PayAnalytics - September 2026 feature release notes

This article lists the updates made available to users in the PayAnalytics application in September 2026.

Statistical significance for pay gaps

This is an note/information message.

This new feature addresses pay equity requirements in the US. A small number of selected customers are testing its initial version; it is turned on case by case, not automatically. US customers interested in testing the feature can reach out through support in the Help Center.

We have added the option to run a pay equity analysis against a statistical significance goal. Instead of only measuring the size of a pay gap, PayAnalytics now states whether the gap is large enough to be real. It also suggests the raises needed to bring this gap down.

A pay gap counts as significant when its confidence interval clears zero, and it is unlikely to result from ordinary variation in pay alone. This is judged against a configurable threshold, entered in whichever unit a legal or reporting team already works in and defaulting to a 95% confidence level. Once suggested raises shrink the gap enough for zero to fall inside that range, the gap is no longer significant. This does not prove the gap is exactly zero; it means the data shows no systematic gap.

How significance works

Significance depends on group size as well as gap size. A larger population needs a smaller gap to count as significant, while a smaller population can carry a larger gap and still fall within the expected range. Therefore, significance cannot be read off the pay gap percentage alone and is separate from a reporting threshold such as the EU Pay Transparency Directive's 5% figure. A gap smaller than that threshold can still be significant, while a larger one may not be.

To use the feature and see how PayAnalytics judges each pay gap's significance, proceed as follows:

1. Start a new pay equity analysis, or re-run an existing one, and select the dataset and analysis as usual. Once the model is validated, open the Set pay equity target step and enable raise suggestions.

2. Under Pay gap reduction method, choose Significance target.

3. Set the threshold under Select significance threshold, as a Significance level (e.g., 0.05) or in Standard deviations (e.g., 1.96). The other value updates automatically. It defaults to a 95% confidence level.

4. Optionally, add subgroups or a demographic variable, such as country or ethnicity, with or without intersecting with gender.

5. Run the analysis. The verdict and supporting figures appear on the Analysis overview, the Pay gaps tab, and the subgroup overview, before and after raises.

Setting a significance target

On the Analysis overview, the two adjusted pay gap tiles include the verdict, and the Results table shows the p-value and threshold in standard deviations before and after raises. In the figure below, the adjusted gender pay gap is Significant at 9.7% before raises and Insignificant at 0.7% after.

Significant gap before raises, insignificant after

When the analysis is split into subgroups, the subgroup overview reports the verdict for each subgroup on its own headcount, where group size matters most. A larger population can be significant at a gap that a smaller population carries comfortably within range, so two subgroups with a similar pay gap can reach opposite verdicts.

Subgroup verdicts by country

The analysis can also be run on a demographic variable such as ethnicity, with each group measured against its reference group. When several groups are compared, the two adjusted tiles summarize how many gaps are still significant rather than a single percentage, before and after raises.

Ethnicity as the demographic variable

Known limitations

This early release has the following limitations:

• Reports do not yet show the significance verdict. It appears on the analysis screens and their Excel exports, but not in the generated report section.

• Manual raise changes are not re-evaluated for significance. Adjusting a raise for a single employee by hand does not recalculate it, and the after-raises verdict is withheld once raises are edited.

• Significance is shown only for the significance target. Closing the gap by target pay gap or by budget, or analyses run before this release, show no significance verdict. The Compensation model still shows the p-value for a demographic variable, which can be used to judge whether its gap is statistically significant.

• Closing pay gaps for each worker category is not yet supported with this goal. The significance target drives the overall and subgroup analyses, not that flow.

We plan to address these limitations in future iterations.

Demo

Language selection for Pay Transparency Export

Previously, every report produced by Pay Transparency Export came out in English, regardless of the language an organization or its employees worked in. Users can currently choose a language for each bulk export, so a batch is generated in a language other than English, with the report text itself rendered in the language selected.

The language defaults to the application's current language when a new bulk export is created, and can be changed to any other available language. This matters most when reports go to employees in different countries. For example, a Rewards team working in the application in Polish can produce a French batch for employees in France and a German batch for employees in Germany in the same session.

The single-employee Pay Transparency report already generated its file in the application's language. With this release, bulk export behaves the same way, so reports for several countries can be created without changing the language settings. The Export list has also gained a Language column, showing the language used for each export alongside its dataset, analysis, employee count, file size, status, and dates.

To generate a bulk export in another language:

1. Go to Inform > Pay Transparency Export and click New Bulk Export.

2. Select the analysis to export. Then, choose the language for the export. It defaults to the language currently in use; change it to any available language you need.

3. Optionally, apply filters, for example by country, to scope the export to the right population, and give the export a name.

4. Generate the export. Once its status shows Completed, download it from the Export list.

Thanks to this enhancement, one export run can now cover as many languages as the organization needs, without adding steps for the team producing it.

Export list showing the language of each generated export

Demo

Audit trail for pay transparency statements

Article 7 of the EU Pay Transparency Directive requires employers to respond to right-to-information requests within a set deadline. The audit trail, which timestamps every action, helps show that each pay transparency statement was prepared within that deadline. Found in System > Audit log and usually available only to admin users, it now records who produced every statement, for which employee, and when, whether from an individual view or as part of a bulk export.

When someone opens an individual employee's statement, the system logs the action. Printing the statement is recorded as a Printed action, and exporting it to Excel – as a Downloaded action. Each of these entries carries the employee's ID in its metadata.

Bulk exports are logged in two places. Generating the export is recorded as a Created action against the export itself, and downloading the generated file – as a Downloaded action. A bulk download entry does not list the individual employees on its own, but the matching Created entry carries a downloadable List of employees, so users can confirm if an employee was part of that export.

The logs can additionally be filtered by action, user, object, IP address, or date range. For instance, to check whether a statement was ever produced or downloaded for a specific employee, administrators should filter by the applicable action and look for the employee's ID, or open the List of employees on the relevant bulk export.

The new feature gives Security and Compliance teams a clear record of every right-to-information statement that leaves the platform, which makes the generation and communication of such documents far easier to audit.

Printed statements in the Audit log, each carrying the employee's ID

Downloaded individual statements and bulk exports in the Audit log

Created bulk export, with a downloadable List of employees

Demo

Population labels on pay gap figures

When access to an analysis is limited to part of the population, the Adjusted pay gap and Unadjusted pay gap figures in Pay Equity Analysis > Analysis overview can describe different groups of employees. The system recalculates the unadjusted gap over only the visible records, and the adjusted gap always comes from a model fitted on the whole population. Previously, only the Unadjusted pay gap figure was labeled as covering visible records, while the Adjusted pay gap figure carried no label. As a result, the two numbers could look directly comparable when in fact they were not.

Currently, each pay gap figure includes a label naming the population it covers, such as (Whole population) or (Visible records only), which makes it possible to read the two gaps correctly at a glance. The same labels appear on the subgroup overview table and for individual subgroup analysis.

Labels appear only when the two figures cover different groups. When access covers every record, or fully covers the group shown, both numbers specify the same employees and no labels are displayed, so nothing changes.

HR professionals and Compensation teams working with restricted data access can now read the adjusted and unadjusted pay gap figures with confidence, knowing immediately whether these numbers describe the same group of employees.

No population label on the Adjusted pay gap

Adjusted pay gap now labeled (Whole population)