---
id: 7f3b2d8e-4a6c-4d91-b5e2-1c9f7a3d6e55
name: Generate segment descriptions in natural language
kind: CustomerInsights
cron: 0 30 7 * * *
analysis_window: 90d
comparison_periods: [7d, 28d]
---

Generate **natural-language descriptions** for customer segments to support marketing, personalization, and business analysis.

Process should transform structured segment metrics into **clear segment narratives** that describe customer value, visit habits, spending behavior, dining context, and behavioral tendencies.

Output must contain **segment descriptions only**, without explanations or descriptions of methodology.

## Known Data Inputs

### Customer Segment Data
Structured segment outputs containing:
- segment name
- percentage of customers
- percentage of revenue
- visit frequency segment
- spending tier
- average visit frequency
- average check value
- dominant dining pattern
- service-context tendency
- preferred payment methods
- menu preference profile
- behavioral indicators.

These segment metrics are derived from customer activity data within last **90 days**.

## Analysis Approach

### Stage 1 - Segment Summary Aggregation
Analyze existing segment data and calculate key summary metrics for each segment including:
- percentage of customers in segment
- percentage of total revenue contributed
- average visit frequency
- average check value
- dominant dining pattern
- service-context tendency
- notable behavioral indicators.

### Stage 2 - Behavioral Pattern Extraction
Identify defining characteristics of each segment such as:
- frequent or infrequent visits
- high or low spending behavior
- preferred dining periods
- preferred service context
- stable payment preferences
- recurring menu-category tendencies.

### Stage 3 - Natural Language Description Generation
Convert each segment's metrics and behavioral traits into concise natural-language descriptions summarizing:
- who segment is
- how often they visit
- how much they spend
- most relevant behavioral tendencies.

## Output Constraints

- Use segment data based on **last 90 days** only.
- Prefer aggregated summaries.
- Avoid more than **2-3 processing steps**.
- Return **natural-language segment descriptions only**.
- **Do not include explanations, reasoning, or methodology**.

## Output Format

Return structured segment descriptions.

Example structure:

Segment Description:
- Segment Name: [segment name]
- Description: [natural-language summary]