Testing & Coverage
Test Coverage helps you validate your agent before changes go live. It shows what has been tested, what passed or failed, and what still needs attention.
Where to test
Open Test Coverage from the sidebar. You can review coverage by:
- Topic and intent
- Flow path
- Benchmark sets
Use the agent filter to focus on one agent at a time.
Creating scenarios
A scenario represents a customer request you want to test. When you create a scenario, you can include:
- The customer message or conversation input.
- The topic and intent it should match.
- Expected behavior (pass or fail).
You can create scenarios one at a time or import them from CSV for faster setup.
Benchmarks
Benchmarks are named sets of scenarios you want to run together. Use them for repeatable tests like regression checks or launch readiness.
Test vs. live conversations
Tests are marked as internal so they do not affect production analytics.
| Type | Purpose |
|---|---|
| Test | Validate quality and gaps |
| Live | Real customer support |
Quick workflow
- Create scenarios for your top topics and intents.
- Run tests and mark outcomes as pass or fail.
- Fix the gaps in Responses or Flows.
- Re-run the same scenarios to confirm improvements.
You can also mark priority intents so they are easier to track over time.
Best practices
- Cover your highest volume topics first.
- Include edge cases that should escalate.
- Re-test after any major change.
If you use benchmarks, group scenarios into named sets so the team can re-run them consistently.