EU Pay Transparency Directive 2026: Getting Salary-Range Ready

The EU Pay Transparency Directive is in force and only 23% of organizations are ready. How to build defensible, data-backed salary ranges for every job post.

Ernest Bursa

Ernest Bursa

Founder · · 12 min read
Head of People marking P25, median and P75 salary band brackets beside a printed job posting on a glass wall in a startup office

The EU Pay Transparency Directive (Directive (EU) 2023/970) is now in force. Since 7 June 2026, employers hiring in the EU must tell candidates the pay level or pay range for a role before the first interview, and they may no longer ask applicants about their salary history. If you post jobs into any EU market and you still set offers ad hoc, every new vacancy is now a compliance decision, not just a hiring one.

This guide covers what the directive actually requires at the recruitment stage, what makes a salary range “defensible” rather than decorative, and how to build a per-role band from real market data that doubles as your audit trail.

The deadline already passed, and only 23% of organizations are ready

The transposition deadline was 7 June 2026, and the European Commission confirmed it would not move. The deadline has now passed, and readiness is strikingly low: only 23% of organizations say they are fully prepared for the directive, according to Payscale’s 2026 Compensation Best Practices Report, a survey of 3,413 compensation and HR professionals.

National rollout is messy, which is tempting cover for doing nothing. Italy, Slovakia, and Lithuania transposed on time. The Netherlands, Sweden, Czech Republic, and Denmark have confirmed delays to 1 January 2027. Around ten member states face potential infringement proceedings for missing the deadline entirely.

Do not mistake that mess for breathing room. Three reasons:

  1. The direction is locked. Every member state must implement the directive; delays change the date, not the obligations.
  2. You probably hire across borders. A startup hiring in Poland, Germany, and Spain complies with the strictest applicable regime in practice, because one job post often serves several markets.
  3. The first gender pay gap reports for 250+ employee companies are due in June 2027, covering calendar-year 2026 data. The pay decisions you make this quarter are already reportable history for larger employers, and the structural requirements trickle down: companies with 100 to 249 employees report every three years.

The gap between “law in force” and “organizations ready” is the whole story. 77% of organizations are now operating under a transparency regime they have not prepared for.

What the EU Pay Transparency Directive requires when you post a job

At the recruitment stage, the directive imposes two concrete obligations. Employers must disclose the initial pay level or pay range for a position, either in the job vacancy notice or before the first interview. And employers are prohibited from asking candidates about their current or past salary. Both apply regardless of company size.

That is the short version, and it touches every single job post. A few practical implications follow directly:

  • “Before the first interview” is the legal floor, not the norm. Several member states (and most candidates) expect the range in the posting itself. Posting it is also the safer operational default: one artifact, publicly timestamped.
  • The range has to be a real range. A band of “EUR 30,000 to EUR 120,000 depending on experience” defeats the purpose and invites scrutiny. Regulators and candidates both read it as evasion.
  • Salary history questions disappear from your interview kits. If your screening call template includes “What are you currently earning?”, delete it now. “What are your salary expectations?” remains fine.

The directive also grants employees the right to ask for their own pay level and the average pay levels, broken down by sex, for categories of workers doing the same work or work of equal value. That is a post-hire obligation, but it is built on the same foundation as the job-post range: a pay structure you can explain.

Why posting a range is the easy part and “defensible” is the hard part

Most companies can paste two numbers into a job ad by Friday. The hard question is where those numbers come from, and whether you can still stand behind them when someone asks you to justify them.

The market is already halfway there on the easy part. 57% of organizations say they post salary ranges in job ads, and 42% do it across all jobs regardless of location, per Payscale’s 2026 CBPR. Measured from the other side, ZipRecruiter’s Economic Research found 53.3% of actual job postings included pay information as of September 2025. Those are different denominators, organizations versus postings, but they corroborate the same picture: just over half the market discloses, and the other half is now non-compliant in the EU.

Here is what “just over half” hides: posting a number is not the same as being able to defend it. Under the directive, two failure modes matter:

Failure mode What it looks like Consequence
No range Job post without pay information, or asking salary history Direct breach of the directive’s recruitment provisions
Indefensible range A number with no documented methodology behind it Exposure in pay gap reporting, employee information requests, and joint pay assessments

The second failure mode is the one that catches prepared-looking companies. If your band for a backend engineer was set by “what we paid the last one, plus a bit,” you have a number but no defense. The directive’s enforcement machinery, from reporting to assessments, runs on whether you can show objective reasons for pay differences. A range without a methodology is just a guess with a deadline.

The four gender-neutral criteria your pay bands must rest on

The directive requires pay structures built on four objective, gender-neutral criteria: skills, effort, responsibility, and working conditions. The European Commission and EIGE published EU-wide guidelines on gender-neutral job evaluation and classification in April 2026 to standardize exactly this.

In plain terms: you may pay two people differently, even in similar roles, if the difference maps to one of those criteria and you apply the criteria consistently. You may not pay them differently because one negotiated harder, anchored low from a previous job (now an illegal question anyway), or joined in a hot quarter.

For a tech company, the criteria translate cleanly:

  • Skills: a Go engineer commanding more than a PHP engineer is a skills-based, gender-neutral differentiator, provided the market data backs it and you apply it to everyone.
  • Effort and responsibility: a staff engineer owning an on-call rotation and architectural decisions sits in a higher band than a mid-level engineer. Document what the band requires, not who is in it.
  • Working conditions: location-based differentiation (Warsaw versus Gdańsk cost-of-labor) is permissible when grounded in market data and applied uniformly.

Notice what every one of those justifications has in common: they only work if you can point to evidence. Which brings us to the practical question every law-firm explainer skips: where does the number actually come from?

How to build a salary band from real market data: a worked example

A defensible band has three components: a market median, a percentile spread, and a documented sample. Here is what that looks like for a real role in a real EU market.

Say you are hiring a backend engineer in Poland, an EU member state directly under the directive. Pull current market listings for the role cluster and you get something like this (illustrative figures from a June 2026 benchmark of roughly 500 salaried listings):

Benchmark component Value (PLN, monthly) What it gives you
P25 ~16,000 Bottom of your posting range
Median ~17,000 Your anchor for a standard offer
P75 ~25,000 Top of your posting range
Sample size ~500 listings Your methodology footnote

That P25 to P75 spread is the pay range your job ad now legally needs. Not a number you reverse-engineered from budget, but a band grounded in what the market actually pays for this work, today.

The breakdowns underneath the band are where defensibility lives:

  • By technology (the skills criterion): in the same benchmark, Go roles showed a median around 25,000 PLN, Java around 20,000 PLN, and PHP around 12,000 PLN. If your Go band sits above your PHP band, that is not bias; it is a documented, job-related skills differential.
  • By region (the working-conditions criterion): Warsaw listings averaged roughly 20,500 PLN against roughly 16,000 PLN in Gdańsk. Location-based band variation is defensible when it tracks market data this directly.
  • By trend: the benchmark trend pointed up, which means a band set today is current, and also means a band set from last year’s anecdotes is already stale.

The procedure generalizes to any role and market: pick the role cluster, read the median and the P25 to P75 spread, post that as the range, and record the sample size and breakdowns. Total time, with the right data source, is minutes per role. Total time without one is a consulting engagement.

Turning a benchmark into your audit trail: the 5% rule and joint pay assessments

The directive’s sharpest tooth is the joint pay assessment. If gender pay gap reporting reveals a gap of at least 5% within any category of workers that cannot be justified by objective, gender-neutral criteria, and the employer does not remedy it within six months, the employer must conduct a joint pay assessment together with worker representatives. That means opening your pay structure to structured scrutiny, with documented follow-up obligations.

Your defense in that scenario is not eloquence. It is paperwork that predates the question:

  1. A job architecture. Clear role categories (“Backend Engineer,” “Data Engineer,” “Product Manager”) so “same work or work of equal value” comparisons happen on your terms, not a regulator’s reconstruction.
  2. A benchmark per band. The median, percentiles, sample size, and capture date for each range you post. This is the document that turns “we felt 17,000 was right” into “the market median for this cluster was 17,000 across ~500 listings in June 2026.”
  3. Documented differentiators. The technology, seniority, and regional breakdowns that explain why two people in adjacent roles sit in different bands, mapped to the four criteria.
  4. A refresh cadence. Benchmarks dated and re-run on a schedule, so nobody can claim your 2026 band was set on 2023 instincts.

None of this is exotic. It is the same artifact, a data-backed benchmark, kept instead of discarded. Most companies already do the work informally when they set an offer; the directive simply requires that the work be objective, consistent, and written down. If you are also navigating the EU AI Act’s hiring provisions, the pattern will feel familiar: regulators reward teams that can produce records, not promises.

Compliance and competitiveness are the same move

Here is the part the legal explainers undersell: the work the directive forces is work that wins candidates anyway. 67.2% of employers say posting pay information helps them attract quality candidates, per ZipRecruiter’s 2025 employer survey. And transparency is becoming table stakes regardless of law: 49% of organizations are targeting organization-wide or public pay transparency in 2026, up from about a third the year before (Payscale CBPR).

A market-grounded band also protects you in both directions. Post below P25 and the data tells you before three weeks of silence does. Anchor offers above P75 across the board and you are quietly overpaying against the market with no offsetting signal. The same benchmark that satisfies an auditor stops you from guessing in either direction.

So the readiness move is not “add a compliance step to hiring.” It is: build job-level ranges from real market data once, attach them to every posting, and keep the output. You satisfy the recruitment-stage obligations, you pre-build your joint pay assessment defense, and you publish ranges competitive enough to actually convert candidates. One action, three returns.

Get salary-range ready with Kit

This bridge from “you must post a defensible range” to “here is the defensible range” is exactly what Kit’s CompensationResearch is built for. Kit continuously benchmarks real job-market listings and turns them into the artifacts the directive demands:

  • Salary benchmarks per role cluster: median plus P25 to P75 band with sample size, the compliant posting range and its methodology footnote in one output.
  • Market trends by region and technology: the objective, gender-neutral differentiators (skills, responsibility, working conditions) that justify band variation in a joint pay assessment.
  • A ready job architecture: 35 role clusters across Engineering, Data, Product, and more, the clear job classification every compliance guide tells you to build first.
  • Role comparisons: side-by-side views for “same work or work of equal value” analysis before a regulator does it for you.

Because Kit is also your hiring pipeline, the range does not live in a spreadsheet nobody updates. You benchmark the role, attach the band to the job posting, and the audit trail accumulates as a byproduct of hiring.

The directive’s deadline has passed and the readiness gap is real: 23% prepared, everyone else exposed. But the fix is smaller than the panic suggests. One defensible, data-backed band per role, documented and refreshed, covers the job-post obligation today and the audit question tomorrow. Start with the next role you post.

Related articles

Ready to hire smarter?

Start free. No credit card required. Set up your first hiring pipeline in minutes.

Start hiring free