Pay equity audits: finding and closing gaps with benchmark data

A company runs a pay equity review and finds that two people with identical titles, similar tenure, and equivalent performance ratings are earning $18,000 apart. Neither manager flagged it. The pay structure wasn't designed unfairly — it drifted there through three years of ad-hoc offers, a couple of counter-offers, and one promotion cycle where the budget ran short. The gap is real. So is the awkwardness when someone finds out.

Pay equity problems rarely start with bad intent. They accumulate through ordinary decisions made without a common reference point. A pay equity audit is how you find those gaps before they find you — and, crucially, how you fix them in a way that's defensible, funded, and sustainable.

What pay equity actually means in practice

Pay equity has a legal definition in most jurisdictions (equal pay for equal or equivalent work, without discrimination on protected characteristics). But for most employers running an internal audit, the practical scope is broader: are people doing comparable work paid comparably, relative to the market and relative to each other?

That involves two distinct checks:

• Internal equity — are people in the same role/level paid consistently with each other, adjusting for legitimate differentiators like tenure, geographic cost-of-living, and documented performance?
• External equity — is the pay for each role competitive with what the market actually pays for equivalent skills, scope, and seniority?

The two are related but not the same. You can be internally consistent and still underpay every engineer relative to market. You can be externally competitive and still have significant internal gaps by gender or ethnicity. A complete audit checks both.

Why benchmarks are the foundation

Without an external reference, internal equity analysis can only tell you whether gaps exist — not whether they're problems. "Engineer A earns $110K and Engineer B earns $92K" is a data point. To know whether either number is defensible, you need to know what the market pays for that role.

Market benchmark data establishes the neutral baseline. It answers: for a role at this scope, seniority, and industry, what does the 25th, 50th, and 75th percentile look like? Once you have that, you can:

1. Classify whether a salary is below market (a fairness and retention risk)
2. Determine whether a gap between two employees is explained by one being above market, one being below, or something else
3. Prioritise which gaps to fix first — those that sit below the market floor should move ahead of gaps where both employees are reasonably placed

The quality of your benchmark data matters a lot. A salary survey that's two years old, covers only one geography, or doesn't distinguish by level will produce baselines that look authoritative but mislead your decisions. EvenBetter's methodology triangulates across multiple live data sources — job postings, salary surveys, open-web data, and our own dataset — weighted by recency, so the baseline reflects what the market pays today.

Running the audit: a step-by-step framework

Step 1: Group roles by comparable work

Start by organising your workforce into job families (Engineering, Sales, Operations, Finance, etc.) and within each family, define levels (IC1/IC2/IC3 or Junior/Mid/Senior, whatever your levelling system uses). The goal is to create groups where people are genuinely doing comparable work.

Watch for two common errors:

• Title inflation — two people with the same title who are actually operating at different scopes or levels. Handle this before pulling pay numbers or the gap analysis will be noise.
• Missing levels — if your levelling hasn't been formally maintained, some roles may be ambiguously placed. Resolve ambiguity now, before the audit, not after you find a gap you need to explain.

Step 2: Pull benchmark data for each group

For each job family and level combination, pull a market benchmark. You're looking for the P25, P50, and P75 of base salary (and total cash if you have reliable variable comp data) for your reference market — industry, company size, geography.

A few practical notes:

• Use the same benchmark source consistently. Mixing sources with different methodologies makes the results incomparable.
• If your workforce spans multiple geographies, pull location-adjusted benchmarks. Remote roles are a judgment call — many companies use the employee's location; some use a national benchmark with location modifiers.
• Document your methodology. When you present findings to leadership or defend decisions to employees, you need to be able to explain where the numbers came from.

Step 3: Find outliers — both directions

For each role/level group, map every employee's salary against the benchmark range. You're looking for:

• Below P25 — a material fairness and retention risk. First fix priority regardless of demographic breakdown.
• P25–P50 — potentially acceptable depending on tenure and performance. Are these employees newer to the role, or have they been here for years without a pay review?
• P50–P75 — the healthy zone for most employees, assuming your comp philosophy targets somewhere in this range.
• Above P75 — not necessarily a problem, but note it. The concern is when the explanation isn't clear.

Within each group, also look at the internal spread — the gap between the lowest- and highest-paid person at the same level. A spread above 20–25% typically means either the levelling is wrong, or someone's pay has drifted significantly.

Step 4: Run a demographic cut

Once you have the market-relative positions mapped, run a demographic cut: does any protected characteristic (gender, ethnicity, age, etc.) predict pay position within a group? This is the heart of a pay equity audit in the legal sense.

You're looking for patterns, not one-off cases. A single outlier in a small group can be noise; a consistent pattern — for example, women in a job family sitting systematically $8K–$12K below men at the same level — is a structural issue.

Control for legitimate differentiators before drawing conclusions: tenure within level, geographic location, documented performance tier, and time since last adjustment. The test is whether the gap remains after controlling for those factors. If it does, you have an unexplained gap — and that's where you act.

Prioritising and funding fixes

The audit will produce more findings than you can fix in one budget cycle. Prioritise in this order:

1. Below-market employees with unexplained demographic gaps — highest urgency; potential legal exposure and the clearest fairness case.
2. Below P25 for any reason — market floor issues create retention risk and are defensible to fund.
3. Systematic internal gaps within groups — where two people doing the same work are more than 15–20% apart and the difference isn't explained by tenure or performance.
4. P25–P50 employees with long tenure and strong performance — these are people who've been quietly underpaid relative to newer hires who negotiated better.

When building the business case for budget, frame it in terms the CFO will respond to: replacing an employee typically costs 50–200% of annual salary in recruiting, onboarding, and lost productivity. Closing a $10K gap for a senior engineer who's likely to leave costs a fraction of replacing them.

Avoid the common mistake of fixing gaps only at review cycles. A mid-cycle adjustment is a legitimate tool — and employees will notice if the company waits eight months to act on a finding it already knows about.

Building in ongoing monitoring

A one-time audit is a point-in-time fix. The gaps will re-emerge unless you build equity monitoring into your compensation cycle.

Practically, this means:

• Running a benchmark refresh at least annually — a benchmark from 18 months ago may no longer reflect reality for fast-moving roles like data science or security engineering.
• Adding an equity check to every offer approval — before extending an offer, check how it positions relative to current employees at the same level. Offers that land below P25 for the group should require explicit sign-off.
• Reviewing the demographic cut at every comp cycle — not just as a one-off project.
• Tracking compa-ratio trends — the ratio of each employee's pay to the band midpoint. A declining compa-ratio for a specific demographic is an early warning sign.

Where EvenBetter fits in

The most time-consuming part of a pay equity audit is usually pulling reliable benchmark data for each role and level. When that data comes from a stale survey or a generic salary range from a job board, the entire analysis is built on a shaky foundation.

EvenBetter lets you benchmark any role in under 60 seconds — paste the full job description, and get a source-cited, signal-strength-rated range triangulated across multiple live data sources and weighted by recency. The output shows P25/P50/P75 for the role as described, not just the title, so you're comparing against genuinely comparable positions rather than a broad label that spans multiple scopes and levels. For teams running equity audits across dozens of roles, that speed and consistency matters: you're working from the same methodology for every data point.

See our methodology for how the benchmarks are built, or check EvenBetter's public benchmarks to get a sense of what the output looks like for common roles.

The honest summary

A pay equity audit isn't about discovering that you've been doing something wrong. Most gaps aren't the product of bad intent — they're the product of inconsistent processes accumulating over time without a shared reference point. The audit is how you find the drift and correct it before it becomes a retention problem, a legal risk, or a headline.

The companies that do this well treat it as a recurring discipline, not a one-time project. They benchmark consistently, document decisions, communicate clearly, and fix gaps mid-cycle when they find them. That combination — market accuracy, internal consistency, and genuine transparency — is what "paying fairly" looks like in practice.