Feedback and Ratings

Understanding how users perceive your AI's responses is essential for continuous improvement. ChatBotKit's Feedback and Ratings system provides a structured way to collect, store, and analyze user sentiment data, transforming subjective feedback into quantifiable metrics that drive better AI experiences.

Traditional feedback collection often happens outside the conversation flow, making it difficult to connect user sentiment to specific responses. ChatBotKit's rating system solves this by linking ratings directly to conversations, messages, and contacts - giving you precise visibility into which interactions succeed and which need improvement.

Key Capabilities

Structured Rating Data

Every rating captures a numeric value along with optional context: the conversation where the interaction occurred, the specific message being rated, the contact who provided feedback, and the bot that generated the response. This structured approach means you can analyze feedback patterns across multiple dimensions rather than treating ratings as isolated data points.

Contact-Scoped Feedback

Enable users to provide and manage their own ratings through contact-scoped abilities. Contacts can rate interactions, view their previous ratings, and update feedback as their experience evolves. This creates a personalized feedback loop where users feel heard while you gather accurate sentiment data.

AI Agent Integration

Your AI agents can collect ratings programmatically during conversations using built-in rating abilities. Install the rating pack on your skillset, and your bot can prompt for feedback at appropriate moments, store ratings with full context, and even analyze patterns in received feedback to adjust its approach.

Flexible Association

Associate ratings with different entities based on your use case:

• Bot-Level: Track overall satisfaction with a specific AI agent
• Conversation-Level: Understand how entire conversations perform
• Message-Level: Identify which specific responses succeed or fail
• Contact-Level: Build user satisfaction profiles over time

Metadata and Reasoning

Capture rich context beyond numeric scores. Ratings support custom metadata for tagging and filtering, plus optional free-text reasoning where users can explain their sentiment. When users provide negative feedback with a reason, you gain actionable insights for improvement.

Export and Analysis

Export rating data through the API for integration with your analytics systems. Filter by time period, bot, value range, or custom metadata to extract exactly the data you need for reports, dashboards, or machine learning pipelines.

Real-World Use Cases

Customer Support Quality Monitoring

Deploy feedback collection in your support bot to understand resolution effectiveness. Each support interaction can be rated by the customer, with ratings tied to the specific conversation. Aggregate this data to identify topics where your AI excels and areas requiring additional training data or refined prompts.

Content Recommendation Tuning

When your AI recommends articles, products, or resources, collect immediate feedback on recommendation relevance. Ratings on individual message responses tell you which recommendation strategies work and which miss the mark, enabling data-driven refinement of your recommendation logic.

Agent Performance Comparison

Running multiple AI agents for different purposes or audiences? Compare rating distributions across bots to understand relative performance. Identify your highest-performing agents and apply their successful patterns to others.

User Satisfaction Tracking

Build longitudinal satisfaction profiles by analyzing contact-scoped ratings over time. Understand whether individual users become more or less satisfied as they interact with your AI, and identify at-risk users whose sentiment trends downward.

Training Data Prioritization

Use negative ratings to prioritize which conversation examples need review. When users rate a response poorly and explain why, you have a clear signal about what to improve. Export these flagged conversations for human review and dataset refinement.

How It Works

Getting started with feedback collection involves a few straightforward steps:

• Enable Rating Abilities: Add the rating pack to your skillset configuration, giving your AI the tools to create and manage ratings
• Configure Collection Points: Update your bot's instructions to describe when and how to request feedback from users
• Set Up Scoping: Choose whether ratings should be user-scoped (visible across your account) or contact-scoped (users manage their own feedback)
• Access Data: Query ratings through the API, filter by relevant dimensions, and integrate with your analytics workflow

Ratings are stored with full relational context, so you can always trace a rating back to its source conversation, message, and user. The API supports pagination, filtering by metadata, and bulk export for efficient data access.

Integration with Messaging Platforms

For Slack integrations, ChatBotKit offers automatic thumbs up and thumbs down buttons that appear after each bot response. These provide one-click feedback collection with optional reasoning for negative ratings. The same underlying rating system stores this data, so you get consistent analytics regardless of whether feedback comes from widget interactions, Slack buttons, or programmatic collection through AI abilities.

Getting Started

To enable feedback collection, navigate to your skillset configuration and add the rating abilities. Configure your bot's instructions to request feedback at appropriate moments - after completing a task, resolving a question, or at natural conversation endpoints.

For programmatic access, the Rating API provides full CRUD operations: create ratings with associated context, list ratings with filtering, fetch individual rating details, update existing ratings, and delete when needed. Export endpoints support bulk data extraction for analytics integration.

Whether you're monitoring support quality, tuning recommendations, or building comprehensive satisfaction metrics, the Feedback and Ratings system transforms user sentiment into structured data that drives continuous AI improvement.