The Movement view is the headline answer to “did things change?”. It follows the same people between cycles: only employees with an extracted performance level in consecutive cycles contribute to the flows, so what you see is genuine transitions rather than population churn.
Like the rest of Compare Reviews, it needs at least two cycles selected in the sidebar, both with AI extraction complete.
What this page answers
How did people flow between performance levels from cycle to cycle? Which individuals changed the most? Which departments improved, held steady, or declined?
What’s on the page
Performance movement
A flow (Sankey) diagram spanning every selected cycle, with one column per cycle in chronological order. Each column splits into the performance levels — Exceptional, Strong, Solid, Concerning, Critical, Insufficient Data — colour-coded consistently with the rest of PI. The bands between columns show how many people moved from each level in one cycle to each level in the next.
Thick bands staying at the same level mean stability; bands sloping up or down are the movement story. With three or more cycles selected, the diagram chains the transitions, so you can trace a group’s path across the whole period.
Biggest movers
A sortable table of the individuals who changed most between the earliest and latest selected cycle:
Name — Click to open the employee detail flyout with their full cross-cycle profile
Department
Prior AI Perf / Current AI Perf — Compass’s performance level in the first and last cycle
Sentiment Δ — How their review sentiment score changed, shown as a trend
Movement — A composite movement score, shown as a trend
The movement score is driven mainly by change in performance level, with a smaller adjustment when review sentiment also shifted markedly. People with no meaningful movement are excluded, and the table is sorted by movement magnitude — biggest changes first, up or down. If nobody moved, you’ll see “No significant movement detected”.
Movement by department
A heatmap of departments against three directions — Improved, Stable, Declined — counting how many people in each department moved each way between the earliest and latest cycle. It’s the quickest way to see whether change is broad-based or concentrated in one or two teams.
Where the data comes from
All three cards compare the stored per-cycle review extraction performance levels and sentiment scores. The Sankey compares each consecutive pair of cycles; the movers table and department heatmap compare the first and last selected cycles directly. Someone missing a performance level in either cycle of a pair simply doesn’t appear in that comparison.
How to read it
Read the Sankey’s diagonal first. A healthy cohort shows thick horizontal bands (people holding their level) with modest upward flow. Heavy downward flow into Concerning or Critical is the signal to act on.
Downward movers deserve the first look. Someone dropping two levels between cycles is either a real performance change, a manager change, or a calibration problem — all three are worth knowing about.
Cross-check big movers against Sentiment Δ. A performance drop with a matching sentiment drop reads like a genuine decline; a performance drop with flat sentiment may be reviewer drift.
Use the department heatmap to separate signal from noise. One department accounting for most of the “Declined” column points at a local cause — a new manager, a reorganisation, a tough year for that team.
Tips
The movers table compares first vs last cycle only. With three or more cycles selected, the middle cycles shape the Sankey but not the table — someone who dipped and recovered will show as unmoved.
Click through before concluding. The employee flyout shows the underlying reviews for each cycle — always read the narrative before treating a mover as a fact.
Remember who’s excluded. Joiners, leavers, and anyone skipped by extraction in one of the cycles don’t flow — a thin Sankey relative to headcount usually means low overlap between the cycle populations.