Test and monitor Quinn

Test Quinn with real customer questions before launch, then track resolutions, escalations, AI CSAT, action outcomes and Copilot usage in Admin > Analytics.

Once Quinn is live, check how it performs with real customers and how much your team uses Copilot. This article covers testing Quinn before launch and the Quinn reporting in Analytics.

Test before you launch

Before you rely on Quinn with customers, try it yourself:

Make sure the Agent is on in Admin > Settings > Agent and the Messages tab is on in Admin > Settings > Widget.
Open your widget on your site or portal and ask the questions your customers actually ask.
Pull a few real transcripts from your inbox and replay the customer's first message.
Check the answers against your help center. If an answer is wrong or missing, fix the article first, then adjust Guidance if Quinn still behaves the wrong way.

Tip: Ask at least one question Quinn shouldn't be able to answer, and confirm it hands off to your team instead of guessing.

Quinn analytics

Open Admin > Analytics and go to the Quinn section. Pick a time period at the top, the same as the other analytics sections.

Outcomes

How the customer-facing Agent handled the conversations it joined:

Conversations involved: how many conversations Quinn took part in
Resolved: split into confirmed resolutions (the customer said so) and assumed ones
Escalated: how often Quinn handed off to a human
Pending: conversations Quinn is still working on
AI CSAT: satisfaction ratings on conversations Quinn handled
System errors: turns that failed for a technical reason

Actions

How often each action was attempted, completed, or failed.

Copilot

How your team used Copilot over the same period. If nobody used it, the section says there was no Copilot activity.

Improve from what you see

High escalation rate: look at the handed-off conversations. Missing articles are the most common cause.
Low AI CSAT: read the low-rated conversations and add Guidance rules for the patterns you find.
System errors: on self-hosted installs, check that your AI endpoint and model settings are valid.