Autoscoring Overview
Autoscoring provides automated assessments of agent performance. It evaluates interaction content to track script compliance, measure quality management criteria, and calculate quality scores across a large volume of interactions without requiring manual human review.
How does it work?
Administrators configure evaluation forms containing specific criteria organized into areas. Autoscoring evaluates the portions of an interaction where an agent is engaged in conversation with the customer. A service may have multiple Autoscoring form assignments. At runtime, the system reviews the assignments from top to bottom and applies the first form whose assignment filters match the interaction segment. Administrators can also configure the percentage of eligible interaction segments processed by Autoscoring. Each service can use the default coverage percentage or a custom coverage value.
These forms use "Intents" to act as yes-or-no questions based on the conversation's content. Intents are extracted using either predefined keyword grammar or AI provider analysis.
When the system processes an interaction transcript, it checks for the presence of these intents. Each intent has a configured positive or negative weight, which the system uses to automatically calculate an overall score for the interaction. Users can view aggregated scoring data and trends in Autoscoring Analytics dashboards, or review individual score breakdowns within specific interaction records.
Use Cases and Benefits
- Assess agent performance automatically.
- Measure agent performance consistently across interactions.
- Track script compliance across interactions.
- Prevent evaluator bias in quality management.
- Prioritize specific interactions for manual review based on pass or fail criteria.
Dependencies
- Interaction transcripts are required.
- For voice services, call recording must be turned on globally or at the service level.
- For voice services, either real-time or historical Speech-to-Text transcription must be configured for the service.
- For chat services, no additional configuration is required.
- If using AI Intents, an AI provider must be configured.
< Previous | Next >