← All resources Versión en español
Drellia · Template

Call QA questionnaire template

A starting point for evaluating customer calls, whether your reviewers are people or AI. Keep the questions that fit your process, rewrite the rest, and add the disclosures your sector requires.

Questionnaire name
Team or department
Owner
Version and date

How to read each question

A. Opening and identification

#QuestionTypeCompliant answerAlertApplies when
1Did the agent greet the customer and say their name and the company's name?Yes/NoYesNoEvery call
2Did the agent tell the customer that the call is recorded?Yes/NoYesYesWhere a recording notice is required
3Did the agent verify the customer's identity before discussing account details?Yes/NoYesYesAccount details were discussed

B. Understanding and resolution

#QuestionTypeCompliant answerAlertApplies when
4Did the agent confirm they understood the reason for the call, for example by summarizing it back?Yes/NoYesNoEvery call
5How was the customer's request handled?Multiple choice: Resolved on the call · Escalated · Follow-up scheduled · Not resolvedResolved on the call, Escalated, Follow-up scheduledNoThe customer made a request
6Was the information given about products, fees or terms consistent with the approved script or documentation?Yes/NoYesYesProducts, fees or terms were discussed

C. Compliance and conduct

#QuestionTypeCompliant answerAlertApplies when
7Did the agent give every mandatory disclosure for this type of call? (List them in the question, e.g. purpose of the call, cost, cancellation rights.)Yes/NoYesYesThe call type has mandatory disclosures
8Did the agent avoid pressure, threats or misleading statements?Yes/NoYesYesEvery call
9Did the agent avoid asking the customer to say data that should not be spoken on the call, such as full card numbers or passwords?Yes/NoYesYesEvery call
10When the customer complained or asked not to be contacted again, did the agent acknowledge it and explain what happens next?Yes/NoYesYesThe customer complained or opted out

D. Closing

#QuestionTypeCompliant answerAlertApplies when
11Did the agent summarize the agreed next steps?Yes/NoYesNoNext steps were agreed
12Did the agent ask whether the customer needed anything else before closing?Yes/NoYesNoEvery call

E. Context (not scored)

#QuestionTypeCompliant answerAlertApplies when
13In one or two sentences, why did the customer call and how did it end?Open—NoEvery call
14What was the customer's mood at the end of the call?Multiple choice: Positive · Neutral · Negative—NoEvery call

Writing questions that are evaluated consistently

  1. One behavior per question. "Did the agent verify identity and explain the fees?" hides which part failed. Split it.
  2. Make it observable in the conversation. A reviewer, or an AI reading the transcript, must be able to point to the moment that answers it.
  3. Define the compliant answer before you start. If two reviewers would disagree on what "good" looks like, the question needs rewording.
  4. Say when it applies. Otherwise calls where the situation never came up are counted as failures.
  5. Use open questions for context, not for scoring. Summaries are useful, but pass-or-fail checks belong in yes/no or multiple-choice questions.
  6. Ask for evidence. Whoever answers should be able to point to the moment in the call that supports the answer, so the result can be checked.
  7. Calibrate before you rely on it. Score the same set of calls with two reviewers (or a reviewer and the AI), compare, and clarify the questions where they differ.

This structure (question type, compliant answers, which answers raise an alert, and when each question applies) is the same one Drellia uses to evaluate calls and chats automatically. See drellia.com/resources for more templates.

Template provided by Drellia as general guidance. It is not legal advice and does not guarantee compliance with any law or regulation. Adapt it to your processes and confirm the requirements that apply to you with your compliance or legal team.