Skip to content
Supportqutee.aiOpen app

Check your agent with simulated test calls

About 20 simulated calls show you before going live how your agent handles typical requests and which setting you should adjust.

Before real callers reach your agent, you can check it with simulated test calls. qutee plays through about 20 typical calls: questions about your topics, transfers, callback requests, appointment requests, calls outside your hours and difficult conversations too. Afterwards you see which calls went well and where a setting does not fit.

  • You may edit the agent.
  • Test calls are available for agents that take incoming calls.
  • The Test calls tab comes together with the Quality tab, depending on your plan.
  • How many runs you can start per day also depends on your plan.
  1. Open Agents and choose your agent.
  2. Open the Test calls tab.
  3. Click Start test calls.

A run takes about three minutes. While it runs, you see how many test calls are done. You can leave the page in the meantime; the result appears here later.

The agent is tested as it is set up right now. No real calls are made: they do not count as calls, no emails go out and no data goes to connected systems.

At the top you see a summary with the number of test calls per result, for example “17 passed” and “3 failed”. Below it, every test call is listed with its result:

  • Passed: the agent handled the request the way your settings intend.
  • Failed: the problem happened on the second try too. You see what went wrong and a recommendation.
  • Inconsistent: the first try went wrong, the second did not. Read the conversation if the request matters to you.
  • Aborted: the test call could not be finished. Start the test calls again later.

Click a test call to read the whole conversation between caller and agent.

For a failed test call, qutee shows you which setting is behind it, for example a knowledge topic that contradicts another one, or a transfer without a phone number.

  1. Click Open in editor on the test call. The matching place of your agent opens.
  2. Adjust the setting and save.
  3. Start the test calls again to see whether the problem is fixed.

Confirmed problems from test calls are also listed under Quality, with the source Test call. View in Quality takes you there. When a later run passes the same case, the finding there is marked as resolved.

Test calls are simulated as text. They do not check:

  • voice, pronunciation and speech recognition,
  • real transfers to phone numbers,
  • your real calendars (a free slot is assumed for appointment requests),
  • emails and connected systems after the call.

The Start test calls button is greyed out: either a run for this agent is still going, or today’s runs are used up. Below the button you see how many you can still start today.

A test call failed although the setting is right: an AI plays the callers, and an AI judges the calls too. Read the reason and the conversation. If it says the settings are clear and the agent still reacted wrongly, the case does not appear under Quality.

Was this page helpful?