Customer Feedback Categorizer

---
title: "Customer Feedback Categorization System"
emoji: "📊"
category: "Customer Service"
author: "Siya"
created: "2024-11-05"
---

## Overview

This prompt creates a structured system for analyzing and categorizing customer feedback based on sentiment and topics. It helps maintain consistency in feedback analysis and provides clear justification for categorizations.

## Key Features

- Sentiment analysis (Positive, Negative, Neutral)
- Topic categorization (Product, Service, Support)
- Structured XML output format
- Built-in justification requirement
- Example-based learning

## Prompt Template

<Prompt content={`You are an AI assistant tasked with categorizing customer feedback. Your job is to analyze the given feedback and categorize it based on sentiment and topic. Here's how to proceed:
1. First, you will be presented with a piece of customer feedback:
<customer_feedback>
{{CUSTOMER_FEEDBACK}}
</customer_feedback>
2. Analyze the sentiment of the feedback. Determine if it is:
   - Positive
   - Negative
   - Neutral
   Consider the overall tone, language used, and any explicit statements of satisfaction or dissatisfaction.
3. Identify the main topic(s) of the feedback. The primary categories are:
   - Product: relating to the features, quality, or performance of the product
   - Service: relating to customer service, delivery, or overall customer experience
   - Support: relating to technical support, troubleshooting, or assistance
   Note that a piece of feedback may touch on multiple topics. In such cases, identify all relevant topics.
4. After your analysis, present your categorization in the following format:
<categorization>
<sentiment>[Insert sentiment here: Positive, Negative, or Neutral]</sentiment>
<topics>
<topic>[Insert first identified topic]</topic>
<topic>[Insert second identified topic, if applicable]</topic>
</topics>
<justification>
[Provide a brief explanation for your categorization, referencing specific parts of the feedback]
</justification>
</categorization>`} />

## Variables

The prompt contains one main variable:

- `{{CUSTOMER_FEEDBACK}}`: The customer feedback text to be analyzed

## Example Usage

Here's an example of how to use this prompt:

<Prompt content={`<customer_feedback>
I love my new smartphone! The camera quality is amazing and the battery lasts all day. However, I had some trouble setting it up and customer support wasn't very helpful.
</customer_feedback>
<categorization>
<sentiment>Positive</sentiment>
<topics>
<topic>Product</topic>
<topic>Support</topic>
</topics>
<justification>
The feedback is primarily positive, with the customer expressing love for the product and praising specific features (camera and battery). However, there is a negative aspect related to the setup process and customer support, which is why "Support" is also included as a topic.
</justification>
</categorization>`} />

## Optimization Tips

1. **Clear Feedback Text**: Ensure the customer feedback is clearly formatted and complete before insertion into the prompt.

2. **Multiple Topics**: Remember that feedback can be categorized under multiple topics - don't limit to just one if multiple are relevant.

3. **Balanced Analysis**: When providing justification, reference both positive and negative aspects if present.

4. **Consistent Format**: Always maintain the XML structure for easy parsing and standardization.

## Best Practices

- Always include specific references from the feedback in the justification
- Consider the overall context when determining sentiment
- Be thorough in topic identification
- Maintain consistent XML formatting
- Provide clear reasoning in the justification section

## Use Cases

- Customer service feedback analysis
- Product review categorization
- Support ticket triage
- Customer satisfaction monitoring
- Feedback trend analysis
- Quality assurance reviews

## Technical Details

The prompt uses a structured XML format for output, which makes it ideal for:

- Integration with automated systems
- Database storage
- Analytics processing
- Report generation
- Trend analysis

## Limitations

- Limited to three main topic categories
- Three-tier sentiment classification may not capture nuanced feedback
- Does not account for mixed sentiment within individual topics