Salary benchmarking for pay transparency means comparing each of your jobs with current market pay for the same work, level and location, and using that evidence to set pay ranges you can disclose and defend. Under the EU Pay Transparency Directive, candidates must be told the starting pay or pay range before the interview and employees can ask how their pay compares, so every range you share becomes a claim you may have to justify.
The Directive does not oblige you to benchmark. It does require pay to rest on objective, gender-neutral criteria, and it shifts the burden of proof to the employer in pay discrimination claims. A documented benchmark, built with a consistent method, is the most practical way to show where a range comes from.
This guide explains how to benchmark salaries for compliance: how to match jobs to market data, which percentiles to use, how to age data and build ranges, what the law allows market data to justify, and what TalentUp’s 2026 job ad data shows about how wide pay ranges really are.
Key takeaways
- Benchmark job content and level, not job titles: in TalentUp’s 2026 data, pay for the same title in one country typically spans 33–57% between P25 and P75.
- Advertised ranges are much narrower: the median posted range is 16–22% wide in Spain, the UK, France, Italy and Poland.
- Anchor ranges to a chosen percentile (usually P50), age the data to the date of use and cut it by country, industry and company size.
- Market data can inform pay, but it cannot on its own justify paying men and women differently for work of equal value.
- Document every step: matches, sources, percentiles, aging factor and range rules.
Why Salary Benchmarking Matters for Pay Transparency
Directive (EU) 2023/970 had to be transposed by member states by 7 June 2026. Its obligations apply to employers as each national law takes effect, and transposition is uneven across countries; our article on the 2026 EU pay transparency law tracks the timeline. Once your national rules apply, each obligation creates a concrete data need:
| Obligation (once transposed) |
What salary benchmarking provides |
| Starting pay or range given to candidates before the interview |
A current, market-based range for each job and level |
| No questions about pay history |
An offer anchored to the role’s range, not to the candidate’s last salary |
| Pay setting and progression criteria accessible to employees |
A written rule linking ranges to market percentiles and levels |
| Employee right to information on average pay by sex |
A job architecture that groups “same work or work of equal value” consistently |
| Gender pay gap reporting: 250+ workers yearly and 150–249 every 3 years from 7 June 2027; 100–149 every 3 years from 7 June 2031 |
Clean, comparable pay data by job and level |
| Joint pay assessment if an unexplained gap of 5% or more is not remedied within 6 months |
Evidence of the objective criteria behind each range |
Without a benchmark, the range in a job ad is often an inherited number nobody can explain. With one, it is the visible output of a documented method. For the full compliance plan beyond pay data, see our guide to preparing for the EU Pay Transparency Directive.
Can Market Data Justify Pay Differences?
This is where many benchmarking projects go wrong. The Directive requires pay structures that let employers assess whether workers are in a comparable situation regarding the value of their work, based on objective, gender-neutral criteria. Article 4(4) names them: skills, effort, responsibility and working conditions, and, where appropriate, other factors relevant to the specific job.
Market rates are not on that list. They still have a legitimate role, with limits:
- Market data sets the level of a range; job evaluation sets which jobs share a range. Two roles of equal value under your evaluation should not end up in different bands simply because one title looks better paid on a salary website.
- Shortage premiums can be justified, but only to the extent you can show them. In Enderby (C-127/92), the Court of Justice held that the state of the employment market can be an objectively justified economic ground for a pay difference between jobs of equal value, but only to the extent the market effect can be determined, applying proportionality.
- Market data is never a justification for individual gaps. Paying a woman less than a man in the same job because “she accepted a lower offer” is negotiation, not market evidence.
- Check the benchmark itself for bias. If female-dominated roles are matched to lower-paying market references because of how titles are worded, the benchmark imports the gap you are trying to remove.
Manoranjan Kumar Mahto, a total rewards and job architecture specialist who founded Total Rewards Hub, imagines an employer explaining a £10k gap between two differently titled roles with nothing but the market:
Manoranjan Kumar MahtoTotal Rewards & Compensation Strategy, Founder, Total Rewards Hub
Now imagine the organisation’s answer is: “The market is different.” Possibly. But which market? What is different about the work?
Balancing external competitiveness and internal fairness is a strategic choice; our article on market salaries vs. internal equity explains the trade-offs.
How to Benchmark Salaries in 8 Steps
1. Build or Clean Your Job Architecture
List every job with its function, family, level and location. Write or update job descriptions that capture responsibilities, scope, decision authority and required skills. Evaluate jobs with a gender-neutral method based on skills, effort, responsibility and working conditions, so you know which jobs are of equal value before you look at the market.
René Paparusso, Team Lead Talent Acquisition Management Switzerland at Georg Fischer, explains why this internal groundwork has to come before the market data:
René PaparussoTeam Lead Talent Acquisition Management Switzerland, Georg Fischer
One of the most persistent myths in compensation management is: “If we pay at market levels, we’re paying fairly.” […] Market data ultimately answers just one question: What are other companies paying for similar roles? […] The market provides a reference point. The compensation logic comes from within.
2. Match Jobs to the Market by Content, Not Title
Job matching is the step that most affects accuracy. Compare each internal job with a market reference that has the same core responsibilities and scope. A “Project Manager” who runs a team and a budget is not the same job as a “Project Manager” who coordinates schedules. Record each match and the reason for it; if no good match exists, combine references or use a blended match and say so.
3. Level Every Match
Market data for a title without a level mixes juniors and seniors. Match junior, mid, senior, lead or manager levels explicitly, using your own levelling guide. TalentUp’s data below shows why: for the same title and country, the top quartile of advertised pay is often 50% or more above the bottom quartile.
4. Choose Sources and Cuts
Pick data that reflects the market you actually recruit from:
- Country (and region or city where pay differs materially), never a European average for a local role.
- Industry, where pay differs, for example tech, financial services or retail.
- Company size, because large employers usually pay differently from small ones for the same job.
- Minimum sample size per cut. If a cut is too thin, widen it (for example, national instead of city) and document the decision.
5. Age the Data to the Date of Use
Market data is a snapshot. If data was collected 9 months ago and pay in that market is moving by about 3% a year, age it forward: 3% × 9/12 = 2.25%, so multiply by 1.0225 (illustrative figures). Apply the same aging rule to every job and keep the market movement assumption on file.
6. Pick Your Market Percentile
Decide your pay philosophy: lead, match or lag the market. Most employers anchor the range midpoint at P50 and use P75 for a small set of critical roles. Apply the rule consistently across jobs of equal value.
7. Build the Range Around the Anchor
Set a minimum, midpoint and maximum for each level. The midpoint follows your chosen percentile; the width reflects how much progression you expect within the level. The range you advertise to candidates is usually the hiring part of the band, not the whole band; our article on transparent salary job postings covers how to publish it.
8. Compare With Current Pay, Check Gaps and Document
Place every employee in their range. Flag people below the minimum and check pay by sex within each category of workers doing work of equal value. Fix unexplained gaps before they appear in a report, and document matches, sources, cuts, aging and percentile choices in one file you can show to employees, worker representatives or a court.
Percentiles: P25, P50 and P75 Explained
Salary benchmarks report pay as percentiles of the market distribution:
| Percentile |
Meaning |
Typical use |
| P10 |
10% of the market pays less |
Floor check: below this, hiring will be hard |
| P25 |
First quartile |
Range minimum for entry into a level; lag strategy |
| P50 (median) |
Half the market pays less, half more |
Range midpoint for a “match the market” philosophy |
| P75 |
Third quartile |
Midpoint for critical roles; lead strategy; range maximum |
| P90 |
10% of the market pays more |
Premium or hard-to-fill roles only, with a documented reason |
Two practical rules help with compliance. Use the median rather than the average, because a few very high salaries pull the average up. And use the same percentile for jobs of equal value: choosing P75 for a male-dominated role and P50 for a female-dominated role of equal value is exactly the kind of decision the Directive targets.
TalentUp Data: How Wide Are Pay Ranges in 2026 Job Ads?
To show what benchmarking adds over a single data point, we analysed the salaries employers advertised in European job ads between January and September 2026.
80–97%
of ads with structured minimum and maximum salary fields give a range rather than a single figure (Italy to Poland)
16–22%
median width of advertised ranges in Italy, Spain, the UK, France and Poland
33–57%
gap between P25 and P75 of advertised pay for the same job title in one country
Source: TalentUp job postings database, January–September 2026. See methodology below.
| Country |
Ads with structured salary fields |
Share of those that publish a range |
Median range width |
| Poland |
~25,800 |
96.8% |
22.1% |
| Spain |
~98,100 |
89.1% |
18.7% |
| Netherlands |
~10,400 |
88.2% |
29.4% |
| United Kingdom |
~44,600 |
86.9% |
18.8% |
| France |
~20,600 |
84.0% |
20.6% |
| Italy |
~19,000 |
79.7% |
15.9% |
Among ads whose salary comes in structured minimum and maximum fields (typically job boards with a salary form), most give a range, and the typical range is roughly one fifth wide (for example €40,000–€48,000). The Netherlands is the exception, with wider ranges (median 29.4%), partly reflecting public-sector salary scales in the Dutch data. Counting salaries written only in the ad text as well, which are more often single figures, the share of ranges is lower: at least 44% of ads that state pay use a range.
Market spread vs posted range, by role
The table compares the spread of advertised pay for one job title in one country (P25 to P75 of the midpoints of all ads) with the width of a typical single ad.
| Role and country |
Ads |
P25 |
P50 |
P75 |
P25–P75 spread |
Median posted range |
| Software engineer, Poland |
~870 |
€51,400 |
€64,200 |
€79,400 |
55% |
30% |
| Project manager, UK |
~470 |
€53,500 |
€67,300 |
€79,500 |
49% |
17% |
| Project manager, Spain |
~400 |
€28,500 |
€36,400 |
€44,600 |
57% |
19% |
| Project manager, Poland |
~380 |
€39,300 |
€51,800 |
€61,200 |
56% |
25% |
| Accountant, UK |
~400 |
€42,500 |
€52,200 |
€60,400 |
42% |
15% |
| Accountant, France |
~280 |
€30,500 |
€36,900 |
€44,200 |
45% |
19% |
| Accountant, Italy |
~290 |
€23,900 |
€27,600 |
€31,900 |
33% |
16% |
| Nurse, Spain |
~860 |
€22,100 |
€28,700 |
€34,100 |
54% |
10% |
| Customer service, Spain |
~300 |
€19,600 |
€22,200 |
€26,900 |
37% |
11% |
Gross annual pay, converted to euros. The lesson for benchmarking is clear: the market for one job title is two to five times wider than any single range. A project manager in Spain at P25 and one at P75 are, in practice, different jobs with different scope and seniority. If you benchmark by title alone, you can justify almost any number; if you match by content and level, you get a range narrow enough to publish and defend.
Choosing Salary Data Sources
No single source is perfect. Most compensation teams combine two or three and document why.
| Source type |
Strengths |
Limitations |
Best for |
| Traditional salary surveys |
Employer-reported actual pay, detailed levelling, total cash and bonus |
Expensive, participation required, data often 6–12 months old, limited coverage of small markets |
Large employers with established job architectures |
| Job ad data |
Large samples, near real-time, wide country and role coverage, shows what employers offer today |
Advertised rather than actual pay, titles need normalising and levelling, uneven salary disclosure by country |
Setting hiring ranges, tracking market movement, roles and countries surveys miss |
| Government and official statistics |
Free, official, large samples, good for macro context |
Broad occupation groups, published with long delays, rarely levelled |
Context and sanity checks, not role-level ranges |
| Collective agreements and public pay scales |
Legally binding minimums by category |
Minimums only, not market rates |
Floors in countries and sectors covered by agreements |
| Internal pay data |
Actual pay for your own jobs |
Reflects your past decisions, including any bias |
Comparison with the market, not as the market itself |
TalentUp combines job ad data from over 100 validated sources with normalisation, deduplication and validation by seniority, skills and experience, and lets you cut results by country, role, level, industry and company size. You can try a role in the free pay transparency salary benchmark or see country medians in the European Salary Benchmark Report.
Common Benchmarking Mistakes That Weaken Compliance
| Mistake |
Why it is a risk |
What to do instead |
| Matching by job title only |
Mixes levels and scopes; the range can be challenged as arbitrary |
Match by content and level, record the match |
| Different percentiles for jobs of equal value |
Can reproduce a gender gap through the pay policy itself |
One documented percentile rule for each job family or level |
| Old data, not aged |
Ranges fall behind the market and stop being an objective reference |
Age data to the date of use; refresh at least yearly |
| A European average for local roles |
Over- or under-pays in every country |
Benchmark each country separately |
| Using market data to explain individual gaps |
Not an objective, gender-neutral reason under the Directive |
Fix the gap or document a genuine, job-related reason |
| No written methodology |
The burden of proof lies with the employer |
Keep one file with sources, cuts, matches, aging and range rules |
Publishing ranges without this groundwork is how transparency backfires; see whether pay transparency can backfire and how much pay information to share.
Conclusion
Salary benchmarking is what turns pay transparency from a risk into a routine. Match jobs by content and level, use consistent percentiles for jobs of equal value, age and cut the data to your market and document every choice. The 2026 data shows why precision matters: the market for a job title is far wider than any range you can credibly publish, so the quality of your matching decides whether your ranges hold up when candidates, employees or courts ask how they were set. Explore plans on our pricing page.
Sources