Customer Segmenting

Generate strategic customer segment definitions for strategy/canvas/04.segments.md .

Prerequisites

Before proceeding, verify:

• strategy/canvas/03.opportunity.md exists (TAM/SAM/SOM data required)

If missing, inform user:

Canvas 03.opportunity.md required before defining segments. Use fnd-researcher agent to establish market sizing first.

Optional context (read if exists):

• strategy/canvas/01.context.md — KBOS framework

• strategy/canvas/05.problem.md — Problem severity data

Core Principle

Segments must be observable and strategic:

Criterion Test

Observable Can identify via searchable database query

Sizeable Market size estimable from public data

Accessible Reachable through known channels

Differentiable Distinct needs from other segments

Process

1. Load Context

Read available canvas files:

strategy/canvas/03.opportunity.md # Required: TAM/SAM/SOM strategy/canvas/01.context.md # Optional: strategic context strategy/canvas/05.problem.md # Optional: pain data

Extract: market size, trends, existing customer hypotheses.

2. List Segment Hypotheses

From market research, identify 3-5 potential customer groups.

For each, capture:

• Who they are (role, company type)

• Why they might buy (problem fit)

• How big the group is (rough estimate)

3. Define Observable Filters

For each segment, identify 2-4 searchable criteria.

Valid filters (can query in databases):

• Company size: "50-200 employees"

• Industry: "E-commerce, NAICS 454110"

• Technology: "Uses Shopify Plus"

• Geography: "US-based, tier-1 cities"

• Behavior: "Monthly GMV >$100K"

Invalid filters (not searchable):

• "Innovative companies"

• "Growth-minded founders"

• "Customer-centric organizations"

See references/filters.md for comprehensive examples.

4. Score Pain Intensity

Rate each segment's pain 1-5:

Score Signal

5 Hair-on-fire, actively buying solutions

4 Significant pain, budget exists

3 Recognized problem, no urgency

2 Mild inconvenience

1 Unaware of problem

Require evidence for each score — job postings, market reports, interview quotes.

See references/scoring.md for detailed rubric.

5. Estimate Segment Size

For each segment, calculate:

• Total matching filters (from industry data)

• Portion within SAM (addressable)

• Derivation source (cite report or calculation)

Use 03.opportunity.md TAM/SAM as ceiling.

6. Prioritize Segments

Rank by: Pain Intensity × Willingness to Pay × Accessibility

Select:

• 1 Primary (P0) — Immediate focus, highest score

• 1-2 Secondary (P1) — Expansion path

Document rationale for prioritization.

7. Write Output

Format per references/template.md.

Write to: strategy/canvas/04.segments.md

Quality Checklist

Before writing output, verify:

• Each segment has 2+ observable, searchable filters

• No psychographic traits in filters

• Segment sizes quantified with sources

• Pain scores have evidence justification

• 1-3 segments total (not 5+)

• Clear prioritization rationale

• Cross-references 05.problem.md if exists

Common Mistakes

Mistake Example Fix

Too many segments 5+ with blurry boundaries Consolidate to 1-3 focused segments

Vague sizing "Large market" "~12,000 US companies matching filters"

Missing pain evidence "Pain: 4" "Pain: 4 — 340 job postings for this role"

Psychographic filters "Forward-thinking retailers" "Retailers >$1M GMV on modern platforms"

No prioritization logic "Both equally important" "Primary: highest pain (5) + proven WTP"

Output Location

strategy/canvas/04.segments.md

Boundaries

• Does NOT validate segment existence (requires outreach)

• Does NOT guarantee segment accessibility

• Does NOT interview customers (provides framework)

• Segment sizes are estimates from available data

• Pain scores require evidence — flag when assumed

• Does NOT handle persona creation (behavior, not demographics)

• Observable filters must be searchable in databases

• Psychographic traits are NOT valid filters

Resources

• Output template: references/template.md

• Filter examples: references/filters.md

• Scoring rubric: references/scoring.md