Scorecards
Scorecards allow QA reviewers to score interactions against a structured scorecard, with each question answered by AI based on the call transcript. Every score is linked directly to what was said, giving reviewers clear, evidence-backed results to work from.
This article covers creating, managing, and applying scorecards within Conversation Analytics.
Creating and Editing Scorecards
When you select to create a new scorecard or edit an existing one, you'll be taken to the Scorecard Builder. Here you can build scorecards tailored to your requirements, which can then be applied to transcribed calls within Conversation Analytics.
Scorecard Details
Scorecard Status. Set the scorecard as active or inactive. Inactive scorecards will not be available to apply to interactions.
Scorecard Name. A required field, visible to users within Management Hub. Must be between 1- 200 characters.
Description. An optional field to help with managing scorecards and understanding their purpose. Visible to users within Management Hub. Can be up to 500 characters.
Pass Rate. The minimum percentage required to be classed as a pass, based on the scorecard questions.
Business Context. The context provided here is used by the AI to evaluate calls and grade them against the scorecard. This is effectively the prompt the AI will use when assessing each interaction, so providing clear and accurate context will directly affect the quality of the scoring.
Sections and Question Types
Sections. Sections allow you to group and split questions within your scorecard into logical categories, making it easier to structure and review results.
Question Types. The following question types are available when building a scorecard:
| Question Type | Description |
| Yes or No | A definitive yes or no question. Set the score to be applied if the answer is yes. |
| Rating | An adjustable scale of 1-10. Used to rate how well an agent performed a particular behaviour. Set a minimum score required to pass the question. |
| Observations | Generates and collates comments, opinions, and observations on the transcribed call. Please note that observations do not contribute to the overall score. |
Within each question type, you'll also find:
Evaluation Criteria. Automatically generated from the question and business context provided. This is used by the AI to evaluate and score the question. You can review and edit this as required.
Auto Fail. When enabled, if this question fails, the entire scorecard will be marked as failed, regardless of the overall score.
Applying a Scorecard to a Transcribed Call
Within the Interactions tab, locate the call you wish to score. Interactions that have already been scored are indicated by a Scored label next to the Interaction ID. Hovering over this label will display an overall summary and the scoring result.
Select an interaction to open the side panel. On the right-hand side, switch from the AI Assistant tab to the Scorecards tab. Select the scorecard you wish to run against the transcribed call and apply it.
Please note this may take a moment, as the AI analyses the transcript against the scorecard and scores each question accordingly to produce the final result.
Once the evaluation is complete, the full scorecard will be displayed, showing the overall score, pass or fail result, and a detailed breakdown against each question.