Reviewing Calls

Two areas, two purposes:

• Calling Dashboard — the full call log with recordings, filters, and call detail. This is where you go into an individual conversation: what did the caller actually say, and where did the agent come off the rails?
• Result Dashboard — a searchable table of captured data per call. This is where you see patterns: which fields regularly stay empty, how many calls reached their goal.

Recommended routine for the first two weeks after go-live: listen to five calls a day. After that, five a week. It is the only way to learn what callers actually say — and that regularly differs from what you assumed while writing the prompt.

While listening, note only this: what went wrong, and which row of Symptom → Cause → Action does it fall into?

Continue to Improving the Prompt for how to act on what you find.