The Fast Feedback tab turns the firm’s feedback exchange into two streams of AI signal. A per-feedback quality analysis scores how well each piece of feedback is written; a separate language-analysis layer extracts what the feedback is about — the topics, subjects, and descriptors — and the sentiment attached to each. This section turns those signals into cohort views.
What this module answers
What’s the overall sentiment of the firm’s feedback, and is it improving? What topics is the firm talking about, positively and negatively? How well-written is the feedback being given? And what does the underlying feedback actually say?
The pages
The Fast Feedback sidebar has four pages:
Introduction — Explains how the analysis works before you read any charts: the six quality dimensions (1–10 each), the overall quality rating (Excellent / Good / Fair / Poor), the sentiment score, the fixed vocabulary of ten quality-issue flags, and the language-analysis layer’s topics, subjects, and descriptor-topic pairs. A queue status banner at the top shows when quality analysis jobs are still queued or processing.
Sentiment — Net Sentiment Index, sentiment trends, trending topics, sentiment by segment, subjects discussed, and the strengths / growth-areas word clouds. Covered in its own article.
Quality — The quality dimensions radar, quality distribution and trends, and common quality issues. Covered in its own article.
Research — The interactive topic explorer: pick a topic, filter by sentiment, and read the actual feedback. Covered in its own article.
Pick a date range
Unlike the review-based modules (which use review-cycle pickers), Fast Feedback uses a generic date range. Use the Date range dropdown in the sidebar — 3 months (default), 30 days, 60 days, 6 months, 12 months, or 2 years — plus the Department, Role, and Location filters below it. All three analysis pages scope to that window.
About the Net Sentiment Index (NSI)
NSI is the headline sentiment number used across this module. It combines the positive / neutral / negative classifications into a single score between −1 (uniformly negative) and +1 (uniformly positive), where more positive feedback lifts the score and more negative feedback lowers it. The Sentiment page shows it firm-wide, over time, and per segment.
Where the data comes from
Both streams are produced in the background as feedback entries are completed:
Quality analysis — Compass scores each feedback entry on calibrated 1–10 scales across six dimensions.
Language analysis — Extraction of sentiment, topics, subjects, and descriptor-topic pairs from the feedback text.
Entries that haven’t been analysed yet are excluded from the charts. The queue status banner on the Introduction page shows quality analysis jobs still queued or processing.
Tips
Start on the Introduction page. It explains why quality and sentiment are independent signals — high-quality feedback can be critical, and vague feedback can be positive.
NSI is a cohort signal, not an individual one. Don’t draw conclusions about a single person from NSI; it’s most reliable at segment level with enough volume.
Aggregates point, the text confirms. When a chart on Sentiment or Quality flags something, use the Research page to read the actual feedback behind it before acting.