The Risk lens surfaces where you may be losing people. Compass extracts risk indicators from every review (retention concerns, burnout language, disengagement signals) and this page rolls them up so you can see where the concerns concentrate.
What this page answers
Where are we losing people? Who’s at risk? Is the risk concentrated in one department, one performance level, or scattered? Who are the most unusual people in this cycle?
What’s on the page
Four headline KPIs
Total Risk Signals: count of all risk signals Compass found across the cycle
High Severity: count of signals Compass classified as high severity
Retention Risk: count of retention-related signals specifically
Burnout Risk: count of burnout-related signals specifically
If all four are zero, the page shows a positive alert (“No risk signals detected”) instead of the KPIs, heatmaps, and at-risk table — your cycle has no flagged risk. The Outliers table still appears if anyone is flagged as unusual.
Risk by department
A heatmap with risk type across the columns and department down the rows. Cells are counts. The colour intensity shows where the risk types concentrate.
A row dominated by one risk type tells you that department has a specific issue (e.g., burnout); a column dominated by one department tells you that risk type is concentrated there.
Risk by performance level
A second heatmap with risk severity (Low / Moderate / High) across the columns and performance level down the rows. This is the most important risk view if your worry is losing your best people: if High severity signals show up in the Strong and Exceptional rows, you’re at risk of losing your top performers.
A risk profile that piles up in Concerning / Critical is uncomfortable but expected — under-performers often disengage. A risk profile that piles up in Strong / Exceptional is the one you should act on first.
At-risk employees
A filterable, paginated table — one row per employee, showing their most severe signal:
Name: employee (clickable)
Department
Risk Type: multi-select filterable (retention, burnout, disengagement, etc.)
Severity: multi-select filterable (High / Moderate / Low, colour-coded)
Signal: the free-text description of the risk signal from the review
AI Performance: their performance level, colour-coded
This is your concrete intervention list. Filter to High severity + Strong+ performance to see the people to prioritise for a conversation.
Outliers
A filterable table of every employee who is unusual on at least one dimension, sorted with the lowest sentiment first. Each row shows:
Name
Department
Flag: the flag type (Low sentiment, Over-rater, Performance concern)
AI Performance: performance level
Sentiment: sentiment score (−1 to +1)
Perspective Alignment: their self-awareness gap category
Outliers aren’t necessarily at risk in the formal sense — they’re just unusual. Worth looking at because the unusual cases are often the ones worth a closer read.
Where the data comes from
Compass reads each review and looks for signals — language that suggests the person is thinking about leaving, exhausted, disengaged, or under-supported. Each signal is tagged with a type and a severity.
The outliers table additionally pulls from sentiment scores, self-awareness gap, and performance level — flagging anyone whose values are unusually low or who shows a strong over-rating pattern.
How to read it
Read the performance-level heatmap first. Risk concentrated in your top performers is your biggest threat.
Then read by department. If one department has a much higher risk count than the rest, dig in — usually it’s a specific manager issue, workload issue, or recent change.
Use the at-risk table as your follow-up list. Filter to High severity + the performance levels you care most about. Click into each name to read the underlying review.
Don’t treat low severity as “ignore”. Low-severity retention risk is the early-warning signal — easier to act on at low than high.
What Compass means by “risk type”
The types Compass uses aren’t a fixed taxonomy — they emerge from the review content. Common types you’ll see:
Retention: language suggesting they’re considering leaving
Burnout: workload, exhaustion, sustainability concerns
Disengagement: reduced involvement or motivation
Communication breakdown: friction with manager or team
Career stagnation: stuck in role, not progressing
Engagement decline: specifically a change from prior cycles
The exact types vary by review content; the heatmaps will show whichever types Compass detected.
Tips
The “no risk detected” message is worth taking seriously. If you genuinely have no risk signals, you’re either in a very healthy cycle or your reviews aren’t capturing enough information to spot risk. Read a few reviews to check.
The outliers table is designed to support moderation. Edge cases get debated longest — having the unusual list ready accelerates the conversation.
Act on risk signals promptly. Don’t let a backlog build up — risk indicators degrade quickly if you don’t follow up.