Group Insights

Group Insights computes two kinds of pay gaps: within an equal-work group (comparing a female-dominated role — where women’s share reaches your threshold, 60% by default — against higher-paid non-dominated roles in the same group) and across groups at the same location (comparing against higher-paid roles in a lower-demand group). The unexplained gap is the full raw gap. The former flat experience, tenure and performance weights are no longer subtracted, because they were not based on any employee data (see System Parameters).

If job-evaluation criteria are recorded (for equal-work groups under Salary Mapping → Job evaluation, or on a job category), they are shown alongside the cross-group comparison. If the recorded criteria contradict the ranking — the comparator group ranked lower (assumed lower requirements) has a higher weighted criteria score (1–5 scale, with the weights from Job evaluation → Categories) than the female-dominated role’s group — a warning is also shown on the comparison row so the grouping can be reviewed. The warning is advisory only and never changes the gap calculation itself.

How groups are ranked in the cross-group comparison. A role is only compared against higher-paid roles in a group ranked lower (lower rank = lower requirements):

A regular equal-work group (EQUAL WORK GROUP) ranks by its number — group “2” has rank 2.
A job category ranks according to the Job category ranking setting (Settings → Salary Mapping – settings → System parameters): either by the category’s Level (the default, set per category under Salary Mapping → Job evaluation → Job categories) or by its Criteria points (the weighted score of its recorded job-evaluation criteria, 1–5 scale).
Both kinds are compared on the same scale: a job category at level 3 is compared against group “2” as 3 vs 2, so group 2 counts as lower requirements. If you mix both, set the levels so they line up with your group numbers. With Criteria points as the basis a category’s rank is its weighted score on the 1–5 scale (it used to be the 4–20 sum), which is rarely the same scale as group numbers — pick that basis mainly when every employee has a job category.
A group with no rank is left out of the cross-group comparison: a job category with no level (or, with Criteria points, without complete criteria), or a regular group with no number such as “Unknown”. It is still analysed within its own group. A job category is never ranked by digits in its name — “Engineering/Level 2” with no level set has no rank.

If the Variable pay feature is enabled, an “Include variable pay” checkbox appears at the top of the page (the same checkbox also appears on the Overview, Classification table, Individual Analysis, Salary Band and Benchmark tool). Checking it adds each employee’s this-year bonus/commission (converted to a monthly amount) on top of base salary before the gap is computed — otherwise base salary alone is shown, exactly as before. Both views (base salary only / total compensation) stay available at all times; checking the box just switches which one you’re currently looking at, it doesn’t remove the other.

Part-time salaries are converted to full-time for the comparison. Every comparison figure — medians, pay gaps, and the targets shown — is based on the full-time equivalent salary: an employee’s actual pay divided by their working-time fraction (FULL-TIME EQ.). Without it, a part-time employee would be compared directly against a full-timer, which systematically manufactures gaps that don’t exist — particularly in female-dominated roles, where part-time work is more common.

The salary shown for an individual employee is always their actual pay, as is the suggested increase. No figure in the product is a salary someone doesn’t have. If the working-time fraction is missing, zero or invalid, the employee is treated as full-time — that can never inflate someone’s converted salary, and so can never manufacture a gap against them.

Who is counted. An employee is part of the analysis unless “Include in analysis” is explicitly set to no — a missing value (for example after an import without that column) counts as included, exactly as on the Overview and in Individual Analysis.

Small roles. If you are allowed to see individual salaries (Admin or a role allowed to see salaries), every role takes part in the comparisons, including roles with one or two people, and the page says “Showing all figures – you are allowed to see individual salaries”. For everyone else only roles with at least Minimum group size employees (at least 3) take part, because a median over one or two people effectively is their salary.

Pay policies per group. When the “Scoped policies” feature is on, each analysis card shows which pay policies apply to the role in the group, and how many of its people a policy covers when it does not cover all of them. This is the basis when a pay difference needs to be justified.

How it is calculated. Each role’s median of full-time-equivalent pay is calculated. A role is female-dominated when the share of women among those with a salary is at least the threshold (normally 60 %). Each female-dominated role is compared with the roles that are not female-dominated and have a higher median — within the same group and, at the same location, in groups with a lower rank. The gap = (comparator role’s median − female-dominated role’s median) ÷ female-dominated role’s median × 100. The card also shows the female-dominated role’s median as a percentage of the highest-paid comparator. Example: Assistant Nurse (4 women, 1 man, median SEK 30,000) against Operations Technician (1 woman, 4 men, median SEK 34,500) in the same group: (34,500 − 30,000) ÷ 30,000 = 15.0 %, and 30,000 ÷ 34,500 = 87 %. Note that the gap is measured against the lower-paid role; the EU report’s pay gap is measured against men’s pay and gives 13.0 % for the same salaries. The whole difference is unexplained until you justify it. There is no minimum gap: every higher-paid comparator is shown. Every concept, with one shared worked example, is in the article Concepts and calculations.