A salary benchmarking peer group is the set of employers whose pay you treat as “the market” when you price a role. Choose it well and your ranges reflect what it really costs to hire and keep people; choose it badly and every number built on it, from offers to merit budgets, is wrong before you start.
Most benchmarking mistakes are not data errors. They are comparator errors: the wrong industry, the wrong size band, the wrong city, or a group of aspirational peers that pushes pay up. This guide explains how to define a compensation peer group by role, the criteria and the mistakes to avoid, and shows with TalentUp data from 2026 job ads how far the median moves when only the peer group changes.
Key takeaways
- Your peer group should reflect who you compete with for talent (labour market), not only who you compete with for customers (product market).
- Use role-specific cuts: engineers, sales people and accountants are hired from different employers and locations.
- In TalentUp’s 2026 data, the median advertised salary for US software engineers ranges from €143,300 to €191,400 depending on the peer group: a 34% swing for the same role.
- Location often moves the benchmark more than company size; size effects are not even linear outside executive roles.
- Avoid aspirational peers and “peers of peers”: research on executive pay shows firms tend to choose higher-paying peers.
- Document objective criteria and review them yearly. Where the EU Pay Transparency Directive applies through national law, that documentation supports your pay structure.
What Is a Salary Benchmarking Peer Group?
A peer group (also called a comparator group or market reference) answers one question: whose pay should ours be compared with? In practice it is a set of filters on market data: industry, company size, location, and sometimes a named list of companies.
It sits between two other benchmarking steps. First you match jobs by content, not title (see what salary benchmarking is). Then you choose the peer group. Only then do you read percentiles and set ranges. If the peer group is wrong, accurate job matching and good data still produce the wrong answer.
There are two kinds of peer groups, and they serve different purposes:
|
Labour-market peer group |
Product-market peer group |
| Question it answers |
Who do we compete with for talent? |
Who do we compete with for customers and investors? |
| Typical members |
Any employer hiring the same skills in the same locations, across industries |
Direct competitors with a similar business model |
| Main use |
Broad-based pay: engineers, sales, finance, operations |
Executive pay, board reporting, performance comparisons |
| Usual form |
Market data cuts (industry, size, location) |
Named list of companies, often 12–20 |
| Main risk |
Too broad, mixing very different employers |
Too narrow, missing the employers that actually hire your people |
A fintech hiring data engineers competes with banks, consultancies and big tech for the same candidates, even though none of them sell its product. For most roles below the executive team, the labour-market view should lead.
Criteria for Choosing Your Comparators
No single criterion defines a good peer group. Combine them, in roughly this order of importance for broad-based roles:
| Criterion |
Why it matters |
How to apply it |
| Talent competitors |
The employers you hire from and lose people to are your real market |
List the top destinations in exits and the top sources of hires, and offers lost by employer |
| Location |
Pay levels differ between cities, regions and countries |
Benchmark at the level where you recruit: city for office roles, country or region for remote roles |
| Industry |
Some sectors pay a premium for the same skills |
Use a sector cut for roles where the premium is real (often tech and finance roles), and a cross-industry cut for general roles |
| Company size |
Scope and complexity usually grow with headcount or revenue |
Use a band around your own size; for executives, commonly about 0.5–2x your revenue or headcount |
| Role family and level |
Different jobs have different labour markets |
Allow a different peer group per job family; keep levels comparable |
| Data sufficiency |
A precise but tiny cut is noisy |
Set a minimum number of data points per cell and widen the cut when you fall below it |
Two practical rules help. First, start narrow, then widen until the sample is big enough: same industry and city, then neighbouring industries or the national market. Second, keep a core peer group and add role cuts, so you can explain the logic in one page instead of one rule per job.
For a full step-by-step process, see how to implement salary benchmarking.
How Much the Peer Group Moves the Benchmark: TalentUp Data 2026
To show the effect, we took 2026 job ads with a stated salary and recomputed the median for the same roles under different peer-group definitions. Nothing changes except the comparators.
34%
gap between the broadest and the narrowest peer group for US software engineers (€143,300 vs €191,400)
+20%
London vs the rest of the UK for software engineers (€105,900 vs €88,100)
+33%
London vs the rest of the UK for account managers (€65,500 vs €49,400)
No city premium
in Spain: Madrid and Barcelona software ads have a median of €42,600 vs €44,000 elsewhere
Source: TalentUp job postings database, 1 January–30 September 2026. Median advertised annual salary in EUR, employee contracts.
Worked example: one role, five peer groups
Imagine a 400-person software company in the US pricing a software engineer role. These are five defensible peer groups and the median each one produces:
| Peer group |
Ads (n) |
Median advertised salary |
vs broadest group |
| All US employers |
14,552 |
€143,300 |
– |
| Tech-sector employers |
3,148 |
€165,600 |
+15.6% |
| Tech, 201–1,000 employees |
479 |
€169,700 |
+18.4% |
| Tech, 201–1,000 employees, excluding New York City and San Francisco |
401 |
€163,300 |
+14.0% |
| Tech, New York City and San Francisco only |
367 |
€191,400 |
+33.6% |
The same job can be priced anywhere between about €143,000 and €191,000, depending only on the comparators. If the company hires nationally and remotely, the national tech cut is defensible; if it competes for engineers in San Francisco, the city cut is the real market. Either way, the choice must be deliberate and written down.
Size, industry and city cuts by role
| Country and role |
Cut |
Ads (n) |
Median |
| US, software engineer |
Employers with 1–200 staff |
1,142 |
€138,700 |
|
Employers with 1,001–10,000 staff |
2,587 |
€162,100 |
|
Retail and consumer goods sector |
154 |
€128,100 |
| US, account manager / account executive |
All ads |
4,844 |
€102,800 |
|
Tech sector |
1,088 |
€134,400 |
|
Employers with 1–200 staff |
350 |
€85,200 |
|
Ads posted by recruitment agencies |
401 |
€70,800 |
| US, data analyst / scientist / engineer |
All ads |
5,032 |
€120,800 |
|
Finance, banking and insurance |
583 |
€140,600 |
|
Public sector, health and education |
249 |
€110,400 |
| UK, software engineer |
All ads |
3,060 |
€90,200 |
|
Employers with 201–1,000 staff |
95 |
€68,700 |
|
Employers with 1,001–10,000 staff |
129 |
€114,400 |
|
Tech sector |
135 |
€74,100 |
| Spain, software engineer |
All ads |
2,267 |
€43,500 |
|
Employers with 201–1,000 staff |
222 |
€48,600 |
|
Employers with 1,001–10,000 staff |
108 |
€50,400 |
|
Tech sector |
702 |
€41,100 |
Three lessons stand out:
- Size is not linear. In the UK and Spain, mid-sized and very large employers do not simply pay more than small ones. A rule like “bigger peers pay more” is safe for executives, much less so for engineers.
- Industry premiums depend on the role. In the US the tech sector pays a clear premium for software and sales roles; in the UK and Spain, tech-sector ads for software engineers have a lower median than the overall market, because banks, consultancies and other employers also hire engineers and some pay more.
- Who posted the ad matters. In the US, account manager ads posted by recruitment agencies have a median of €70,800, against €102,900 for ads posted directly by employers. Decide whether agency data belongs in your market and apply the rule consistently.
Location: The Peer Group Criterion Teams Underestimate
In our data, location often moves the median as much as industry or size. London ads pay 20% more for software engineers and 33% more for account managers than ads elsewhere in the UK. In the US, New York City and San Francisco ads pay about 27% more for software engineers (€176,700 vs €139,600).
But the city premium is not universal. In Spain, software ads in Madrid and Barcelona have a slightly lower median than the rest of the country (€42,600 vs €44,000), possibly because some ads outside the two cities are national or remote roles from well-paying employers. That is why you should check the data rather than assume a fixed city uplift.
Your location decision is also a pay policy decision. If you hire remotely, you must choose between benchmarking each employee’s local market, a national market, or a single reference market. We cover the trade-offs in offering the same salary regardless of location and, for multi-country teams, in salary benchmarking for international companies.
Common Peer Group Mistakes
| Mistake |
What happens |
Fix |
| Too narrow |
Five direct competitors give tiny, volatile samples and miss the employers that actually hire your people |
Widen to labour-market competitors; set a minimum sample per cell |
| Too broad |
National, all-industry data blends very different employers and underprices scarce roles |
Add sector or city cuts for roles where the premium is real |
| Aspirational peers |
Benchmarking against the companies you would like to be inflates every range |
Pick peers you actually compete with today, at a comparable size |
| “Peers of peers” |
Copying competitors’ peer lists creates a self-reinforcing upward loop |
Build the list from your own hiring and attrition data |
| One peer group for all roles |
Engineers priced against retail employers, or store staff against tech |
Keep a core group plus role-family cuts |
| Mixing executive and broad-based logic |
Revenue-matched product-market peers used for engineers or sales staff |
Use product-market peers for executives, labour-market cuts below |
| Ignoring location |
A national median applied to a city with a large premium, or the reverse |
Benchmark where you recruit; check city vs national data |
| Changing the group every year |
Ranges jump because comparators changed, not because the market did |
Keep criteria stable; document and explain any change |
The aspirational-peer problem is well documented for executives. In a study of S&P 500 and S&P MidCap 400 companies after the SEC’s 2006 disclosure rules, Faulkender and Yang found that, after controlling for industry, size and other factors, firms appeared to select highly paid peers to justify CEO pay, especially when the peer group was small or the CEO also chaired the board.
Executive Pay Peer Groups
Executive peer groups follow a different logic. Executives are usually benchmarked against a named list of companies with a similar industry, business model and scale. Practitioner guidance commonly suggests around 12 to 20 peers of roughly 0.5 to 2 times the company’s revenue or market capitalisation, reviewed annually and after major mergers or divestitures.
In the US, listed companies must disclose this in the Compensation Discussion and Analysis of their proxy statement: SEC Regulation S-K, Item 402(b)(2)(xiv), asks whether the company benchmarked total compensation or any material element and, if so, to identify the benchmark and its component companies. That makes US executive peer groups public and open to scrutiny by investors.
Keep the two worlds separate. A revenue-matched list of 15 product-market competitors is the right tool for the CEO and the board, but it is the wrong tool for pricing a junior data analyst, who is hired in a much broader labour market.
Governance: Document, Review and Explain Your Peer Group
A peer group is a policy, not a one-off analysis. Treat it like one:
- Write the criteria down. For each job family: industries included, size band, locations, data source, minimum sample and the percentile you target.
- Link it to your compensation philosophy. “We pay at the median of tech employers in our hiring cities” is a peer group statement and a pay position in one sentence (see how to build a compensation philosophy).
- Validate with real hiring data. Compare the peer group against where offers are lost and where leavers go. If those employers are missing, the group is wrong.
- Review annually. Change criteria only for a documented reason, such as a new hiring location, a shift to remote work or a major market change, and record the impact on ranges.
- Feed the result into your bands. Peer group changes should flow into the annual review of salary bands and the balance with internal equity.
What the EU Pay Transparency Directive changes
The EU Pay Transparency Directive (2023/970) does not tell employers which comparators to use. Member States had to transpose it by 7 June 2026, and its obligations apply through national laws once transposed. Where they apply, three points make a documented peer group policy more valuable:
- Pay structures must allow work of equal value to be assessed on objective, gender-neutral criteria (Article 4).
- Employers must make the criteria used to set pay and pay progression available to workers (Article 6).
- In pay discrimination cases the burden of proof shifts to the employer (Article 18).
Market rates can play a role in explaining pay differences, but they must be applied consistently. In Enderby (C-127/92), the EU Court of Justice accepted that the state of the labour market could objectively justify part of a pay difference, to the extent the employer could show it. A peer group chosen case by case, or changed to justify individual salaries, is hard to defend; one written down and applied to every role is much easier. Reporting duties start from 7 June 2027 for employers with 250+ workers (yearly) and 150–249 workers (every three years), and from 7 June 2031 for those with 100–149 workers. If a gender pay gap of 5% or more in a category of workers cannot be justified by objective, gender-neutral criteria and is not remedied within six months, a joint pay assessment is required.
Sources
- TalentUp job postings database, ads published 1 January–30 September 2026 in the United States, the United Kingdom and Spain. See TalentUp methodology.
- European Union. Directive (EU) 2023/970 on pay transparency. EUR-Lex.
- Court of Justice of the EU. Case C-127/92, Enderby v Frenchay Health Authority. EUR-Lex.
- US Code of Federal Regulations. 17 CFR 229.402, Executive compensation (Regulation S-K, Item 402). Legal Information Institute, Cornell Law School.
- Faulkender, M. and Yang, J. (2010). Inside the black box: The role and composition of compensation peer groups. Journal of Financial Economics, 96(2), 257–270.