Choose a scorecard and run the test

A scorecard defines what a successful conversation looks like. Choose relevant checks so the result measures your actual goal.

Navigation pathAgent Testing → New test → Scorecard → Review

PrerequisitesAn agent and persona selected in the first two steps.

Usage steps

Choose one or more Template criteria, a Saved scorecard, or Create custom scorecard.
For example, select checks for capturing contact information, responding to an objection or closing with clear next steps. Avoid grading actions the agent is not meant to perform.
Select Continue. On Review & run, review Test name, Agent, Persona, Grading on and Max duration.
Use the edit controls or Back to correct the scenario before starting it.
Select Run test when ready. Return to the scenario list to check the last-run status and open the scenario to inspect its available results.

Choose a scorecard and run the test — Figure 20. Step 3 offers individual checks and saved scorecards; Step 4 reviews the complete test before Run test.

Expected result. A test run is created with the chosen evaluation criteria.

Practical notes and troubleshooting. A run marked Failed may indicate execution failure. Read the run details before treating it as evidence that the agent failed the behavioural check.

Configure call analysis and scorecardsCall analysis classifies conversations into outcomes. Structured outputs extract data, while scorecards evaluate quality. Configure each for a distinct purpose.Create call analysis criteriaCall Analysis Criteria defines the rules used to recognise outcomes in a conversation. Clear, non-overlapping descriptions make classifications easier to review.