Skip to main content

Scorecard Builder

Build and curate your own AI automated scorecards using Scorecard Builder

A
Written by Animesh Raparia

Accessing Phrase libraries

  1. To start developing the scorecards, we need to ensure the correct phrase libraries are uploaded so the phrases in Scorecard Builder can be correctly mapped. Navigate to the Phrase Library tab on the left-hand side menu.

Uploading Phrase Libraries

2. Click Upload New Library.

3. Ensure the phrases being uploaded in the .CSV format follow the same structure as the example provided.

  • Phrase group: the type/category of phrases being defined

  • Phrases: 1 or more words put together under phrase group

  • Slops: the number of spaces/gaps allowed between words to expand search possibility

  • Channel: agent or customer channel

4. Rename the Library and upload as a CSV file then hit Submit.

5. We can also upload a revised version of an existing library through clicking the Upload button. This gives us the ability to create another version.

6. We can also see the previous version history of the scorecard if required.

Accessing Scorecard Builder

7. Now let’s navigate back to the side menu and access Scorecard Builder through the Configurations tab.

Creating Aisles

8. Aisles are the different teams (queues) that are linked to the available metadata values in the system and can be mapped to different scorecards. Each scorecard we design can be individually applied per Aisle. The new scorecard builder has the capability to create an aisle so you can easily map different scorecards to different teams/queues. Click Create Aisle.

9. Name your Aisle to the relevant team in your contact centre. Apply an existing scorecard and its relevant version. Click ADD.

10. Once added, navigate back to the Aisles tab to view/modify the latest default scorecards applied to each Aisle created.

11. Click Update to implement the new scorecard to its appropriate Aisle.

12. Select a new default scorecard and its associated version from the pre-existing collection if required.

13. Now that we have finalised our Aisles, lets adjust our current scorecard collection. Click on the Scorecards tab to navigate.

14. In this tab you will see a collection of all the scorecards and their version history. Any scorecard versions will appear here. All these scorecard versions are also visible in the Aisles tab if we would like to implement a new scorecard and set as default. We also have the ability to sort the versions alphabetically.

15. Each scorecard has a versioning history attached to it. Click View History to track the previous versions made to the individual scorecard.

16. See example version history below.

17. You can view and update your existing scorecard through clicking the action icon below.

18. Or you can create a brand-new scorecard to add to our collection. Click the Scorecard Builder button.

19. After naming your scorecard in the Config Name section, navigate to Question Builder to customise the questions

20. Recall that the standard daisee scorecard has 3 categories: Communication, Compliance and Conduct. Each category has its appropriate pool of questions that are customisable.

We can click on each section to adjust their settings such as the weighting and criticality. Click the “i” button for the detailed view of each category. And create category specific questions through clicking the “+” button.

Definition of each category above:

  • Communication: How effectively the agent communicated with the caller.

  • Compliance: How well the agent complied with regulatory requirements and internal policies and procedures.

  • Conduct: How well the agent conducted themselves and represented their organisation.

21. Under the Communication section, we can create a question evaluating agents’ communication skills through clicking the “+” button.

For this example, we are assessing whether the agent provided an appropriate greeting to the caller. Rename the “average” to your desired question label. In this case, we have renamed it as “Did the agent greet the customer?”. The fields on the right-hand side determine the scorecard elements of each question.

  • Question type: a component of scorecard which determines how the question will be scored. There are six question types for the main question, and these are Average, Bauble, Binary Threshold, Measure, Phrase Search, and Script Section. The default question type is Average.

  • Weight: Represents the importance of the question when calculating the score; Weights must be replaced with a number that is greater than or equal to 0. If this field is omitted from your node, a default value of 1 will be used.

  • Criticality: Calculates the criticality of the question and will determine the auto-fail of the categories; Criticality must be replaced with a number that is greater than or equal to 0. If this field is omitted from your node, a default value of 0 will be used.

  • Evidence Description: This displays the label in the media player where instance happens.

  • Conditional: A scorecard component that is mapped to the Conditional Builder. This defines the mapping of the questions to the appropriate metadata. (i.e which questions only apply to which teams/disposition/queues)

22. Once the details in the main question were defined. Setting up of the subquestion will be next. To do that, we need to click the “+” in the main question (Did the agent greet the customer?) and redo the process as above. Rename the sub question. For this example, we named it as a “Greeting” sub question.

It has the same components as above. The only difference is the question type we desired to use. The most common type of question that are being used in the sub questions are Phrase Search and Script Section.

  • Question type: Phrase Search. A mechanism used in the sub questions that refers to a list of key words.

  • Phrase library: Refers to the uploaded phrase groups. This needs to be mapped to the libraries we have uploaded in the Phrase Library Loader (see item 22).

  • Weight: Represents the importance of the sub questions when calculating the score; Weights must be replaced with a number that is greater than or equal to 0. If this field is omitted from your node, a default value of 1 will be used.

  • Criticality: Calculates the criticality of the sub questions and will determine the auto-fail of the categories; Criticality must be replaced with a number that is greater than or equal to 0. If this field is omitted from your node, a default value of 0 will be used.

  • Evidence Description: This displays the label in the media player where instance happen.

23. After this, we need to set the version of the Phrase Library we want to map our scorecard to.

24. We can define each of the sub-question phrase groups through selecting the appropriate Phrase Library section version.

25. Once we select the correct Phrase Library version, we can then choose the appropriate Phrase Group for the sub question. For example, we are using the phrase group “Intro” from “Default Phrase Library” for the “Greeting” sub question.

26. We can then define the explanation description displayed in the scorecard when the question passes or fails.

Setting Up Conditionals

27. If you wish you apply certain scorecard questions to different teams, queues, disposition, call duration & direction, we can create a scorecard logic to set conditionals. To do that, we select the condition expression builder.

28. To add/delete a conditional Click Add Node or Delete Node.
To expand/minimise an existing conditional in an existing scorecard. Select the “+” on the side (e.g the “+” or “-“ button before “Default”) it will prompt to the custom metadata values that are being used and the conditional logics that are set up on the values.

The expression types include:

  • Equals is used if there is only one metadata value you would like to include in the contiditionals.

  • any_of is used if there are multiple metadata values to include in the conditional.

  • call_duration is used to exclude or include questions in the scorecard specific to the duration of the call.

  • And is an expression that needs to pass multiple conditions in the scorecard.

  • Or is an expression that picks any of the multiple conditions in the scorecard.

  • Not is an expression which is used to exclude single or multiple conditions in the scorecard.

  • Custom Metadata is a metadata that is specific to the customer about the call.

  • Direction the call direction, which can be one of two values, outbound or inbound.

  • Queue the queue the call took place on, this can be any text value.

29. As an example, we can use a custom metadata value such as “Scorecard” to define which scorecard values will be used in the conditional.

30. We can then select the “Scorecard” values to pick the exact values we want the questions to be applied.

Did this answer your question?