Make your calls better with QA and the Simulation Lab
Read real calls, rehearse against difficult customers, grade what happened and feed the fixes back.
Before you start
- Some finished calls to read, or an agent ready to rehearse.
- Call QA is on Professional and Enterprise, as are the Simulation Lab and Coach. Evals are on every plan.
Read real calls first
Open Call Logs and read a few transcripts, starting with calls that went nowhere. Missed means nobody was reached, so filter on Outcome for calls someone picked up. Listen to recordings if you have them switched on.
Where:Call Logs
Turn what you found into tests
On a call in Call Logs, press Add to tests to keep what that caller said. In the agent's Evals tab you can also press New suite and write your own cases. Evals check fixed answers to set questions.
Where:Call LogsAgent Studio
Rehearse in the Simulation Lab
Pick an agent or team and run it against AI-played customers: difficult, stalling, angry or off-topic. These are text practice calls, so no number is dialled. Each is graded on your QA scorecard. Add customers of your own for the people you really meet.
Where:Simulation Lab
Know what a pass means
A pass is the AI's judgement of an AI conversation, in text, without the agent's knowledge base or tools. Treat a fail as a real problem to fix, and a pass as ready to try on a real test call.
Set up your scorecard
In Call QA, go to Setup. Scoring for shows which scorecard each agent is really graded on. Ready-made packs for collections, healthcare and admissions only add items. Changes apply to calls graded from now on.
Where:Call QA
Choose which calls get graded
In Call Settings, under AI after the call, pick Automatic, Risky calls, On demand or Off. On demand charges nothing until someone presses Grade. Automatic runs on every answered call and is charged each time. The numbers in Call QA describe only the calls that were graded.
Where:Call Settings
Close the gaps, then run it again
Intelligence lists questions nobody could answer. Answer teaches the agents using the knowledge base you pick. Change the script, save, and run the suite again. Pass rate by run shows whether it helped, and Versions lets you go back if it did not.
Costs and what to watch
- Grading, insights, translation and rehearsal can each carry a price. The price is shown before you press, and a run that did not finish was not charged.
- A low count of graded calls is not a verdict. Read the coverage line before you act on an average.
Read next
Build your first agent
Create one agent for one job, give it what it needs to know, hear it yourself, and then let real callers reach it.
Read more →Start the Telecaller Workspace
Give your staff one screen to dial customers, take hand-overs from the AI and save outcomes.
Read more →Get ready to go live
A last check before real customers hear your agent: words, hours, opt-outs, recording, money and who is watching.
Read more →