Salary Benchmarking: How to Implement It Step by Step (2026)

Salary benchmarking is the process of comparing what you pay for each role with what the market pays for the same work, at the same level and in the same location, so that pay ranges and offers rest on evidence rather than habit. Implementing it well means turning that comparison into a repeatable process: a defined scope, roles matched on content, data from several sources aged to a common date, clear calculations and a documented refresh cycle.

This guide walks through that process step by step, with a worked example for one role, and TalentUp data from about 3,700 job ads showing why matching on job titles alone gives misleading results. If you need the basics or a compliance-first view, see our guides to salary benchmarking for pay transparency and international salary benchmarking.

Step 1: Set the Scope and Your Market Position

Start by writing down what the benchmarking is for: setting ranges for a new pay structure, preparing the annual salary review, pricing a hard-to-fill role or checking competitiveness after attrition. The purpose decides the scope.

• Roles: you do not need to price every job. Pick benchmark jobs, roles that are common in the market, easy to match and cover each job family and level, then slot the remaining roles relative to them.
• Markets: the labour markets you actually hire from: country, city or region, and sometimes sector or company size.
• Pay elements: base salary only, or also target bonus and total cash. Keep the same definition across all sources.
• Effective date: the date your new ranges or increases take effect. All data will be aged to it.

Then set your market position, the percentile of the market you want to pay at. Matching the market usually means the median (P50), leading means around P75, and lagging means below the median, often offset by benefits, flexibility or equity. Many companies lead for critical roles and match for the rest. Our guide to writing a compensation philosophy covers this decision in detail.

Step 2: Build a Job Architecture and Match on Content

Job matching is the step where most benchmarking errors start. Before you can compare pay, every internal role needs a place in a job architecture: a job family (for example Finance), a function or sub-family (Accounting) and a level defined by scope, autonomy, complexity and impact, not by years of service.

Then match each benchmark job to the market role that does substantially the same work at the same level:

1. Read the job content, not the title. Compare main responsibilities, scope (team, budget, clients), required skills and decision rights.
2. Match the level separately. A “Project Manager” who runs a single internal workstream and one who owns a multi-million euro client programme are different levels, even with the same title.
3. Use the closest match and record its quality. If no market role fits well, use a close benchmark and adjust, or slot the role between two benchmarks. Never force a poor match.
4. Keep a match log. For each role, record the market role, level, source, a match quality rating (good, partial, weak) and who approved it. You will reuse it next year and it is your audit trail.

Job matching is also where internal equity and market competitiveness meet: if two internal roles share a level, their market matches should be at comparable levels too.

Why Job Titles Mislead: TalentUp Data 2026

To show why matching on titles alone fails, we analysed job ads published in 2026 with a stated annual salary range for five common titles in the United Kingdom, Spain and France.

Source: TalentUp job postings database, 1 January–30 September 2026, about 3,700 employee job ads with a stated annual salary range in the UK, Spain and France.

Most titles carry no level at all. Across the sample, 84.6% of ads use a bare title such as “Accountant” or “Software Engineer”, with no junior, senior, lead or equivalent. The share is lowest in Spain (11.1% state a level) and highest in the UK (25.2%), with France at 17.5%.

The same title spans a wide pay range. The table shows the spread of advertised salaries (midpoint of each ad’s range) for the same title in the same country.

When the level is stated, pay moves. Where we have at least 20 ads per group:

Medians of advertised range midpoints; “–” means fewer than 20 ads.

Responsibility keywords matter too, and not always in the same direction. In Spain, project manager ads with an IT, software or digital keyword have a median of €45,000 (n=39), against €39,000 for plain or other project manager titles (n=139) and about €35,000 for construction and engineering titles (n=19, indicative). In the UK the order flips: construction and engineering project managers advertise a median of £60,000 (n=42), IT project managers about £57,300 (n=38) and other titles £52,500 (n=143).

The French accountant figures are a useful warning in the other direction: a “confirmé” or “senior” label adds almost nothing to the median, because many unlabelled ads are already for experienced roles. A title, with or without a level, is a weak proxy for the work. Benchmark on job content and level, and treat any title-only source with caution.

Step 3: Choose, Age and Blend Your Data Sources

No single data source is complete. Most compensation teams combine two or three and weight them according to how well each matches the role and how current it is.

Our methodology page explains how TalentUp turns job-ad data into salary benchmarks.

Age the data to a common date

Surveys and statistics describe the market at an earlier date. Aging moves each data point forward to your effective date with an assumed annual rate of market movement:

Aged value = market value × (1 + annual rate × months between data date and effective date / 12)

Pick the rate from a source you can explain, such as salary increase budget forecasts for your countries (see salary increase budgets for 2027 in Europe) or official wage indices from Eurostat or national statistics offices, and use the same rate for all sources in the same market.

Blend the sources with documented weights

Give each source a weight based on match quality, sample size and freshness, then calculate a weighted market reference. Write the weights down before you see the result, so they are not tuned to reach a number someone wanted.

Worked example: pricing one role (illustrative)

A software company in Madrid is pricing a Backend Software Engineer at level P3 (works independently, owns services end to end, no line management). Its policy is to pay at the market median, and new ranges take effect on 1 January 2027. All figures below are illustrative, not market data, and the aging rate is an assumption.

The calculation is 0.5 × 42,490 + 0.3 × 41,400 + 0.2 × 44,129 = €42,491, rounded to €42,500. Survey B gets a lower weight because its match is only partial.

Step 4: Calculate Compa-Ratio and Market Position

With a market reference for each role, you can measure where people and roles sit. Two ratios do most of the work:

• Compa-ratio = employee base salary ÷ range midpoint. It shows where one person sits against the reference for their role: 1.00 is at the midpoint, 0.90 is 10% below.
• Market index (or market position) = your average pay for a role ÷ market reference at your target percentile. It shows where your company sits for that role.

Continuing the example, the midpoint is set at the market reference of €42,500:

The average pay of the three, €38,833, gives a market index of 0.91: for this role the company pays about 9% below the median it says it targets. That gap, not individual cases, is what the budget conversation should start from.

Compa-ratio answers “how far from the midpoint”; range penetration answers “how far through the range”, which is more useful once ranges have different widths. Use both, and look at results by gender, level and team before deciding on adjustments.

Step 5: Turn Results Into Pay Ranges

The market reference becomes the midpoint of the range for that level. Then set the width:

• Range spread: narrower for junior and operational roles (around 20–35%), wider for senior and specialist roles (around 40–60%), where experience and impact vary more.
• Minimum and maximum: with a midpoint of €42,500 and a spread of 50%, the range is €34,000 to €51,000 (80% and 120% of the midpoint).
• Overlap between levels: enough to reward strong performers without promotion, not so much that a lower level can out-earn the level above.

In the example, employee C at €33,000 sits below the €34,000 minimum and should be brought into the range, while employees A and B are inside it. Our guide on how to create salary bands covers spreads, overlap and grade structures in depth, and the guide to pay compression explains what to do when new hires land above long-serving staff.

If you hire in several cities or countries, decide whether ranges vary by location before you publish them. See location-based pay, pay zones and single rates.

Step 6: Set Governance and a Refresh Cadence

A benchmark that nobody owns goes stale. Agree who does what and when:

Define off-cycle triggers in advance: repeated offer rejections over pay, regretted leavers who cite pay, or market data moving beyond an agreed threshold. Our article on real-time pay benchmarking and the annual salary review covers this hybrid model, and the salary review playbook shows how benchmarking feeds the review cycle.

Salary Benchmarking Under the EU Pay Transparency Directive

The EU Pay Transparency Directive (Directive (EU) 2023/970) had to be transposed into national law by 7 June 2026, and its obligations apply through each country’s national rules once transposed. It does not require benchmarking as such, but it raises the bar for how pay is set and explained:

• Objective, gender-neutral criteria. Pay structures must allow employers to assess whether workers are in a comparable situation regarding the value of their work, using criteria such as skills, effort, responsibility and working conditions. Your job architecture and match log are the backbone of that.
• Pay ranges for candidates. Applicants are entitled to information on the initial pay or its range, based on objective, gender-neutral criteria, before the interview or otherwise before the employment relationship begins.
• Pay criteria and information rights. Employers must make available the criteria used to set pay, levels and progression (member states may exempt employers with fewer than 50 workers from the progression part), and workers can request average pay levels by sex for their category of work.
• Reporting. Employers with 250 or more workers report the gender pay gap yearly from 7 June 2027; those with 150–249 every three years from 7 June 2027; those with 100–149 every three years from 7 June 2031. If an unexplained gap of 5% or more is not remedied within six months, a joint pay assessment is required.
• Burden of proof. In equal pay disputes, the burden of proof shifts to the employer.

In practice, document every step: scope and market position, the match for each role with its quality rating, sources and their dates, aging rate, weights, the resulting ranges and who approved them. Market data explains where your ranges sit, but differences between people doing the same work or work of equal value still need objective, gender-neutral reasons. Our guides on whether market data can justify pay differences and on running a pay equity audit go further, and you can check your ranges with our free salary benchmarking tool for pay transparency.

Sources

• TalentUp job postings database: about 3,700 employee job ads with a stated annual salary range in the United Kingdom, Spain and France, published 1 January–30 September 2026 (see Methodology above).
• EUR-Lex: Directive (EU) 2023/970 on pay transparency
• Eurostat: Labour costs
• Office for National Statistics: Average weekly earnings in Great Britain