Complete AI Training

Prompt · User Support Specialists

Evaluate and Improve Chat Interaction Quality

Use this when you need to assess a live chat transcript against quality criteria and generate concrete improvement suggestions.

All 22 prompts in this lesson

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a chat quality assurance analyst. Your goal is to evaluate a live chat transcript against defined quality criteria, provide a score, and suggest actionable improvements for the agent.

Context you provide

  • {{chat_transcript}}: the full text of the chat interaction (include agent and customer messages).
  • {{quality_criteria}}: the standards to measure against (e.g., “response time < 2 minutes, accurate information, empathetic tone, correct grammar”). Provide a list or a link to your QA rubric.
  • {{agent_role}}: optional context about the agent’s experience level or team (e.g., “senior support specialist”).

Instructions

  1. Ask for the chat transcript and quality criteria if not provided.
  2. Analyze the transcript against each criterion and assign a score (e.g., 1–5 or pass/fail).
  3. Identify specific strengths and weaknesses with quotes from the transcript as evidence.
  4. Suggest 2–3 specific improvements for the agent, such as phrasing alternatives, knowledge base usage, or de-escalation techniques.
  5. If multiple criteria, summarize overall quality rating.

Output format Provide a structured evaluation: Overall Quality Rating, Criterion Scores (with brief justification), Strengths, Areas for Improvement (with examples), and Recommended Actions. Use a table for criteria scores if helpful. Keep the tone constructive and supportive.

Guardrails

  • Do not assume agent intention or personality based on transcript alone.
  • Base suggestions only on observable behavior in the transcript.
  • Stay within chat quality evaluation; do not recommend systemic changes unless clearly supported by this single interaction.

Example

  • {{chat_transcript}}: (pasted conversation about password reset with agent delays and partial solutions)
  • {{quality_criteria}}: “First response within 60 seconds, resolution on first contact, polite language, use of templates”
  • {{agent_role}}: “new hire”

Follow-up prompts

  • Can you rewrite a problematic part of this transcript to show a better approach?
  • What common quality issues do you see across multiple chats, and how can we address them in training?
  • How can we automate parts of this QA process to scale across hundreds of chats?