How to Conduct a Pay Equity Audit: A Step-by-Step Guide

The Core Design Challenge

The first step is defining the audit scope. At minimum, a pay equity audit covers gender. A comprehensive audit covers gender, ethnicity, age, and disability — though the ability to analyse each characteristic depends on the quality of self-identification data available. An organisation with 40% ethnicity self-identification completion can conduct a directional ethnicity analysis with appropriate caveats but cannot produce statistically reliable conclusions for specific ethnic group comparisons. Data quality is a scope constraint, not a reason to defer the analysis entirely.

Data assembly is the most time-consuming step in the process and the one most likely to produce errors if rushed. The required data set for each employee includes: unique employee ID; grade or evaluation score; base salary; any other measured compensation components (bonus paid, equity vested); gender and any other characteristics being analysed; continuous service tenure; most recent performance rating; contract type (full-time, part-time, fixed-term); and office location. The data must come from authoritative sources — payroll for salary, HRIS for grade and tenure, performance management system for ratings — and must be reconciled across sources before any analysis is run. Missing data must be investigated before the analysis begins: systematic missing data (e.g. performance ratings missing more often for women than men) is itself an equity finding that affects the analytical methodology.

Two methodologies are available. Cohort analysis compares the average or median pay of a protected group against the comparison group within defined cohorts — typically grade and role type. It is explainable, transparent, and appropriate for organisations with insufficient volume for regression. Regression analysis uses a statistical model to control for multiple legitimate pay factors simultaneously, isolating the unexplained pay difference correlated with the protected characteristic. It requires a minimum of approximately 100 employees in the full analysis group and 30 or more in the protected characteristic group to produce statistically reliable results. Organisations below these thresholds should use cohort analysis with explicit disclosure of the sample size limitations.

How the Approach Works

Findings must be categorised by type to determine the appropriate remediation response. A point-of-hire finding — employees in comparable roles hired at different salaries, with the gap correlated with a protected characteristic — is a legal liability that requires salary correction and likely legal advice. A merit allocation finding — employees with comparable performance ratings receiving systematically lower merit increases, correlated with a protected characteristic — requires process intervention in the merit allocation system. A representation finding — the unadjusted pay gap is large primarily because of the concentration of protected groups in lower-grade roles — requires a structural response (pipeline, promotion equity, succession planning) rather than pay corrections alone.

The communication of findings requires careful management of two distinct audiences. For the board and senior leadership: a clear statement of findings with financial impact estimates and proposed remediation. For affected employees: individual conversations about salary corrections, with language carefully reviewed by legal counsel before delivery to avoid unintentional admission of liability. For the broader workforce: a statement of commitment to pay equity and the process being conducted, without specific findings that could cause concern before corrections are in place. The sequence — investigate, correct, then communicate — is the most defensible approach in most jurisdictions.