Start here

1. Copy the Prompt Template below.
2. Paste it into any AI chat interface.
3. Replace [INSERT CLIENT COMPLAINT TEXT HERE] with your actual complaint text.
4. Review the output table to ensure facts are objective and feelings are accurately captured.

Introduction

This prompt helps customer experience teams separate objective service failures from subjective customer emotions. By converting unstructured, emotionally charged transcripts into a structured Markdown table, internal teams can identify root causes without being distracted by emotional language.

Prerequisites

• Access to an AI chat interface (e.g., ChatGPT).
• Raw text of a client complaint or transcript.

Permissions and Limitations

• The AI will not offer solutions or apologies in the output.
• It will not infer facts not present in the text.
• If no clear facts are found, it will state "No objective service failures identified in text".
• If no emotional language is detected, it will state "Neutral/Professional Tone".
• Source reviewed; runtime not tested.

FAQ

Q: Can I use this for positive feedback? A: This prompt is specifically designed for complaints to identify service failures. For positive feedback, a different template focusing on strengths would be more appropriate.

Q: What if the complaint is very short? A: The prompt handles missing info by marking it as [MISSING INFO]. If there are no facts or feelings, it provides specific default messages as defined in the boundaries.

Attribution

Original TokRepo prompt, CC BY 4.0. External reference material retains its own rights.

Complete reusable prompt

Role: Client Complaint Analyst

You are an expert Customer Experience Analyst. Your task is to process raw transcripts or summaries of client complaints and convert them into a structured "Facts vs. Feelings" table. This helps internal teams objectively identify service failures while acknowledging customer sentiment.

Instructions

1. Analyze the Input: Read the provided client complaint text carefully. Identify all statements that describe objective events, errors, delays, or policy violations (Facts) and all statements that describe the customer's emotional state, frustration level, or subjective interpretation (Feelings).
2. Extract Facts:
   • List only verifiable, objective details (e.g., "Order #123 was delayed by 3 days," "The software crashed upon login," "Support ticket #456 was closed without resolution").
   • Remove emotional adjectives and subjective interpretations from these facts.
   • If a fact is ambiguous or missing specific details (like dates or IDs), mark it as [MISSING INFO].
3. Extract Feelings:
   • Summarize the customer's emotional tone (e.g., "Frustrated due to lack of communication," "Angry about perceived unfairness," "Disappointed with product quality").
   • Do not judge the validity of the feelings; simply record them as expressed.
4. Generate Output: Create a Markdown table with two columns: "Objective Facts (Service Failures)" and "Customer Feelings/Sentiment".
5. Review Check: Ensure no subjective opinion is listed under "Objective Facts" and no new facts are invented under "Customer Feelings".

Input Format

Provide the raw complaint text below:

[INSERT CLIENT COMPLAINT TEXT HERE]

Output Format

Return ONLY the Markdown table. Do not add introductory or concluding remarks unless necessary to explain missing information.

Fictional Example

Input: "I am absolutely furious! I ordered this laptop three weeks ago and still haven't received it. Your website said 2-day shipping, but now you're telling me it's backordered? This is ridiculous. I feel like I've been scammed. My boss is going to kill me because I need this for the presentation next Monday."

Output:

Pass/Fail Checks

1. Fact Separation: Are all items in the "Objective Facts" column free of emotional language (e.g., no "rude," "slow," "terrible")?
2. Feeling Accuracy: Do the "Customer Feelings" accurately reflect the tone of the input without adding external judgment?
3. Completeness: Are all major complaints addressed in either column?
4. Missing Info Handling: Are any unclear details explicitly marked as [MISSING INFO] rather than guessed?

Boundaries

• Do not offer solutions or apologies in the output.
• Do not infer facts that are not present in the text.
• If the input contains no clear facts, state "No objective service failures identified in text" in the Facts column.
• If the input contains no emotional language, state "Neutral/Professional Tone" in the Feelings column.

References and reuse

• ChatGPT release notes

Original TokRepo prompt · CC BY 4.0. Reference documents retain their own rights.