What problem does it solve?
A typical pay equity audit is a periodic statistical regression over the whole workforce that finds gaps after they exist. Between audits, new hire offers, merit increases and promotions can each create or widen a gap, and the only remedy is a remediation budget to close them after the fact. Elevance Health's compensation team described the result before it changed its process: "Managers didn't know what they didn't know," so they proposed offers that were competitive with market rates but not always in line with internal pay equity.
Regulation is also moving from annual reporting to real time obligations. The EU Pay Transparency Directive requires employers to give candidates pay information before the interview or otherwise before the contract, and requires larger employers to report gender pay gaps, with a mandatory joint pay assessment when a gap of at least 5% in any category of workers is not justified and not remedied within six months. A compensation team that only checks equity once a year finds out about a breach after it has already made the disclosures the law requires to be accurate.
How does it work?
- Baseline analysis. The system runs a regression across the current workforce, grouping employees doing substantially similar work, and controls for legitimate factors such as experience, performance and location to isolate any unexplained gap by gender, race or other protected characteristic.
- Remediation modelling. For every gap found, it models the cost and equity impact of closing it in different ways (a targeted raise, phased over cycles, by level or by location) so the compensation team can budget and prioritize.
- Decision time check. When a recruiter enters a candidate's proposed starting salary, or a manager enters a raise or promotion increase, the system returns the equitable range for that role, level and location in the same workflow, before the offer or increase is finalized.
- Continuous monitoring. Every governed decision is logged, so the next baseline analysis starts from a smaller, better documented gap instead of rediscovering the same disparities.
- Audience
- Employee facing
- Autonomy
- Assist
- Adoption
- Early adopters
- Channels
- Internal tools
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Cost reduction | Too few to pool | 25% | 1 | 1 vendor |
Value drivers: Risk and loss reduction, Compliance quality, Employee productivity.
Indicative value
An employer with 20,000 employees governing 25,000 pay decisions a year
USD 100,000 to USD 187,500
Remediation cost avoided per year
How this is calculated
Formula: payDecisions * shareWithGap * avgRemediationPerGap * avoidedShare. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Pay decisions governed per year (new hires, raises, promotions) payDecisions, decisions per year | 25,000 | 25,000 | Editorial assumption scaled from Syndio's reported 102,000 pay decisions a year for Salesforce's roughly 80,000 employees, https://synd.io/case-study/salesforce/. |
| Share of decisions that would otherwise create or widen an unexplained pay gap shareWithGap, fraction of decisions | 0.02 | 0.03 | Editorial assumption, replace with your own baseline audit findings. |
| Average later remediation cost per decision that created a gap avgRemediationPerGap, USD per decision | 800 | 1,000 | Editorial assumption. The high end is kept so total baseline remediation for the reference organization stays near Salesforce's reported pattern of about $3 million a year for about 80,000 employees, roughly $37 per employee per year, https://synd.io/case-study/salesforce/. Replace with your own remediation data. |
| Share of that later remediation avoided by catching the decision upfront avoidedShare, fraction | 0.25 | 0.25 | Syndio reports that Elevance Health cut pay equity remediation costs 25% after adopting decision time pay guidance, https://synd.io/wp-content/uploads/2024/01/01_2024_Case-Study_Elevance-Main.pdf. |
What it leaves out: Counts only avoided remediation spend on pay decisions. It leaves out legal and reputational risk from an unremedied gap, recruiter and compensation team time, the software cost, and any effect on offer acceptance or retention.
Who already uses it?
2 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
Salesforce
United States · Technology and software · 2026
Salesforce, with about 80,000 employees, governs about 102,000 pay decisions a year, and Syndio's platform manages the complexity of that analysis across more than 100 countries. Benioff first took a public stand on pay equity in 2015. Salesforce initially ran pay equity analysis in house with a model built by its own data science team, and later replaced it with Syndio's platform as the company scaled.
No outcome disclosed.
Elevance Health
United States · Insurance · 2023
Elevance Health, a US health insurer, partnered with Syndio in 2020 for a gender and race pay equity analysis and began using Syndio's Pay Finder tool in 2021 so that more than 200 internal and external recruiters can check a candidate's proposed salary against an equitable, competitive range before an offer goes out, instead of only auditing pay equity after the fact.
- Cost reduction: 25%
"Pay Finder helped Elevance Health maintain pay equity with every new starting salary, resulting in a 25% reduction in remediation costs."
Claimed by: vendor - Cycle time reduction: 6%
"This resulted in a 6% decrease in time to fill and 6% increase in offer acceptance rate."
Claimed by: vendor - Conversion uplift: 6%
"This resulted in a 6% decrease in time to fill and 6% increase in offer acceptance rate."
Claimed by: vendor
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- Compensation, job level and location data for every employee, kept current
- A job architecture that groups roles doing substantially similar work
- Demographic data collected and stored under local law, with a lawful basis for equity analysis
- Pay bands and a policy for how to remediate a confirmed gap
Systems to integrate
- HR information system for compensation and job data
- Applicant tracking system, to check offers before they are extended
- Compensation planning and merit review tools, to check raises and promotions
- Payroll, for the final governed pay figure
Complexity: Medium
The statistical method is well established, the harder part is data: getting clean, comparable compensation, role and demographic data across countries with different pay data rules, and getting the decision time check into the actual hiring and pay review workflow so managers use it instead of working around it.
- 1
Run the baseline analysis before anything else
Start with a full regression across the current workforce, with legal counsel involved so the analysis and its findings are handled correctly, before any decision time check goes live.
- 2
Put the check at the point of the decision
Elevance Health's recruiters look up the equitable range in Syndio's Pay Finder before making an offer, at the point the decision is made, rather than as a separate report checked later.
- 3
Set a remediation policy in advance
Decide how a confirmed gap gets closed, immediately, at the next cycle, over how many cycles, before the analysis finds one, so a real finding does not stall on a policy debate.
- 4
Train managers on what the range means
A recommended equitable range is not a hiring decision. Explain to managers and recruiters why it exists and what to do when a strong candidate wants more than the range supports.
- 5
Rerun the baseline on a fixed cycle
Reanalyze at least yearly, and whenever a new pay transparency or reporting obligation applies in a country the organization operates in, so disclosures stay accurate.
Guardrails
- Every remediation decision is approved by the compensation team, not applied automatically
- Protected characteristics are used only inside the statistical analysis, never as a matching feature that changes an individual's recommended range
- Access to individual level results limited to compensation and legal roles
- Methodology and grouping logic documented and available for audit
KPIs to instrument
- Remediation cost per cycle, and per employee as headcount grows
- Share of new pay decisions that fall inside the equitable range without an override
- Unexplained gap size at each baseline analysis, trended over time
- Time from a confirmed gap to remediation
Human in the loop
The compensation team owns the methodology, approves every remediation, and decides how to close a confirmed gap. Recruiters and managers see a recommended range but make the actual offer or raise decision; nothing changes an employee's pay without a person approving it.
Common failure modes
- Managers override the range without review
- A recommended range that is easy to ignore gets ignored under hiring pressure. Log every override with a reason and review the pattern, not just the individual case.
- Job architecture hides real differences
- Grouping roles too broadly compares people doing genuinely different work and produces a false gap, or too narrowly hides a real one. Have compensation and legal review the groupings, not just the statistics.
- Demographic data is incomplete or unlawfully collected
- Analysis needs demographic data the organization may not lawfully hold everywhere. Confirm the lawful basis and data quality per country before analyzing, and disclose gaps in coverage.
What are the risks and rules?
EU AI Act
Depends on design
The tier depends on what the range is used for. When it materially influences an individual hiring offer, it falls under Annex III point 4(a) (recruitment); when it materially influences an individual raise or promotion decision, it falls under point 4(b) (decisions affecting the terms of a work relationship). A recommended range that a recruiter or manager can accept or override, as this use case is designed, still counts as materially influencing that decision. A separate system that only produces an aggregate regression report for the compensation team, with no individual recommendation, is either outside Annex III on that narrower use or falls under Article 6(3) as a preparatory task that detects deviations from prior decision patterns rather than a high risk use in itself. Deployers of the high risk part must inform workers' representatives and affected employees before use (Article 26).
Rules that apply
Guidance
- Directive (EU) 2023/970 on pay transparency (European Union, Europe). Requires employers to give candidates pay information, such as in a job vacancy notice, before the interview or otherwise before the contract. Requires employers with at least 100 workers to report gender pay gaps, with a mandatory joint pay assessment when a gap of at least 5% in any category of workers is not justified by objective, gender neutral criteria and not remedied within six months. Member states had to transpose it by 7 June 2026; EUR-Lex blocks automated fetches, so this cites the Wayback copy.
- Annex III, high risk AI systems referred to in Article 6(2) (European Union, Europe). Point 4 covers employment, workers' management and access to self employment, including decisions affecting the terms of a work relationship.
Controls to put in place
- Legal counsel involved in the analysis methodology and any remediation
- Named compensation owner for the model, its groupings and its overrides
- Country by country review of what demographic data may lawfully be collected and analyzed
- Documented, auditable methodology, separate from the vendor's proprietary scoring
Frequently asked questions
- What results do companies report from decision time pay equity checks?
- Syndio reports that Elevance Health cut remediation costs 25% after putting Pay Finder in front of more than 200 recruiters, checking each candidate's proposed salary before an offer goes out. Syndio also reports that Salesforce held its roughly $3 million annual remediation cost steady while governing about 102,000 pay decisions a year and roughly tripling its headcount; Syndio's case study attributes this to pay staying corrected through later merit cycles and promotions, though it does not describe a check at the point of each individual offer or raise.
- Is this different from an annual pay equity audit?
- An annual audit still matters for the aggregate baseline and for reporting, but it only finds gaps after new ones have already been created. A decision time check catches a specific offer or raise before it goes out, so fewer gaps reach the next audit.
- Does the AI decide who gets what raise?
- No. It analyzes patterns and recommends a range; the compensation team, recruiter or manager makes the actual pay decision and can document a reason to go outside the range.
- Is pay equity analysis high risk under the EU AI Act?
- It depends on the design. A range that materially influences an individual hiring offer, raise or promotion falls under Annex III point 4(a) or 4(b). A separate tool that only produces an aggregate regression report for the compensation team, with no individual recommendation, is either outside Annex III on that use or a preparatory task under Article 6(3).
How to cite this page
Blits.ai AI Use Case Library, "AI for compensation governance and pay equity analysis", last verified 29 September 2026, https://www.blits.ai/ai-use-cases/compensation-and-pay-equity-analysis. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 29 September 2026: First published