This article explains what Case Management is, how to configure it, and how to review, justify, and resolve individual cases. Case Management is where you prepare for what employees will be able to see under their right to pay information: it flags the specific people whose pay sits outside an expected range, so you can deal with each situation in advance - either record a clear explanation, or conclude that there is no explanation and the case needs to be remediated.
WHEN TO USE THIS ARTICLE
Use this article when:
you want to review individual employees flagged as pay outliers
you have defined your pay factors and know which ones have the biggest impact on pay in your organisation
you need to record justifications for pay differences (for example, to support EU Pay Transparency Directive obligations)
WHAT IS CASE MANAGEMENT
You find Case Management in the left-hand navigation.
PayGap compares every employee's pay against the pay of their comparable peers (category of worker) and produces an index (more on this below). When an employee's index falls outside the range you have configured, PayGap automatically creates a case for them and flags it as Pending review.
A case is simply one employee flagged for review. Case Management is where you:
see who has been flagged and why
review the individual figures behind the flag
record an explanation (a justification), or conclude there is none and mark the case for remediation
Why this matters - get ahead of the right to pay information.
Under the right to pay information, employees can request the figures that compare their pay with their peers. Case Management lets you find and handle those situations before they reach the employee: for each flagged person you either record a defensible explanation backed by evidence, or you accept that the difference cannot be justified and flag it to be remediated. Nothing should come as a surprise when an employee asks.
Knowledge is retained over time. Every time you upload new data the reasoning a manager records stays attached to the case. So when an index moves again at the next upload, you can see the full history and add a separate justification for that specific change - the institutional knowledge does not disappear when a manager leaves,
STEP-BY-STEP GUIDE
1. Configure Case Management in Settings
Before you review anything, set the rules that decide who gets flagged. Go to Organisation → Settings → Analysis → Case management.
You configure three thresholds and one method:
Lower bound - flag a case when the index falls below this percentage. Default: 80%.
Upper bound - flag a case when the index exceeds this percentage. Default: 120%.
Change threshold - trigger a fresh review when the index moves by at least this amount compared with the previous value, Default: 0 (off).
Index benchmark method - whether the comparable-job benchmark behind the index is based on the Mean or the Median.
Mean or Median - which to choose? The regulations require you to disclose information to employees based on the mean (average), so we suggest setting Mean as your benchmark - it keeps Case Management aligned with what you will actually disclose. It is still worth re-running the analysis on the Median as a cross-check. The mean is sensitive to outliers: a few very high or very low salaries in a comparable job pull the average away from where most people sit, which can flag people whose pay is actually normal for that group. The median is not affected this way. So compare the two: cases that appear under the Mean but disappear under the Median are usually a sign that the benchmark is being distorted by a small number of extreme salaries in the peer group - check that group's pay distribution before treating the case as a genuine outlier. Cases that persist under both measures are robust to the choice of benchmark and more likely to reflect a real pay position that needs a justification. Use the Mean for disclosure, and the Median as a sense-check. For a fuller explanation of how the two measures behave, see Understanding the measures in Base Analysis.
Anyone whose index lands outside 80–120% (with the default settings) is flagged as Pending review in Case Management. Everyone inside the range shows as Within valid range.
These settings are the Global level, but they can be overridden per legal entity. Set a sensible group-wide standard here, then adjust the bounds for any legal entity that needs a different range (for example, a market with very different pay structures).
2. Open Case Management and read the summary tiles
Open Case Management from the left-hand navigation. The top of the page gives you the headline counts before you touch the table.
Employment Cases - the total number of cases, split by gender.
Employment Case States - the number of cases in each status, also split by gender:
Pending review - index is outside the bounds and no one has looked at it yet (your starting point)
Under review - someone is actively working on it
Deferred - review is on hold; action will be taken later or once more data arrives
Resolved - all outstanding issues addressed, case closed
Reopened - previously resolved, but reopened due to new information or expired evidence
Within valid range - index is inside the bounds; no action required
3. Find and filter cases
Use the controls above the table to narrow down to the cases you care about:
Search by employee ID or name
Employment case status - filter to one or more statuses (for example, only Pending review)
Only active - toggle on to hide employees who are inactive (leavers or marked as inactive)
Columns - choose which columns appear in the table. Switch off the indices you do not need for a shorter, simpler view.
4. Understand the table
Each row is one employee. The columns are:
Employee ID, Gender, Job Title, Legal Entity - who the person is.
Index - the ratio of the employee's pay to the average pay (mean or median, depending on your Settings) of everyone in the same comparable job. 100% = exactly at the benchmark; below 100% = paid below peers; above 100% = paid above peers.
Same-gender index - the same ratio, but the benchmark is built only from employees of the same gender.
Opposite-gender index - the same ratio, benchmark built from the opposite gender. Always calculated on total cash, because this is the basis on which employees receive their figures under the right to pay information.
Predicted index - what the pay prediction regression model expects this person to earn, on the same 100% scale (use only if that model has been finalized).
Index difference - the gap between the actual Index (vs all comparable peers) and the Predicted index. Read it as the part of the person's pay position the model cannot explain. It can run in either direction (positive or negative).
Case status - reflects whether the index sits inside the bounds you configured in Step 1.
Active employment - whether the person is still an active employee.
Sort and export:
Click the Index column (or any index column) to sort from lowest to highest, or highest to lowest.
Use the export icon in the top-right above the table to download the list to Excel - useful when you need to share a worklist with the people who will address each case and don't have access to PayGap.
5. Decide where to start
You will rarely review every case at once. Work through them in priority order:
Filter to Pending review and sort by Index, ascending. The people paid furthest below their peers come first.
Work the extremes on both ends - very low indices first (potential underpayment), then very high indices (premiums that need investigation as they bring the average up).
Use Index difference to separate the explained from the unexplained: a large difference means the model cannot account for the position, so it needs a human explanation.
Compare Index, Same-gender index, and Opposite-gender index. If someone sits near the benchmark for their own gender but far from the opposite gender, the gap is gendered and deserves closer attention (it's likely employee will challenge that)
Ask yourself:
Is this person genuinely an outlier, or is the comparable group very small or even incorrect?
Does the index difference point to something the model is missing (a real pay factor) or to a gap we cannot defend?
Is the pattern one-off, or does it repeat across a job, team, or legal entity?
Red flags:
A low index that lines up with gender (e.g. a woman well below both her peers and the opposite-gender benchmark)
A large, persistent index difference with no recorded justification
Extreme indices in comparable jobs with very few people - statistically unreliable, but still worth checking case by case
Extremely low or high indices can also point to a data error or missing data standardisation (for example, a salary entered in the wrong currency, as an hourly instead of annual figure, or on a different FTE basis). Before treating an extreme value as a real outlier, check that the underlying pay data is correct and consistently standardised.
6. Open a case, justify it, and resolve it
Click an employee in the table to open their employee profile, where the full case lives.
The activity log shows every automatic update PayGap has made over time (e.g. "PayGap detected change in data and updated the case to Within valid range for Index 98.92%").
Current employment state shows the live Case status, Index, Predicted Index, and when any evidence expires.
Use Change status… to move the case (e.g. to Under review, Deferred, or Resolved) and leave a comment to record context.
When you resolve a case, attach one or more justification reasons - the objective criterion that explains the pay difference - or pick Other:
Need help deciding? Click the ⓘ info icon to open the guidance panel.
It explains, side by side:
Status definitions - what each case status means
Justification reasons - for each criterion: when to use it, what to think about, what evidence is useful to attach, and a proportionality check
This guidance is there so a manager/HR records the right explanation with the right evidence the first time - and so that reasoning is preserved on the case for the next person who reviews it.
KEY INFORMATION
Case Management exists to get ahead of the right to pay information. Work each flagged case to one of two outcomes: a defensible justification backed by evidence, or an honest conclusion that the difference cannot be explained and must be remediated.
The Lower/Upper bounds are set Globally and can be overridden per legal entity. Cases are flagged using whichever bounds apply to that employee's legal entity.
Whether the index uses Mean or Median depends on the Index benchmark method in Settings - change it there and every index recalculates. We suggest Mean for disclosure (it matches what you report to employees) and the Median as a cross-check: cases that survive both measures are robust, while cases that only appear under the Mean may be an artefact of a skewed peer group.
Before treating an extreme index as a real outlier, rule out a data error or missing standardisation (currency, hourly vs annual, FTE basis).
Cases update automatically with each upload. A case can move from Pending review to Within valid range (or back) on its own as the data changes; your justifications and comments stay attached throughout.
The Opposite-gender index is always on total cash, matching what employees receive under their right to pay information.
A justification is only as strong as its evidence. Use the proportionality check in the info panel: if Market Forces is the only reason, the market premium should match the pay difference for that individual.
COMMON QUESTIONS / TROUBLESHOOTING
What is Case Management actually for?
It is where you prepare for the right to pay information. Every flagged case should end in one of two states before an employee could ever ask: a recorded justification with evidence, or a documented conclusion that the gap is unexplained and needs to be remediated. The goal is no surprises.
Who decides whether someone is flagged?
The thresholds in Settings → Analysis → Case management do. With the defaults, anyone whose index is below 80% or above 120% is flagged as Pending review. Adjust the bounds Globally, or per legal entity, to match your standards.
Should I use the Mean or the Median?
Use the Mean as your live setting - it matches the figure you disclose to employees. Re-run on the Median as a cross-check. If a case appears under the Mean but not the Median, the average is probably being distorted by a few extreme salaries in the peer group, so check that group before treating the case as a genuine outlier. Cases that show under both measures are robust. See Understanding the measures in Base Analysis.
An index looks extreme. Is it real?
Not necessarily. Very low or very high indices are sometimes a data problem - a salary in the wrong currency, an hourly figure entered as annual, or an inconsistent FTE basis - or the result of a comparable job with very few people. Check the underlying data and the group size before treating it as a genuine outlier.
What is the difference between Index and Predicted index?
Index is what the person earns relative to their comparable peers (category of worker). Predicted index is what the regression model expects them to earn given their pay factors. The Index difference between the two is the part the model cannot explain.
Why is the index always on total cash?
Because that is the basis on which employees are given their figures under the right to pay information. Keeping the column on total cash means the number you review is the number they could see.
Can I review fewer columns?
Yes. Use the Columns control above the table to hide the indices you do not need for a shorter, simpler view.
Can I export the list?
Yes. Use the export icon in the top-right above the table to download the cases to Excel - handy for sharing a worklist with the people responsible for each case.