Agent quality: review your team's conversations

Score conversations handled by people with your own rubric and track each agent's performance on a scoreboard. This is not the AI quality review.

Oct 19, 2026

Watch this step in the video · Agent quality: the rubric

Agent quality lets you review conversations handled by the people on your team, score them with a rubric and track each agent's performance. It is available on every plan and works per workspace.

This is not the AI quality review. Here you assess human agents. Scoring and feedback on AI agent replies is done from each AI agent's settings (see related articles).

Before you start

  • To see the screen and review you need access to the workspace settings. To edit the rubric you need permission to manage settings.
  • Only closed or resolved conversations with a human agent assigned are listed.
  • Nobody can review their own conversations.

1. Check the rubric

  1. Go to Settings → Agent quality and open the Rubric tab.
  2. The first time you'll find four sample criteria, each with a max score of 5: Tone and empathy (weight 2), Answer accuracy (weight 3), Response time (weight 1) and Resolution (weight 3).
  3. Click Add criterion to add your own (for example, "Correct use of templates"). Fill in the name, description (what earns the max score), Weight, Max score and order, and click Save criterion.
  4. Use Edit to change a criterion and Deactivate to leave it out of new reviews.

The weight multiplies each criterion's contribution to the total. Editing the rubric doesn't change reviews already saved, and if you delete a criterion that has been used, it is only deactivated so the history stays intact.

2. Review conversations

  1. Open the To review tab. You'll see the contact, channel, agent and close date.
  2. Click Review on the conversation you want to assess.
  3. Give each criterion a score (between 0 and its max). The Weighted total shows at the bottom.
  4. Add an optional Comment with what the agent should repeat or correct.
  5. Click Save review.

3. Track the results

  • Reviews: the history with contact, agent, reviewer, score, percentage, date and comment. Filter by agent and by dates, or click Clear filters.
  • Scoreboard: each agent's average percentage, how many reviews they have, their Best criterion and their Weakest criterion.

Tips

  • Agree with your team on what each score means before you start.
  • Review a similar sample of conversations per agent each week so the scoreboard is comparable.
  • Use the comment to give concrete, actionable feedback.

Troubleshooting

  • Nothing to review: there are no closed or resolved conversations with a human agent assigned yet.
  • You're asked to define a criterion first: activate or create a criterion under Rubric.
  • "You cannot review your own conversations.": ask another supervisor to review them.

Related articles

Was this helpful?

Related articles

Message us on WhatsApp