Automatically assign conversations to your Reviewers

Chris Jewitt avatar
Written by Chris Jewitt
Updated over a week ago

Role: Admin, Account Manager, (Workspace) Manager, Lead

Klaus gives you the possibility to automatically assign conversations to your reviewers on a regular recurring basis.

How to set up Assignments?

To set up an Assignment, you should go to Settings > Users and Workspaces, choose the Workspace, go to Assignments and click on Add assignment.

Now you can complete the following steps to set up your Assignment:


  1. Name of your assignment.

  2. Choose the interval (monthly, weekly or bi-weekly) and decide if this should be recurring.

  3. For weekly and bi-weekly assignments, it's possible to select a custom start date and day of the week. Monthly assignments will repeat on the 1st day of every month.

  4. You can choose between regular assignments where reviewers are paired with reviewees or peer-reviews where participants review each other.


Choose who your reviewers should be and who should be reviewed.

If you are doing peer reviews, the participants will reviewer each other.


The Assignment setup is flexible, so you can adapt it to your current QA process. If you want to set up an assignment based on how many reviews each reviewer must do each cycle, select Goal logic "Based on reviewers", set the number of reviews they need to do and select "in total" from the dropdown.

“Based on reviewers in total” goal logic will maximize the coverage of reviewees covered in the long term. This is an especially useful option if there are a large number of support agents and only a handful of QA people (reviwers) to cover them.

An example of a cycle:

By default, assignment will try to maximize the coverage of agents by rotating the reviewee list one by one. This behavior can be overwritten by the option “Reviews per reviewee in a row”.

If you want for all reviewers to review all reviewees in each cycle, select Goal logic “Based on reviewers”, set a goal for number of reviews each reviewer must do in each cycle and select “per reviewee” from the dropdown.

“Based on reviewers per reviewee” goal logic will ensure that each reviewee get’s a set number of reviews from all reviewers and that everyone gets an equal amount of reviews each cycle and reviewers review exactly the same set of agents.

This option works only for small teams and is not efficient for a large number of support reps because all reviewers will review the same reviewees each cycle


Create conditions to assign the right kind of conversations which should be reviewed e.g. conversations with more than 7 replies from the previous 7 days.

Finally, make sure to double check that everything is correct and then click on "Create".

"Show preview" button shows an estimation of reviewer-reviewee pairings for each cycle. Actual pairings might differ depending on conversation availability and conditions.

The list of Assignments will now show up in the Assignments menu and in the Browser Extension. If you are using the Browser Extension, then the name of the Assignment and the goal will only be displayed if you have navigated to that conversation directly from the Assignments view.

Conversations are assigned to the reviewer one at a time. Reviewers can skip the conversation if they are not suitable for a review.

When all of the assigned conversations have been reviewed, the Assignments tab will be empty until the next week/month!

The Assignments feature is available on available on legacy Professional and Advanced plan, Growth, AI suite and Enterprise.

*To ensure ineligible review content does not end up in your assignments there are minimal content requirements already at work. These are basic sanity checks that pre-qualify conversations before you set any filters. For example the conversation must have at least one public message from both an agent and a customer and the conversation must have messages with actual text (not empty or consisting of punctuations and numbers only). We also do our best to not show conversations where the only participant is a bot.

Did this answer your question?