Docs/Daily Use/Browsing Conversations

Browsing Conversations

Read every conversation your AI had with a shopper. See exactly what it recommended, where it helped, and where it could improve.

Read Every Word the AI Said#

The Conversations page in your dashboard shows every chat your AI has had with a real shopper. You can read the full transcript of each one — what the customer asked, what the AI replied, which products it recommended, and whether the customer ended up buying.

The Conversations page showing a list of real shopper chats with date, outcome (No sale / Purchased), rating, and message count columns. A search bar and All / Converted-only filter pills sit at the top.
Screenshot: The Conversations page on a real dashboard. Each row is a real chat — search by keyword, filter to Converted only, and click any row to see the full transcript.

Why This Exists#

Most chatbot tools hide conversations behind aggregate metrics. We built the opposite — full transparency, because shop owners need to trust what the AI is saying about their brand.

Three reasons to actually read them:

  • Verify quality. Random-spot-check a few conversations a week. If something feels off, you'll catch it early.
  • Find product gaps. When shoppers ask for things you don't have, you'll see exactly what wording they use. Add those products. Or block those topics.
  • Improve your copy. The questions shoppers ask are the questions your product pages should already be answering. Steal the wording.

What Each Conversation Shows#

Click any conversation in the list to open the transcript. You'll see:

  • The full chat history — both shopper and AI messages, in order, with timestamps.
  • Product cards — exactly which products the AI recommended, with the prices and links it showed.
  • Outcome — whether the conversation led to an order (direct or assisted attribution).
  • Customer satisfaction — if the shopper rated the response, you'll see thumbs up / down.
  • Tags — auto-detected sentiment, intent, and topic.

The list page lets you filter by:

  • Date range — past 7 days, 30 days, custom.
  • Outcome — only conversations that converted, only ones that didn't, all.
  • Sentiment — happy, frustrated, neutral.
  • Page-numbered pagination — handles thousands of conversations without breaking.

What to Do If You Spot a Bad Answer#

Two options, depending on how you want the AI to behave next time:

  1. One-off correction. Open the conversation. Look for the question that got the bad answer. Take note of it.
  2. Permanent fix via Knowledge Gaps. If you see the same bad-answer pattern across multiple conversations, head to the Knowledge Gaps page and resolve it once. The AI applies your correction to every future similar question. See Knowledge Gaps.

Privacy Note#

Conversations are stored to power your dashboard analytics, train your store's memory, and help you improve. Customers are not personally identified — we don't capture names, emails, or payment info from chats. Only the conversation content + the order ID if they checked out.

For more on data handling see Data Privacy.

Was this page helpful?
Updated April 2026

Related reading