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

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

A pay equity audit is the most direct way to find out whether your organization pays people fairly for comparable work — and to fix it before a regulator, a plaintiff's attorney, or a candidate with a pay-range screenshot does it for you. You already know the headline numbers. In 2024, women working full-time, year-round earned just 80.9 cents for every dollar earned by men, and the gap actually widened for the second straight year. But aggregate statistics don't tell you what's happening inside your pay structure. An audit does. This guide walks you through the audit end to end: how to scope it, what data you need, how to run the analysis, how to read the results without panicking, and how to remediate in a way that holds up. You'll also learn where audits go wrong, and how to keep the work defensible if it ever surfaces in litigation.

TL;DR — Key takeaways

• A pay equity audit compares pay across protected groups within comparable-work groupings, then isolates gaps that legitimate factors can't explain.
• The defensible method is multiple regression: control for job, experience, performance, and location, then measure the residual gap tied to gender, race, or other protected status.
• Scope the audit with counsel first. Running it under attorney-client privilege protects your findings if you're ever subpoenaed.
• Clean job groupings make or break the analysis. Garbage groupings produce garbage gaps — which is why a sound job evaluation foundation matters.
• Budget 1–3% of payroll for first-year remediation, and never fix a gap by cutting anyone's pay.

What a pay equity audit actually measures

A pay equity audit measures whether employees doing substantially similar work are paid equitably regardless of protected characteristics like sex, race, or age. That's a narrower question than the raw wage gap. The 80.9-cent figure mixes together every job in the economy; it reflects occupational segregation, hours, and career interruptions as much as discrimination. Your audit strips those out and asks a sharper question: among people in the same role, with similar experience and performance, does protected-class status still predict pay?

This is also where pay equity and internal equity diverge. Internal equity is about whether your pay structure ranks jobs fairly relative to each other — that's a job evaluation problem. Pay equity is the regulatory question layered on top: within those fairly-ranked jobs, are individuals paid without bias? You need the first to do the second well, because you can't compare "similar work" if you've never defined what similar work means.

The legal backbone is the Equal Pay Act, which requires equal pay for equal work and covers every form of compensation — salary, bonuses, stock, overtime, profit sharing, and benefits. Pay differences are legal only when they trace to seniority, merit, quantity or quality of production, or a bona fide factor other than sex — and the burden is on you, the employer, to prove the factor applies. The EEOC enforces these rules alongside Title VII, the ADEA, and the ADA. An audit is how you check your own homework before they do.

Step 1 — Scope the audit (and loop in counsel first)

Start by deciding what you're auditing and who's directing it. Decide the protected dimensions (gender is standard; add race/ethnicity where you have reliable data), the population (one country, one legal entity, or the whole company), and the snapshot date for pay data.

Crucially, decide now whether the audit runs under attorney-client privilege. If outside or in-house counsel defines the purpose and scope, directs the data collection, and receives the findings as legal advice, your analysis is far better protected if it's ever subpoenaed. Privilege isn't automatic — it hinges on a deliberate process: counsel sets scope, data is gathered under their direction, the analysis is reproducible, results go into a legal memo, and only a sanitized summary circulates broadly. Make that call before you pull a single row of data. You cannot retroactively wrap privilege around an analysis you already emailed to the whole comp team.

Step 2 — Build clean comparison groups

This is the step that quietly determines whether your audit is worth anything. You can't compare pay across "similar work" until you've defined what counts as similar. Most organizations group employees into pay analysis groups (PAGs) — clusters of roles similar enough in content and level that you'd expect comparable pay.

If your job architecture is messy — inflated titles, overlapping levels, roles that mean different things in different departments — your groupings will be wrong, and every gap you "find" will be noise. A consistent, quantitative foundation fixes this. The point-factor method scores each job against weighted compensable factors — skill, effort, responsibility, and working conditions — so that "similar work" is defined by measured job content, not by title bingo. Jobs with similar point totals belong in the same analysis group. That gives your regression honest inputs.

Step 3 — Gather and clean the data

Pull the variables your model needs:

• Base pay, plus total cash (bonus, commission, equity if relevant) — the EPA covers all of it.
• Protected attributes: gender, and race/ethnicity where available.
• Legitimate explanatory factors: job/grade, time in role, total experience, performance rating, location, and full-time/part-time status.

Then clean it. Standardize pay to a common basis (annualized base, FTE-adjusted). Fix duplicate records and missing ratings. Document every assumption — if this analysis is ever questioned, "we excluded interns and contractors" needs to be a written, consistent rule, not a judgment call you make twice.

Before you run a single regression, sanity-check your groupings the way an auditor would. If you can't explain in one sentence why two roles sit in the same comparison group, your model can't either. PointFactors builds those defensible groupings automatically from scored job content — see how it works in a short demo.

Step 4 — Run the analysis

For each comparison group large enough to support it, run a multiple regression. The model predicts pay from your legitimate factors — job, experience, performance, location — and then tests whether protected-class status still explains a statistically significant slice of what's left over. That residual is the part of the pay gap your legitimate factors can't account for, and it's the number that matters.

This is the gold standard precisely because it simulates a fair comparison: two employees identical on every legitimate dimension except gender or race. If the model says women in a group are paid 4% less after controlling for everything legitimate, that 4% is your unexplained gap. For small groups where regression isn't statistically valid, fall back on structured cohort comparisons — matched pairs of similar employees — and treat the results as directional rather than definitive.

A quick read of what regression output tells you:

Step 5 — Investigate before you remediate

A statistically significant gap is a flag, not a verdict. Before adjusting anyone's pay, investigate what's driving it. Sometimes it's a real, legitimate factor your model didn't capture — a specialized certification, a retention package, a market premium for a hot skill. Capture that factor, document it, and re-run. If a gap survives a good-faith hunt for legitimate explanations, you're looking at a disparity you need to fix.

Step 6 — Remediate without creating new problems

Two hard rules. First, you can never close a gap by reducing anyone's pay — the EPA prohibits it, and morale won't forgive it. You fix gaps by raising the underpaid. Second, budget realistically: organizations typically spend 1–3% of total payroll on first-year remediation. For a company with $50 million in payroll, that's roughly $500,000 to $1.5 million. Knowing that number before you present to the CFO turns a surprise into a plan.

Sequence the fixes by severity, get sign-off, and make the adjustments in a single coordinated cycle where possible so the correction reads as a deliberate equity action, not a series of one-off favors.

Step 7 — Make it a recurring discipline

A one-time audit is a snapshot that's stale the moment you hire, promote, or run a merit cycle. Treat pay equity as an operating discipline: re-run the analysis at least annually, ideally before and after each comp cycle, so new gaps get caught while they're small. Pair the audit with disciplined pay-range governance and the pay-transparency obligations now spreading across states — our 2026 multi-state pay transparency guide covers where posting and reporting rules now apply.

Common mistakes that sink an audit

The audits that fail tend to fail the same ways. They compare by job title instead of job content, so the groupings are fiction. They skip privilege and then can't protect the findings. They control for a factor that is itself tainted by bias — like prior salary — and launder discrimination into a "legitimate" explanation. They find gaps and stall on remediation because no one budgeted for it. And they run once, declare victory, and let new gaps quietly reopen. Avoid those five and you're ahead of most organizations.

FAQ

What is a pay equity audit? It's a structured analysis comparing compensation across protected groups — gender, race, age — within groups of employees doing comparable work, to identify pay gaps that legitimate factors like experience, performance, and location can't explain.

How is a pay equity audit different from the gender pay gap? The headline gender pay gap (women earning about 81 cents per dollar) mixes all jobs together and reflects occupational differences as much as bias. An audit controls for legitimate factors and isolates the unexplained gap within comparable roles inside your own organization.

How often should we run a pay equity audit? At least annually, and ideally around each compensation cycle. Pay equity drifts every time you hire, promote, or adjust pay, so a recurring cadence catches new gaps while they're still small and cheap to fix.

Should we run the audit under attorney-client privilege? For most employers, yes. When counsel directs the scope and analysis and receives the findings as legal advice, your results are far better protected if they're ever subpoenaed. Decide this before you collect data — privilege can't be added afterward.

How much does remediation cost? Plan for 1–3% of total payroll in the first year. A $50 million payroll organization typically budgets $500,000 to $1.5 million for initial corrections. Costs usually fall in later cycles once the structure is clean.

Can we lower someone's pay to close a gap? No. The Equal Pay Act prohibits reducing any employee's wages to equalize pay. You close gaps by raising the underpaid employees, never by cutting others.

Do we need a job evaluation before a pay equity audit? Functionally, yes. You can't compare "similar work" without a consistent definition of job content. A point-factor job evaluation produces the defensible comparison groups your audit depends on.

Run audits on a foundation that holds up

A pay equity audit is only as honest as the job groupings underneath it. PointFactors uses AI-assisted point-factor evaluation to score every role against consistent compensable factors — giving you comparison groups that stand up to a regulator, a plaintiff's expert, and your own board. Book a demo to see how it works, or explore plans and pricing to get started.

Justin Hampton is the founder and CEO of PointFactors, where he helps HR and compensation leaders bring rigor, defensibility, and speed to job evaluation and pay equity.