The Calibration tab of Analytics > Performance reviews in Workleap shows which groups rate their people differently from the rest of the organization, and which barely tell their performers apart. Use this article to read the two signals it flags, Spread and Gap, and to know when a flag means something.
Who can use it: Executives. The tab reads the same period, cycles, and filters as the rest of the Analytics page (see How do I analyze performance reviews across cycles as an Executive?).
Read the Calibration signals
The Calibration signals block sums up the selection before the group table.
IN SCOPE: how many people and how many submitted reviews. They are shown side by side, not as a fraction, because reviews come from every reviewer type.
AVG REVIEW SCORE on the scale derived from the selection.
REVIEW SCORE DISTRIBUTION: the share of people in the lowest rating, in the middle ones, and in the top two. The three ranges are fixed to the scale, so a reading compares across cycles and groups.
RATING AVG BY TYPE: the average given by managers, in self-reviews, by upward reviewers, and by peers. A wide difference between types is itself a signal; for example peers rating lower than everyone else.
Group the table
The table of the Calibration tab has one row per group.
In Group by, pick Direct manager (the default), Department, Segment, Job title, or Tenure.
Read each row: the group, PEOPLE (how many rated people are behind the number), AVG, SPREAD, and GAP. Sort by average, spread, or gap.
Expand a row for its SCORE DISTRIBUTION (one bar per score) and the PEOPLE in it with their individual score.
A person on two teams counts in both groups; a person with no manager, team, or segment is left out of that grouping. A group scored only on performance or potential, with no overall score, is dropped rather than shown as zero.
Read Spread
Spread is how widely a group's own ratings differ, in points. A small spread means the group rated almost everyone the same, so it isn't telling its performers apart.
On a 5-point scale, a spread of 0.4 or less is flagged (red) and 0.6 or less is worth a look (yellow). On another scale, the same rule scales with it: flagged at 20% of the widest possible spread, worth a look at 30%.
Only a low spread is flagged; a high spread is not a problem.
Groups under 5 rated people are never flagged: one rating would swing them too far. The row still shows its spread.
Read Gap
Gap is how far the group's average sits from the organization's average, with a sign: positive means the group rates above the organization.
On a 5-point scale, a gap of 0.5 or more in either direction is flagged and 0.25 or more is worth a look. On another scale, flagged at 10% of the scale, worth a look at 5%.
The organization baseline is always the whole organization over its whole history, whatever filters you set, so a gap means the same thing in every view.
Sorting by gap sorts by size, so a group 0.5 below and one 0.5 above tie. Groups under 5 rated people are never flagged.
Tip: A flag is a reason to compare examples and rating criteria across groups before making people decisions, not a verdict. A manager with a high gap may have a strong team, or a lenient hand; the expanded row and the reviews themselves tell you which.
Open the 9-box
The 9-box of the selection, performance against potential, is on the Scores tab under the 9-box grid view. It uses the same bands as each cycle's calibration: low, medium, and high at about 22%, 56%, and 22% of each scale. See How do I read the results of one review cycle? for the cell names.
Limitations and workarounds
Small groups aren't hidden, just not graded. Treat a row with fewer than 5 people as anecdote.
The baseline can't be filtered. To compare two departments to each other rather than to the organization, use + Break down by on the Scores tab.
A comparison can't be shown per row. Comparing by the same dimension the table is grouped by is refused; the overlay applies to the chart instead.
Related
How do I analyze performance reviews across cycles as an Executive?
How is the Review Score calculated, and how do I set its formula?
How do I read the results of one review cycle?