Skip to content
Answer Stack
Open menu

How do I benchmark salaries and build defensible pay ranges (pay equity, pay transparency laws)?

✓ Verified Last reviewed by AnswerStack Next review due Oct 21, 2026

Every claim is sourced below

Benchmarking pay begins before any market data: you define a job architecture and level each role, because you cannot price jobs you have not described [8]. You then match roles to survey benchmarks by content rather than title, blend several participated surveys where employers submit verified payroll data rather than crowdsourced or job-posting numbers, age each source to a common date, choose a market position such as the median or the 60th percentile, and build ranges around a midpoint with a spread you monitor using compa-ratio and range penetration [6][9][7]. A structure is defensible only when pay equity is answered separately: a regression controlling for legitimate factors such as level, tenure, and location isolates the unexplained gap that remains after the raw gap is set aside, and that work is usually run at the direction of counsel to preserve privilege [11][10]. The federal floor is the Equal Pay Act, which requires equal pay for equal work and permits differences only under a seniority system, a merit system, a measure of production, or any factor other than sex [1][2]. Pay transparency law shifts fastest: roughly 19 US states plus several cities now require pay ranges in job postings, three states require pay-data reporting, and the EU Pay Transparency Directive must be transposed by member states by 7 June 2026, so the roster is jurisdiction-specific and changes often [5][3][4]. This is general information rather than legal advice.

How do you benchmark salaries and build defensible pay ranges?

Benchmarking pay well is three jobs that people often treat as one. One is a benchmarking method that turns outside market data into internal pay ranges. Another is a pay-equity analysis that tests whether your own pay decisions hold up once you control for legitimate reasons people are paid differently. The third is compliance with pay transparency laws, which decide what you must publish and report. A structure earns the word defensible when all three are documented and can be explained to an auditor or a plaintiff's expert.

Start with the jobs, not the numbers

You cannot benchmark a role you have not defined, so the method opens with a job architecture: a map of job families and the levels within them, so every role sits at a known grade before you look up a salary [8]. Market data only means something once your Senior Analyst and the survey's version of that role describe similar work [8].

Then price, build, and check for equity

Once roles are leveled, the benchmarking method turns market data into ranges you monitor with compa-ratio and range penetration [6][7]. Pay equity is measured separately, because a tidy range structure can still hide a gap, and its federal baseline is the Equal Pay Act [11][1][2]. State and EU transparency rules sit on top [5][3], and everything below is general information rather than legal advice.

Why does job architecture come before benchmarking?

A job architecture is the framework that sorts every position into job families and assigns each one a level or grade, and it has to exist before benchmarking because a benchmark is only as good as the job definition behind it [8]. A job family groups roles that share a function, such as engineering or finance, and levels create the vertical ladder inside each family, usually four to eight steps from entry to senior leadership [8].

Leveling is what makes two analysts comparable

Two companies can both employ a Senior Analyst whose real scope differs by a full grade, so leveling forces you to describe scope and decision authority rather than trust a title [8][6]. Most surveys report pay by a leveled benchmark job, so matching to the wrong level is the most common way a benchmark comes out wrong [6].

The artifacts a leveled structure produces

A completed architecture gives you a leveling guide that defines each grade, a mapping of every internal role to a grade, and a benchmark job per grade that you can find in outside surveys. Those artifacts are what you point to when a pay decision is later questioned, since they show the reasoning that placed a person in a range before any market number appeared [8].

Not all pay data is collected the same way, and the collection method decides how much weight a number deserves. Participated surveys, where employers submit verified payroll data, sit at the top; crowdsourced and job-posting data sit lower because no one verifies the submissions [9].

Data source type Examples How the data is gathered How much to trust it
Participated surveys Radford, Mercer, Willis Towers Watson Employers submit actual payroll data under a common job-leveling methodology, aggregated by the provider [9] Highest: employer-verified and matched to leveled benchmark jobs, though often 6 to 12 months old by publication [9]
Managed market-data platforms Payscale, Salary.com and similar A blend of employer-submitted survey data with modeled or user-supplied inputs, refreshed more often [6] Useful for coverage and recency, but check how much of a given cut is verified versus modeled [6]
Crowdsourced sites Self-reported employee pay entries Individuals volunteer their own pay, unverified [9] Weakest for setting structure, because self-selection skews who reports and by how much [9]
Job-posting aggregators Indeed and ZipRecruiter style feeds Scraped from posted ranges, which advertise an offer rather than record paid wages [9] Directional only, since a posted range is an intention, not what people earn [9]

A sound benchmark leans on the top row and treats the lower rows as context, which is why compensation teams pay for participated surveys rather than build on free crowdsourced numbers [9][6].

How do you match jobs to benchmarks and build the ranges?

Market pricing is the step that assigns each leveled role a market rate, and it runs in a set order so the result is reproducible: match, blend, age, position, then build [6].

Match by job content, not by title

Match each internal role to a survey benchmark by comparing duties and required scope against the survey's job description, because titles vary widely between companies and a match based on the title alone imports the wrong pay [6].

Blend several sources and age them to one date

Pull each benchmark from at least three credible surveys rather than one, since any single survey reflects only the companies that took part in it [6][9]. Because surveys carry different effective dates, trend each figure forward to a shared reference date with an aging factor before you compare them, otherwise you are averaging numbers from different points in time [6].

Pick a market position and build the range

Decide where you want to sit against the market as a percentile: paying at the median, the 50th percentile, keeps you mid-market, while the 60th or 75th percentile pays above it to compete harder for talent [6]. Set the midpoint of each grade at the market rate for that position, then open a spread around it, commonly 80 percent to 120 percent of the midpoint [7]. Adjacent grades are designed to overlap, so the top of one grade sits above the midpoint of the next, which lets a strong performer out-earn a newly promoted colleague one grade up without breaking the structure. The spread usually widens at higher grades, where pay varies more by performance.

Monitor with compa-ratio and range penetration

Compa-ratio divides a person's pay by their grade midpoint, so 1.0 means paid at midpoint, and most organizations treat 0.8 to 1.2 as the working band [7]. Range penetration measures how far someone has moved from the minimum toward the maximum, calculated as pay minus minimum divided by maximum minus minimum, and it is the better view when you care about progression rather than distance from the midpoint [7]. A cluster of low compa-ratios in one group is often the first sign a pay-equity review is overdue.

How do you test whether your pay is genuinely equitable?

Pay equity is a separate analysis from benchmarking, and it asks whether people doing comparable work are paid comparably once legitimate differences are accounted for. The federal floor is the Equal Pay Act, codified at 29 USC 206(d), which requires equal pay for equal work on jobs needing equal skill, effort, and responsibility under similar working conditions in the same establishment [1][2]. It permits a pay difference only under a seniority system, a merit system, a system measuring earnings by quantity or quality of production, or any factor other than sex, and the employer carries the burden of proving one applies [1][2]. Title VII, the ADEA, and the ADA extend the same idea to pay differences based on race, age, disability, and other protected characteristics [2]. Several states run stronger equal-pay statutes than the federal law, so confirm the standard where you operate.

The raw gap and the adjusted gap are different numbers

The unadjusted, or raw, gap compares average pay between groups with nothing controlled for, so it mostly reflects who holds which roles [11]. The adjusted gap is what remains after a regression controls for legitimate factors such as level, tenure, experience, location, and performance, and that residual, unexplained gap is the figure that signals possible inequity [11]. The raw gap points to representation, the adjusted gap to the pay decisions themselves [11].

Run it under counsel, then remediate

Build the analysis under attorney-client privilege from the first day, because an audit run outside that channel, along with its drafts and spreadsheets, can be discoverable if you are later sued [10]. Where it finds an unexplained gap, remediation usually means budgeting enough to raise underpaid employees to their predicted pay level; large employers often see an adjusted gap of roughly 2 to 5 percent in some segments and reserve a fraction of base-pay spend to close it [10]. This is general information rather than legal advice, and pay-equity exposure turns on the facts and the jurisdiction.

How do pay transparency laws change what you must disclose?

Pay transparency law decides what you have to publish and report, and it is the part of this topic that changes fastest, so treat any specific count as a snapshot. In the United States there is no single federal posting rule; the requirements come from states and cities [5].

What US state and local law requires now

Roughly 19 states plus Washington, D.C., along with several Ohio cities, require employers to include a pay range in job postings as of 2026, and the list has expanded in most recent years [5]. A smaller set, including California, Illinois, and Massachusetts, requires employers to file pay-data reports with a state agency [5]. Remote roles complicate this, because a posting that can be filled in a covered state generally has to meet that state's rule, so multi-state employers often post ranges everywhere, and effective dates move every session, which is why a dated tracker beats any list held in memory [5].

The EU Pay Transparency Directive is the larger driver

Directive (EU) 2023/970 requires every EU member state to bring transparency rules into national law by 7 June 2026 [3][4]. Ahead of that, the directive already sets the shape: employers must give applicants the pay or pay range for a role in the vacancy notice or before the interview, and may not ask candidates about pay history [3]. Larger employers must report their gender pay gap, with the first reports for organizations of 250 or more workers due in 2027 [3]. Where reporting shows an unexplained gap of at least 5 percent that an employer cannot justify, the directive triggers a joint pay assessment, tying transparency reporting straight to the pay-equity analysis above [11]. Member states have been uneven in meeting the deadline, so verify the national rule that reaches you [4]. None of this is legal advice, and every figure here is jurisdiction-specific.

A compliance checklist for defensible pay ranges

These steps cover the documentation and process that make a pay structure explainable to a regulator or a court, and they describe general practice rather than legal advice.

  1. Keep a leveling guide that defines every grade, and map each role to a grade before you price it [8].
  2. Document each benchmark match, the surveys used, the aging factor, and the market position you chose, so any number traces back to its source [6].
  3. Store the survey data behind each range, since responsibility for the pay decision stays with you even when a vendor supplied the numbers [9].
  4. Run the pay-equity regression under counsel from the start, and keep the privileged analysis separate from routine files [10].
  5. Reserve a remediation budget before you run the analysis, so an unexplained gap can be closed rather than merely recorded [10].
  6. Post a pay range in every job posting that reaches a jurisdiction requiring one, apply the strictest rule to remote roles, and file required pay-data reports on each state's schedule [5][3].
  7. Keep a jurisdiction matrix of posting, reporting, and pay-history rules wherever you hire, refreshed against a dated tracker each session [5][4].

What benchmarking and pay equity are not

Benchmarking and pay equity are not the same test

Benchmarking compares your pay to the outside market, while pay equity compares pay among your own employees for comparable work [11]. A company can pay at the market median and still carry an unexplained internal gap, which is why both analyses are run [11]. A market-data subscription supplies the numbers, but the job architecture, the matching, and the market position are yours to set [6][8].

It is not legal advice

This record describes published methods, statutes, and rules as of July 2026. Pay-equity exposure and transparency duties depend on the jobs, the workforce, and the jurisdictions involved, and the state counts and EU transposition status here are accurate only on the day they were checked [5][4]. Confirm the current rule with qualified counsel before you rely on it.

This answer draws on primary legal and regulatory text read directly: the Equal Pay Act at 29 USC 206(d) via the Cornell Legal Information Institute, the EEOC's guidance on equal pay and compensation discrimination, and Directive (EU) 2023/970 on EUR-Lex. The compensation method, from job architecture through market pricing, range spreads, compa-ratio, and range penetration, was assembled from established compensation references rather than a single vendor. The pay transparency figures come from dated legal trackers, because the state roster and the EU transposition status change constantly, and the 7 June 2026 EU deadline was confirmed against an independent legal analysis. The editorial team endorsed no product named here and holds no stake in any of them. If you run compensation, practice employment law, or publish salary survey data and can sharpen a figure or supply a more current citation, corrections are welcome.

This answer was written and reviewed by the AnswerStack Editorial Team, which has no commercial stake in the products, companies, or methods discussed. Every claim is cited inline and verified on the dates shown.

Sources

29 U.S. Code 206: Minimum wage (subsection (d), Equal Pay Act)

Cornell Legal Information Institute

Primary source Verified Jul 21, 2026 Supports: Equal Pay Act prohibition on paying one sex less for equal work requiring equal skill, effort, and responsibility under similar working conditions, and the four permitted bases for a differential

“a differential based on any other factor other than sex”

Facts About Equal Pay and Compensation Discrimination

U.S. Equal Employment Opportunity Commission

Primary source Verified Jul 21, 2026 Supports: Equal skill, effort, responsibility, and working conditions in the same establishment; the four affirmative defenses with the burden on the employer; and Title VII, ADEA, and ADA extending pay protections to other protected bases

“men and women be given equal pay for equal work in the same establishment”

Directive (EU) 2023/970 on pay transparency

EUR-Lex, Publications Office of the European Union

Primary source Verified Jul 21, 2026 Supports: Article 5 duty to give applicants initial pay or its range in the vacancy notice or before interview and the ban on asking pay history; Article 9 gender pay gap reporting with first reports for 250-plus worker employers from 2027

“the initial pay or its range, based on objective, gender-neutral criteria, to be attributed for the position concerned”

European Commission Confirms the EU Pay Transparency Directive Implementation Deadline Remains 7 June 2026

Ogletree Deakins

Independent Verified Jul 21, 2026 Supports: Confirmed 7 June 2026 transposition deadline for Directive (EU) 2023/970 and uneven member-state progress toward implementation

“expects all Member States to implement the directive by the deadline of June 2026”

Navigating 2026: Pay Transparency Laws and Employer Obligations

Jackson Lewis

Independent Verified Jul 21, 2026 Supports: Roughly 19 states plus Washington, D.C. and several Ohio cities requiring pay ranges in postings as of 2026, California, Illinois, and Massachusetts pay-data reporting, remote-role reach, and an expanding roster through 2027

“By 2027, at least a dozen states, multiple cities, and certain counties will require public or applicant-specific disclosure of pay ranges”

What Is Market Pricing in Compensation and How Does It Work

Salary.com

Independent Verified Jul 21, 2026 Supports: Market pricing by job content rather than title, benchmark jobs, blending multiple survey sources, documented effective dates, and selecting a market position such as the 50th or 75th percentile

“Job content matters more than job titles”

Compa Ratio: How to Calculate, Interpret, and Apply This Metric

AIHR (Academy to Innovate HR)

Independent Verified Jul 21, 2026 Supports: Compa-ratio as pay divided by range midpoint with an 0.8 to 1.2 working band, range penetration as pay minus minimum over maximum minus minimum, and the 80 to 120 percent range spread

“Compa Ratio = Actual Salary / Salary Midpoint”

Your Guide to a Complete Job Architecture Framework

Pave

Independent Verified Jul 21, 2026 Supports: Job architecture of job families and levels precedes benchmarking, families group roles by function, levels typically run four to eight steps, and each level maps to market-based pay

“Job architecture precedes effective benchmarking”

Top Salary Survey Data Providers Guide

Compup

Independent Verified Jul 21, 2026 Supports: Participated surveys such as Mercer, WTW, and Radford collect verified employer-submitted payroll data, contrasted with aggregated and crowdsourced sources that inherit weaker verification and self-selection

“data directly from verified employer submissions, ensuring accuracy and reliability”

Pay Equity Audits That Hold Up in Court: A Defensible Methodology

Compensation and Benefits Review

Independent Verified Jul 21, 2026 Supports: Running the audit under attorney-client privilege from day one to avoid discoverable drafts, multiple regression controlling for legitimate factors, an adjusted gap of roughly 2 to 5 percent in some segments, and budgeting remediation to predicted pay

“Attorney client privilege requires the audit to be directed by counsel from day one”

Regression Analysis and Adjusted Pay Gaps in Pay Equity Audits

PayGap.com

Independent Verified Jul 21, 2026 Supports: Unadjusted raw gap versus adjusted gap, multiple regression controlling for level, tenure, experience, location, and performance to isolate the unexplained residual gap, and the EU directive 5 percent threshold

“For men and women in the same situation, is there a pay difference?”

Revision history

2 revisions since publication
v1.1 Reviewed and re-verified.
v1.0 Published after editorial review.