Segments & Personas

This guide covers defining audience segments and composing them into personas for adaptation targeting.

Segments

A Segment represents a single audience characteristic across three categories:

Category	What It Captures	Example Vectors
Demographic	Who the audience is	Household Income, Age Group, Education Level
Behavioral	How the audience acts	Usage Pattern, Purchase Frequency, Brand Loyalty
Psychographic	Why the audience acts	Emotional Driver, Lifestyle Values, Aspiration

Each segment has three identifying fields:

• Category — one of DEMOGRAPHIC, BEHAVIORAL, PSYCHOGRAPHIC

• Vector — the dimension being segmented (e.g., “Household Income”)

• Value — the position on that dimension (e.g., “Middle-Income”)

The combination of (category, vector, value) must be unique.

Creating a Segment

1. Navigate to Audiences > Segments > Add Segment

2. Select the Category

3. Enter the Vector (dimension) and Value (position)

4. Optionally add a Description and Insights

5. Save

Segment Insights

Like geographic models, segments support structured insights:

[
  {
    "heading": "Advertising Approach",
    "points": [
      "Value-focused messaging resonates strongly",
      "Emphasize cost-effectiveness and practical benefits",
      "Avoid luxury or aspirational positioning"
    ]
  }
]

These insights are included in the adaptation pipeline context when the segment is part of an active persona.

Personas

A Persona combines geographic targeting (Region, Country, Language) with non-geographic segments to create a complete audience profile:

Persona: "Budget-Conscious First-Timer"
├── Geographic: North America / United States / en-US
├── Demographic: Household Income → Middle-Income
├── Behavioral: Usage Pattern → First-Time Users
└── Psychographic: Emotional Driver → Nostalgia

Creating a Persona

1. Navigate to Audiences > Personas > Add Persona

2. Enter a Name (e.g., “Budget-Conscious First-Timer”)

3. Set geographic targeting:

   • Region — e.g., North America

   • Country — e.g., United States

   • Language — e.g., English (United States)

4. Add Segments via the inline PersonaSegment section — select from existing Demographic, Behavioral, and Psychographic segments

5. Save

Cascading Admin Selectors

The admin uses AJAX-powered cascading dropdowns:

• Selecting a Region filters the Country dropdown to show only countries in that region

• Selecting a Country filters the Language dropdown to show only languages for that country

This prevents invalid combinations (e.g., selecting Japanese language for a German country).

Linking Personas to Ad Units

When creating an adaptation ad unit, select a Persona in the targeting section. This:

1. Pre-fills the Region, Country, and Language from the persona

2. Passes the persona’s segment insights to the adaptation pipeline

3. Provides audience context for the cultural research and writing nodes

See Origin & Adaptation Ad Units for details on configuring adaptation ad units.

How Personas Feed the Pipeline

The compose_insights() function in src/cw/lib/insights.py aggregates insights in order:

1. Region insights — broad cultural patterns

2. Country insights — regulatory and local cultural rules

3. Language insights — localization guidance

4. Persona segments — non-geographic audience insights (Demographic, Behavioral, Psychographic)

This hierarchical composition provides comprehensive context to the pipeline’s cultural research and writer nodes, enabling audience-appropriate adaptations.

Import & Export

Segments and personas have separate import/export commands:

# Segments
uv run manage.py export_segments                # → data/segments.json
uv run manage.py export_segments --dir custom/
uv run manage.py import_segments                # ← data/segments.json
uv run manage.py import_segments --dry-run

# Personas
uv run manage.py export_personas                # → data/personas.json + persona_segments.json
uv run manage.py export_personas --dir custom/
uv run manage.py import_personas                # ← data/personas.json
uv run manage.py import_personas --dry-run

Persona export produces two files:

• personas.json — persona records with geographic references (by code)

• persona_segments.json — persona-to-segment mappings (by category/vector/value)

Import resolves references by code/value rather than by primary key, making data portable across environments.