How does sentiment analysis work?

Simple Chat scores analyzed customer messages from 1 to 5, with lower scores indicating more negative sentiment. The analysis also identifies topics and product mentions.

Reading the report

Open Reports > Sentiment, choose a date range, and use customer filters to narrow the results. Only analyzed messages with valid sentiment scores contribute to the totals.

Low scores can help you find conversations worth reading, but they don't explain the cause on their own. A frustrated message might concern the product, a delayed shipment, or a misunderstanding about your policy.

Checking a pattern

1. Find a product or topic associated with low scores.
2. Open several linked conversations and read the customer's wording.
3. Identify the issue your team can address, such as unclear delivery expectations.
4. Compare a similar period after making the change.

Treat the scores as AI assessments, not customer survey responses. Read individual complaints before making decisions about an order or customer.