Interview Analysis

Yazi AI Analysis Product: Complete Explainer

Overview

Yazi's AI Analysis transforms raw interview data into structured, actionable insights through an intelligent 4-phase categorization system. The product automatically analyzes interview conversations, categorizes questions and answers, and presents findings through an interactive interface that allows users to explore patterns, drill down into specific insights, and validate AI classifications.

How It Works: The 4-Phase Analysis Process

Phase 1: Question Category Generation

When: Automatically triggered once 20-30 questions have been collected Purpose: Create meaningful question categories that reflect the research objectives

Process:

• AI analyzes the research brief, survey questions, and sample of actual questions asked

• Generates 10-18 specific, actionable question categories (avoiding overly broad categories)

• Categories are tailored to the specific research context and objectives

Example Categories:

• "Affordability and Financial Considerations"

• "Product Features and Policy Options"

• "Claims Experience and Resolution"

Phase 2: Question Categorization

Purpose: Classify every question into exactly one category

Process:

• Each AI interviewer question is analyzed and assigned to a single question category

• Uses single-select classification (one question = one category)

• Updates the messages table with category assignments

Phase 3: Answer Category Generation

Purpose: For each question category, identify patterns in how people respond

Process:

• Groups all answers by their associated question category

• AI analyzes answer patterns within each question category

• Generates 12-18 specific answer categories per question category

• Categories are granular and specific (e.g., "Capitec Bank pricing" rather than "competitor pricing")

Example Answer Categories for "Affordability" Questions:

• "Direct Cost Sensitivity"

• "Premium vs. Cost Comparison"

• "Specific Competing Expenses"

Phase 4: Answer Categorization

Purpose: Tag each answer with relevant categories (can be multiple)

Process:

• Each individual answer is processed and can receive multiple category tags

• Uses multi-select classification (one answer = multiple possible categories)

• Calculates proportions showing what percentage of answers fall into each category

The Interactive Interface

Question Categories Sidebar (Left Panel)

• Displays: All question categories with count of questions in each

• Numbers: Show how many questions have been classified into that category

• Interaction: Click any category to filter the view and see its specific answer categories

• Active State: Selected category highlighted in blue theme

Answer Categories & Key Themes (Center Panel)

• Displays: Answer categories relevant to the selected question category

• Progress Bars: Show proportion of answers in each category (green theme)

• Percentages: Display both count and percentage of total answers

• Interaction: Click any answer category to see specific citations/examples

• Active State: Selected category highlighted in green theme

Citations (Right Panel)

• Displays: Actual question-answer pairs that exemplify the selected categories

• Color-Coded Tags:

  • Blue tags = Question categories

  • Green tags = Answer categories

• Editable: Users can click tags to reassign questions/answers to different categories

• Copy Function: Each citation can be copied for use in reports

Key Features & Capabilities

1. Hierarchical Navigation

Users can drill down from broad themes to specific examples: Question Category → Answer Categories → Individual Citations

2. Real-Time Classification Updates

• Users can modify category assignments by clicking tags

• Changes are immediately reflected in the interface

• Proportions and counts update automatically

3. Quality Control & Validation

• All AI classifications can be manually reviewed and corrected

• Color-coded system makes it easy to distinguish question vs. answer categories

• Clear visual hierarchy helps users understand relationships

4. Flexible Analysis

• Switch between different question categories to explore different aspects of the research

• Each question category reveals its own unique set of answer patterns

• Multiple answer categories per response capture nuanced insights

Business Value

For Researchers

• Speed: Automated categorization reduces analysis time from days to hours

• Consistency: AI ensures consistent categorization criteria across large datasets

• Depth: Captures nuanced patterns that might be missed in manual analysis

• Flexibility: Easy to explore different angles and drill down into specific insights

For Clients

• Actionable Insights: Specific, granular categories provide clear direction for action

• Evidence-Based: Every insight backed by actual customer quotes and examples

• Comprehensive: Captures the full spectrum of customer feedback and sentiment

• Professional Presentation: Clean, interactive interface suitable for stakeholder presentations

This AI Analysis product represents a significant advancement in qualitative research, combining the depth of human insight with the scale and consistency of artificial intelligence.