Theme
Answer Feedback
Answer feedback lets you rate the overall quality of an AI-generated suggestion. Your ratings help the system learn which types of answers are useful and which need improvement.
Rating an Answer
Suggestion card footer with thumbs-up and thumbs-down feedback buttons
Each suggestion card has two feedback buttons at the bottom: thumbs up and thumbs down.
Thumbs Up (Positive)
Click the thumbs-up button to indicate the answer was helpful. The button highlights to confirm your rating. No additional input is required.
A positive rating signals that:
- The answer was relevant to the customer's question.
- The content was accurate and well-structured.
- You were able to use the answer (directly or adapted) in your response.
Thumbs Down (Negative)
Click the thumbs-down button to indicate the answer was not helpful. This opens a feedback modal where you select a reason and optionally provide a comment.
Feedback modal with reason code dropdown and optional comment field
Reason Codes
When you give a thumbs-down rating, select one of the following reason codes:
| Reason Code | When to Use |
|---|---|
| Irrelevant | The answer does not relate to what the customer asked. The system retrieved content about the wrong topic entirely. |
| Incorrect | The answer contains factual errors. The information is wrong, even though it may be on the right topic. |
| Too Generic | The answer is vaguely related but does not provide enough specific detail to be useful. It reads like a generic response rather than a targeted answer. |
| Misleading | The answer could be interpreted in a way that leads the agent or customer to the wrong conclusion. The content is technically present in the sources but presented out of context. |
Select the reason code that best describes the issue. If multiple apply, choose the most significant one.
Comment Field
Below the reason code selection, an optional comment field lets you provide additional detail. Comments are especially valuable when:
- The reason code alone does not capture the specific problem.
- You can describe what the correct answer should have been.
- You noticed a pattern (for example, the same wrong answer appearing repeatedly).
TIP
Even a brief comment like "answer referenced the wrong product line" gives administrators actionable context when reviewing feedback reports.
Removing Feedback
If you rated an answer by mistake, click the same button again to remove your rating:
- Click a highlighted thumbs-up to remove your positive rating.
- Click a highlighted thumbs-down to remove your negative rating and dismiss the reason.
The card returns to an unrated state.
What Happens with Feedback Data
Your feedback is stored against the specific suggestion and its associated conversation. The data is used in three ways:
- RAG tuning. Aggregated feedback helps the system adjust retrieval parameters and re-rank sources for similar queries in the future.
- Analytics dashboards. Administrators see feedback trends in the Feedback Analytics view, including the most common negative reason codes and which knowledge bases have the lowest satisfaction rates.
- Knowledge gap identification. Patterns of negative feedback on specific topics signal to administrators that the knowledge base needs new or updated content.
WARNING
Feedback is tied to your agent profile. Administrators can see which agents provided feedback, but this data is used for quality improvement -- not performance evaluation.
Next Steps
- Learn how to rate individual sources: Source Feedback
- Return to the feedback overview: Feedback