How to Conduct a Pay Equity Analysis: 5 Steps and a Free Checklist

KEY TAKEAWAY

A pay equity analysis compares compensation for employees doing similar work, controlling for legitimate pay factors like tenure, performance, and location, to identify unexplained pay gaps by gender, race, or other protected characteristics. 5 steps: collect your data, define comparable work, control for legitimate variables, measure the gap, fix what you find and document it.

Pay equity is not the same as equal pay. Equal pay means paying the same rate for the exact same job. Pay equity is more nuanced: it controls for legitimate factors that justify pay differences (seniority, performance ratings, scope of role) and then asks whether any remaining gap is explained by protected characteristics like gender or race.

The distinction matters because most pay gaps in modern organizations are not the result of explicit discrimination. They accumulate over time through biases in starting salary negotiations, smaller percentage increases for historically underrepresented groups, and differential access to high-visibility roles that come with compensation premiums.

A pay equity analysis surfaces these gaps before they grow into legal exposure, retention problems, or reputational damage when they become public.

What Is Pay Equity?

Pay Equity Definition

Pay equity means that employees performing comparable work receive comparable compensation, regardless of gender, race, age, ethnicity, disability status, or other protected characteristics. It is distinct from the raw pay gap (which compares average pay for all men versus all women without controlling for role or seniority) and from pay equality (which compares pay for identical roles). Pay equity controls for legitimate variables to isolate unexplained compensation differences.

Pay Equity vs Pay Equality vs the Raw Pay Gap: The Differences

The controlled pay gap (pay equity) is the most legally significant metric because it isolates compensation differences that cannot be explained by legitimate business factors. A 3 percent unexplained gap is a compliance and legal risk signal. A 15 percent raw gap may simply reflect occupational segregation (different roles being dominated by different demographic groups) rather than pay discrimination within roles.

Both metrics are worth tracking. The controlled gap drives legal exposure. The raw gap drives broader workforce strategy about representation and role access.

Step 1: Collect Your Compensation Data

A pay equity analysis requires a comprehensive compensation dataset. At minimum, you need the following fields for every employee in scope:

• Employee ID (anonymized for analysis; linked to protected characteristic data separately)
• Current base salary or hourly rate
• Total cash compensation (base plus variable pay, bonus, commission)
• Job title, job family, and job level within your job architecture
• Department and business unit
• Location (city, state, country)
• Tenure in current role and total company tenure
• Most recent performance rating (from the last completed review cycle)
• Employment type (full-time, part-time)
• Gender, race/ethnicity, and any other protected characteristics you are analyzing (handled with strict privacy controls)

Data privacy note

Protected characteristic data must be handled under strict privacy controls and separated from performance or compensation analysis workflows. In most organizations, HR maintains a separate dataset linking employee IDs to protected characteristic data that is used only for equity analysis and never stored in the core compensation system. Consult your legal and privacy teams on data handling requirements before collecting or processing this data.

Step 2: Define Comparable Work

Defining comparable work is the most technically complex step in a pay equity analysis. You cannot compare pay across roles that require fundamentally different skills, experience, or accountability. The analysis requires grouping employees into 'pay equity cohorts': groups of employees whose work is sufficiently similar to make compensation comparisons meaningful.

Two approaches to defining comparable work:

Job architecture-based grouping

If your organization has a documented job architecture with job families and levels, you can use it directly as the basis for pay equity cohorts. All Software Engineers at Level 3 within the Engineering job family form one cohort. All Senior HR Business Partners form another. This approach is clean and defensible because the groupings are already established in your compensation structure.

Statistical regression grouping

For organizations without a documented job architecture, a statistical regression analysis groups employees based on the objective characteristics of their roles (scope, skills, accountability) rather than job titles, which can vary inconsistently across the organization. This approach requires more statistical expertise but produces more rigorous results when job titling is inconsistent.

Common mistake

Defining cohorts too broadly. Grouping all 'managers' together regardless of function, team size, or level produces a noisy analysis because the cohort contains roles with legitimately different pay ranges. The more precisely your cohorts reflect comparable work, the more meaningful your gap findings will be.

Step 3: Control for Legitimate Pay Factors

Within each comparable work cohort, some pay differences are legitimate. The equity analysis must control for these factors before calculating the unexplained gap:

After controlling for these legitimate factors, any remaining pay difference correlated with a protected characteristic (gender, race) is the unexplained gap that requires remediation. An unexplained gap of 1 to 2 percent is within the statistical noise range for most analyses. A gap of 3 percent or more is typically flagged for remediation.

Step 4: Identify and Measure the Gap

The most rigorous method for calculating the unexplained pay gap is a multiple regression analysis: a statistical technique that models pay as a function of legitimate factors and then measures whether gender or race explains any remaining pay variation after those factors are controlled.

For HR teams without in-house statistical expertise, there are three practical approaches:

1. Engage a compensation consulting firm: organizations like Mercer, Willis Towers Watson, or a specialized pay equity firm conduct the regression analysis and produce a legally defensible report. This is the highest-cost approach and the most defensible in litigation.
2. Use compensation management software with built-in pay equity analytics: CompBldr's pay equity module runs the controlled gap analysis against your compensation and performance data within the platform, removing the need for external statistical tools or manual data exports.
3. Conduct a simplified cohort comparison: group employees into defined cohorts, calculate average pay for each demographic group within the cohort controlling for tenure quartile and performance rating level, and identify cohorts where the gap exceeds your threshold (typically 3 percent or more). This is less statistically rigorous than regression but accessible for organizations beginning their pay equity journey.

Step 5: Fix the Gaps and Document Everything

Identifying pay equity gaps is the analysis. Fixing them is the compliance and retention work.

Principles for remediation:

• Address gaps through salary increases to underpaid employees, never through salary reductions to overpaid employees. Cutting pay to achieve equity creates legal exposure under prior pay agreements and destroys trust.
• Prioritize the largest gaps and the most statistically significant findings first. Not every small gap requires immediate remediation. Focus resources on gaps that are both large enough to be meaningful and statistically significant enough to be unlikely to be random variation.
• Build remediation into the annual merit cycle where possible. Many organizations address pay equity adjustments as a separate budget item within the merit cycle, funded at a dedicated equity budget rate (typically 0.5 to 1.0 percent of payroll in addition to the standard merit pool).
• Document every decision: which gaps were identified, how they were calculated, which employees received adjustments, and why. This documentation is your defense if a pay equity claim is filed.

In TraineryHCM, pay equity findings from CompBldr feed directly into the compensation planning module. When the merit cycle opens, employees flagged as having a pay equity gap appear with that context visible alongside their performance rating and compa ratio, so HR can address equity and merit in a single workflow rather than in separate processes.

Free Pay Equity Audit Checklist

Pay Equity Audit Checklist

PRE-ANALYSIS [ ] Employee compensation dataset compiled with all required fields [ ] Protected characteristic data secured under privacy controls [ ] Job architecture documented with all employees mapped to a job family and level [ ] Legitimate control variables identified and available in dataset [ ] Legal counsel briefed on analysis scope and data handling approach  ANALYSIS [ ] Comparable work cohorts defined and documented [ ] Regression analysis or cohort comparison completed [ ] Unexplained gaps identified by protected characteristic and cohort [ ] Statistical significance of gaps assessed [ ] Findings reviewed by legal counsel before internal distribution  REMEDIATION [ ] Remediation priority list created (gaps by size and significance) [ ] Remediation budget approved [ ] Adjustment amounts calculated for each affected employee [ ] Adjustments implemented in payroll [ ] Adjustments documented in compensation system [ ] Communication plan created for affected employees  ONGOING [ ] Pay equity analysis cadence established (annual recommended) [ ] Pay equity review built into annual merit cycle workflow [ ] New hire salary setting process reviewed to prevent new gaps from forming [ ] Promotion and transfer pay adjustment process reviewed for equity [ ] Pay equity training delivered to hiring managers and compensation team