (unadjusted) gap is the simple difference between the mean or median pay of women and men. It is the basis for reporting. The adjusted gap accounts for factors such as job level, tenure or working time and shows how much of the difference remains unexplained. For the report, you need the raw gap; the adjusted gap helps you understand the causes and defend against a discrimination claim.

What is “work of equal value” and how do you define categories?

Work of equal value is work that differs in job title but is comparable in required skills, effort, responsibility and working conditions. The directive requires comparing pay precisely within these categories, set by objective, gender-neutral criteria. In Levelly.ai, levels L1–L5 are the starting point. In a demo, we show how to map your job structure to categories and how to document the criteria.

What happens if the gap reaches 5%?

If reporting shows a gap of at least 5% in a category of workers that cannot be justified by objective, gender-neutral criteria and is not remedied within 6 months, the employer must carry out a joint pay assessment with worker representatives. The platform flags threshold breaches at various levels and provides data for that assessment.

Do I have to raise salaries if we detect a gap?

The directive does not require mechanically equalising all differences. It requires that differences be justified by objective criteria or removed. You may not, however, lower pay to “close” the gap. Levelly.ai helps identify which differences require action and how much a correction would cost before you decide.

Who bears the burden of proof in a dispute?

The employer. In a pay discrimination claim, the company must show that the difference results from objective, gender-neutral factors. That is why documentation matters so much: report history, justifications of differences and a trail of actions taken. The platform stores and monitors all of it.

Can the results be trusted enough to make a pay decision?

This is the most important and entirely fair question. That is why we show not only the result but also the method: which variables enter the model, when we apply the Oaxaca–Blinder decomposition, how we treat bonuses, part-time work and small groups, and how the explained part differs from the unexplained one. It is best to verify this on your own or anonymised data during a demo.

Where is our pay data stored?

This is a frequent and legitimate question from HR and security teams. We discuss the details — data location, the data processing agreement (DPA), encryption, access control and retention period — individually, to meet your internal requirements. The platform processes HR and payroll data securely.

How long does implementation take and how much work does it require from HR?

The starting point is a CSV or Excel file. The platform guides you through uploading the data and detects errors before the analysis, so you don’t need an IT department or spreadsheets. The metrics appear right after validation. The scope of data preparation depends on how well your pay data is organised, and we discuss it during a demo.

We have entities in several EU countries — does each country implement this differently?

Yes. The directive sets a common minimum, but details such as thresholds, deadlines and penalties differ by country. Levelly.ai generates country-tailored reports, and the plan for larger organisations covers multiple entities (parent companies together with subsidiaries).

Check it on your own data

Marketing claims look similar across every vendor. The result on your data does not. In 30 minutes, we’ll show how the platform calculates the raw and adjusted gap, identifies risk groups and documents a decision on data close to yours. You’ll leave the meeting with a ready checklist of compliance points to tick off — not another slide.

See your compliance matrix live — book a 30-minute demo.

Prefer to explore the tool yourself first? Start for free, no card required.

Informational material. Not legal advice. The scope of compliance in a specific case depends on the final shape of the national act implementing Directive 2023/970 and on the organisation’s data and configuration. The “to confirm in a demo” statuses indicate features whose exact scope we show live.