Back to Resources
Guide

How Our AI Analyzes Your Feedback

Learn how sentiment, intent, emotion, urgency, and drivers are detected.

The AI Engine

We use a multi-stage analysis pipeline designed specifically for customer feedback (B2B and B2C).

1. Sentiment Analysis

We don't just look for "good" or "bad" words. We look at context.

  • "The app is fast, but the support is slow."
  • Result: Mixed Sentiment. Positive for "App Performance", Negative for "Support".

2. Driver Extraction

This is our propriety logic. We identify the noun or concept that is driving the sentiment.

  • Comment: "I can't believe how expensive the enterprise plan is."
  • Driver: "Enterprise Pricing" (Negative).

3. Urgency Detection

We categorize every comment into specific urgency levels:

  • Critical: churn risks, legal threats, severe bugs ("I'm cancelling", "Data loss").
  • High: broken core features, login issues.
  • Medium: feature requests, minor bugs.
  • Low: general praise, minor UI feedback.

4. Summarization

For every question in your survey, we generate a "TL;DR" (Too Long; Didn't Read).

  • Instead of reading 100 comments, read: "Users generally like the new dark mode, but 15% report contrast issues on the settings page."

Privacy & Security

Your data is processed in a secure environment. We do use LLMs (like OpenAI) to process text, but check our Privacy Policy for details on data retention and training (we don't train on your data).