Sentiment analysis across locations
Star ratings tell you something's wrong; they don't tell you what. This playbook runs sentiment analysis on review text across every location weekly, surfacing exactly what guests are praising and complaining about - often catching a problem (like a new hire sinking service scores) weeks before it shows up in the aggregate rating.
Pull review text, not just star counts
Aggregate the actual written reviews per unit, not just the numeric rating, since the sentiment signal lives in the language.
Run topic and sentiment extraction
The AI tags recurring themes - wait times, food temperature, staff friendliness - and tracks sentiment trend per theme, per unit.
Build the weekly heatmap
A simple grid of units vs. themes, color-coded by trend direction, gets reviewed in a 10-minute weekly ops meeting.
Worth knowing
A sudden negative sentiment spike at one unit, isolated to a specific theme, is often tied to a single recent staffing change. Cross-reference with recent hire/schedule dates before assuming it's a broader operational issue.