How can we help?

The Random Sampling feature randomly picks samples of calls handled by agents so that supervisors can evaluate and/or review them weekly or monthly. 

This feature is available to all Talkdesk Quality Management™ customers and reduces customer effort by:

• Randomly picking and choosing which calls should be selected for evaluation. 
• Automatically selecting calls according to the rules defined by the users and placing them ready to be evaluated in the Quality Management (QM) application.
• Setting evaluation goals so that supervisors may meet their goals/quota for the week or month.

 In this article, you will find information on:

• Creating a Random Sampling Rule 
• Viewing Random Sample Results

Creating a Random Sampling Rule 

To create a new sampling rule to pick up calls to be evaluated, please follow the steps below:

1. Once you have logged in to your Talkdesk account, go to "Quality Management" > "Assign Evaluations" section > Random sample tab [1].

2. Click on the Create sample button [2].

3. On the “Create random sample” modal, insert a “Name” [3] for the random sample. Optionally, you can also add a “Description” [4]. 

4. Press the Continue button [5]. If you want to cancel the creation, click Cancel [6].

5. In the “Required Settings” [1] section, define the configurations for this new call sampling rule:

• “Evaluation form” [2]: Select an evaluation template from the dropdown menu.
• “Evaluators” [3]: Select one or more available evaluators from the dropdown menu.
• “Pick up agents from” [4]: Choose your agents from the “Queue” (Ring Group) or “Team” options.
• “Queues” [5]: Select the queue(s) and then the available “Agents” [6] from the dropdown menu.

Note: Please be aware that the evaluator will not get any randomly selected calls to evaluate if you do not select an agent from the evaluator's team/queue.  

• “Include AI pending evaluations to pick up” [7]: Define if the AI Pending evaluation has priority when doing the distribution [8], or if the system should first pick interactions that have not been picked by the AI [9]. Note: This option is only available when using an “AI enabled” evaluation form.
• “Distribute calls by” [10]: Choose if you wish to distribute calls by “Number” or “Percentage”.
• “Number”[11]: Choose if you want to distribute calls every “week” or “month” [12], and the number of calls each agent can receive [13].

• “Percentage” [14]: Insert the percentage of calls from the previous day [15], that is going to be attributed to the “Evaluators” selected. 

Notes:

• If there is more than one rule applied to pick calls by percentage, and the agents selected to have their calls sampled are in the rules, then, the overall percentage will be the sum of those rules (e.g. if an agent is in one rule that has 5%, and the same agent is in another rule that has another 5%, then, 10% of that agent's calls will be sampled).
• You can choose a minimum of 0% and a maximum of 100% of the calls, and only whole numbers are accepted (e.g. 35%, 91%, 26%, and so on).
• The total number of “Evaluators” appears by default [16], and changes according to the selection made on the “Evaluators” field.

• “Due date to finish evaluations” [17]: Choose if you want to finish the evaluations until the “Last day of each month”, “Last Friday of each month”, or on a “Specific day”.

6. In the “Optional Settings” section [18] you can define the following:

• “Calls duration” [19]: Insert the “Minimum” and “Maximum” call duration.
  • If you select a duration range, the random sampling does not pick out calls within the range you defined. 
• “Only pick up calls from” [20]: If you only want to define a period of picked up calls from the previous week or month. 
• “Delete incomplete evaluations after” [21]: In case you select this option, the incomplete evaluations (“To do” and/or “Draft” evaluations) originated from this random sampling rule will be automatically removed from the “Evaluations” list, after the amount of days chosen is reached.

At the bottom of the page, the “Overview” section [22] informs about the current configuration status.

7.  When you finish defining the required settings and the optional ones you want to set, click on the “Publish” button.   

After creating a new sample, it will be “Inactive” by default. To activate it, go to the "Assign Evaluations" section > Random sample tab > Click on the toggle [1] next to the name of the random sample you created. After activating it, a success message [2] appears to inform you that the random sample rule is updated, and the changes will be applied on the next run.

Notes: 

• The system does not pick up calls from inactive agents or calls that have already been evaluated.

• It is possible to edit a sample configuration name and description after the creation and edit an active random sample by clicking on the pencil icon [3]. When it is an active rule, the changes are only applied on the next run.

• The system will pick up calls randomly from the available calls, if the Optional “Only pick up calls from” setting is defined, then it will only use this time period, and distribute as many calls as possible to achieve the goal.

• When a rule is enabled for the first time or re-activated (e.g. after an edit), it will run nightly (time might vary due to different time zones).

Once you publish a Random Sample, you can select its “Name” [1] to view more details, such as:

•  “Goal” [1]: Rule goal. 
• “Last run” [2]: Date/time of the last run.
• “Assignment progress” [3]:  Total interactions already assigned out of the intended goal.
• “Available interactions” [4]: Total of interactions available from the last run, according to the rule criteria. 
• “Assigned for evaluation” [5]: Interactions selected to be distributed in the current run. Each day it runs, it will display a new value.
• “Next run” [6]: Date/time of the next run.

Viewing Random Sample Results

After publishing a Random Sample, it will appear on the “Evaluations” page of the evaluator with the “To Do” evaluation status [1]. Agents will not see the “To do” evaluations until they are published. 

The behavior of this status is similar to the “Draft” one, but no questions are filled out until an evaluator starts to fill them. The interaction is then moved to the “Draft” status.

Note: The “To Do” evaluations cannot be deleted directly. When edited, they go to “Draft” status and if the draft is discarded, they go back to “To Do”. So, the supervisor needs to submit the evaluations first, before they can be deleted.