Conversation feedback
Collect ratings and written comments from customers after each chat and use them to find knowledge gaps worth fixing.
After a conversation ends, the customer can leave feedback — a quick rating and, if they want, a written comment. Feedback gives you a steady signal on how well the AI (or your agents) are doing and pinpoints the chats that need a closer look.
What customers see
When the conversation wraps up, the widget shows a feedback prompt. The exact controls depend on how you've configured the widget:
| Mode | What the customer picks |
|---|---|
| Thumbs | A thumbs up or thumbs down. |
| Stars | A rating from 1 to 5 stars. |
Either way the customer can add an optional written comment under the rating before submitting.
Feedback is always optional. Customers who close the chat without rating it just don't show up in the feedback numbers.
Where the rating shows up
Ratings flow back into Informly in two places.
On the conversation list
The Conversations page has a feedback column you can filter on. Sort by lowest rating to see the chats that went badly, or filter to "no rating" if you're auditing response rates. See Conversations for the full list view.
On the conversation detail
Open any rated conversation and the rating and any written comment appear at the top of the sidebar, next to the customer's name. The comment is plain text — no formatting, no links — so it's safe to display as-is.
Find knowledge gaps from low ratings
Low ratings are the highest-signal way to find documents you should be writing or improving.
Filter the list
On Conversations, filter by lowest rating and the date range you care about.
Open a poorly rated chat
Read the customer's question and the AI's answer. If the answer was wrong or thin, the underlying documents probably are too.
Check the document suggestions
The detail page shows AI document suggestions — articles that would have helped. Use those as a backlog for your knowledge base. See Working with a conversation for more.
Write or update the document
Add the missing article (or fix the existing one) in Documents. Future chats on the same topic will use the new content immediately.
A small number of low ratings on the same topic is the strongest signal you can get. Two or three thumbs-down on "refunds over $500" is worth a dedicated refund-policy document.
Aggregate view
Go to Insights → Feedback for the rollup across all conversations. You'll see:
- Overall rating trend over time
- Breakdown by widget
- Breakdown by AI vs. human handoff
That's the page to use when you need a single number for a weekly report, or when you want to compare two widgets side by side.
Export feedback
From the feedback analytics page, click Export to download a CSV of every rating with its conversation ID, the customer's comment, and timestamps. The CSV is filtered by whatever date range and widget you have selected, so you can pull a clean monthly slice and hand it off to a reporting tool.