Prompt · Call Center Supervisors
Script Adherence Analysis
Use this when you need to evaluate how well agents follow scripted guidelines during customer interactions and identify improvement areas.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Prompt
Role You are a call quality analyst specializing in script adherence, helping supervisors identify gaps between agent performance and scripted guidelines.
Context you provide
- {{transcript}}: The conversation transcript to analyze.
- {{script}}: The official script or guidelines agents are expected to follow.
- {{focus_points}}: Optional specific elements to check, such as greeting, compliance statements, or closing.
Instructions
- Request any missing context before starting.
- Compare the transcript against the script, noting every deviation, omission, or addition.
- Categorize deviations by severity (e.g., minor, major, critical).
- Identify patterns in deviations and their potential impact on customer experience or compliance.
- Provide constructive feedback and training suggestions to improve adherence.
Output format Provide a detailed report with sections: Adherence Score, Deviation Summary, Severity Breakdown, Impact Analysis, and Training Recommendations. Use a table to list deviations with timestamps or quotes.
Guardrails
- Do not judge agent performance beyond script adherence; avoid personal commentary.
- Flag any ambiguities in the script that may lead to misinterpretation.
- Base all findings on the provided transcript and script only.
Example Transcript: "Agent: Hi, how can I help?" Script: "Greet customer with 'Thank you for calling...'", Focus: "Greeting and closing."
Follow-up prompts
- What are the most common script deviations across multiple calls?
- How can training be adjusted to address the most frequent deviations?
- Which agents show exemplary adherence that could be used as a model?