User Segmentation

User Segmentation

You are a user segmentation specialist. Divide your user base into meaningful groups so you can deliver the right experience, messaging, and upgrade path to each user. In PLG, segmentation is the difference between a generic product that sort of works for everyone and a personalized experience that converts each user type optimally.

1. Diagnostic Questions

Before building your segmentation strategy, answer these:

1. What data do you collect about users? (Profile data, usage events, billing data, firmographic data)
2. How many active users do you have? (Segments need sufficient sample sizes -- at least 100-200 users per segment)
3. Do you have a way to act on segments? (Can you target messages, emails, features, or experiences by segment?)
4. What decisions are you trying to inform? (Onboarding, messaging, pricing, sales outreach, feature development)
5. Do you already have implicit segments? (Different plans, roles, use cases that naturally separate users)
6. What is your activation metric? (Needed to segment by lifecycle stage)
7. What engagement data do you track? (Feature usage, session frequency, depth of usage)
8. Do you have churn prediction signals? (Declining usage, support tickets, failed payments)

2. Segmentation Dimensions

2.1 Behavioral Segmentation

Example behavioral segments:

Segment: Power Users
  Criteria: Sessions/week >= 5, Feature breadth >= 70%, Active 30+ days
  Size: ~10% | Action: Expansion offers, beta access, advocacy program

Segment: Regular Users
  Criteria: Sessions/week 2-4, Feature breadth 30-69%, Active 14+ days
  Size: ~30% | Action: Feature adoption campaigns, engagement deepening

Segment: Casual Users
  Criteria: Sessions/week 0.5-1, Feature breadth <30%, Active 7+ days
  Size: ~35% | Action: Activation nudges, value demonstration, use case guidance

Segment: At-Risk Users
  Criteria: 50%+ session drop week-over-week OR no sessions in 7+ days
  Size: ~15% | Action: Re-engagement campaigns, feedback requests, support outreach

Segment: Dormant Users
  Criteria: No activity in 30+ days, previously had 3+ sessions
  Size: ~10% | Action: Win-back campaigns, sunset communications

2.2 Firmographic Segmentation

2.3 Lifecycle Segmentation

1. New User (Day 0-3): Hasn't completed activation. Goal: activation.
2. Activated (Day 1-14): Completed activation, not yet habitual. Goal: build habit.
3. Engaged (Day 7-90): Regular usage, multiple sessions/week. Goal: expand value, prepare upgrade.
4. Power User (Day 30+): Top 10% usage, broad feature adoption. Goal: expand seats, upgrade, advocacy.
5. At-Risk (Any time): Usage declining from personal baseline. Goal: re-engage before churn.
6. Churned (After inactivity threshold): No activity beyond threshold. Goal: win-back, learn from churn.

2.4 Intent Segmentation

2.5 Plan/Tier Segmentation

3. Behavioral Cohort Creation

3.1 Power Users

Identification criteria (use 2-3):

• Top 10% by session frequency, feature breadth, or content created
• Consistent usage for 30+ consecutive days
• Using advanced/premium features regularly

Actions: Beta access, advisory board, referral prompts, expansion paths, study behavior to improve onboarding for others.

3.2 Champions

Power users who also exhibit expansion and advocacy signals:

• Invited 2+ team members
• Share content externally
• NPS 9-10
• Engage with community positively

3.3 At-Risk Users

Detection signals:

Signal: Usage frequency decline
  Rule: Current week sessions < 50% of 4-week trailing average

Signal: Feature breadth contraction
  Rule: Features used this week < 50% of peak week

Signal: Login gap
  Rule: Days since last login > 2x typical inter-session gap

Signal: Support escalation
  Rule: 2+ support tickets in 7 days, or negative sentiment

Composite at-risk score: Each signal = 1 point. 2+ signals = at-risk.

3.4 Dormant Users

Thresholds (adjust for your product's usage cadence):

• Daily-use product: No login for 7+ days
• Weekly-use product: No login for 21+ days
• Monthly-use product: No login for 60+ days

Sub-segments: Lapsed after activation (higher win-back potential), never activated (diagnose why), churned after paying (highest value to win back).

4. User Scoring Models

4.1 Engagement Score

Engagement Score Components (0-100):

Usage frequency (0-30 points):
  7+ sessions/week: 30 | 4-6: 20 | 2-3: 12 | 1: 5 | <1: 0

Feature breadth (0-25 points):
  80%+ features: 25 | 50-79%: 18 | 25-49%: 10 | <25%: 3

Engagement depth (0-25 points):
  Avg session > 30 min: 25 | 15-30: 18 | 5-15: 10 | <5: 3

Recency (0-20 points):
  Today: 20 | 1-3 days: 15 | 4-7 days: 8 | 8-14 days: 3 | 14+: 0

Tiers:
  80-100: Highly engaged (power user)
  60-79: Engaged (regular)
  40-59: Moderate (casual)
  20-39: Low (at risk)
  0-19: Disengaged (dormant/churning)

4.2 Activation Score

Activation Score (customize milestones per product):

Example for a project management tool:
  Created first project: 20 pts | Added first task: 15 pts
  Invited team member: 25 pts | Completed first task: 15 pts
  Used template: 10 pts | Connected integration: 15 pts

Tiers:
  80-100: Fully activated
  50-79: Partially activated (nudge remaining steps)
  20-49: Early activation (guide next steps)
  0-19: Not activated (immediate onboarding attention)

4.3 Expansion Readiness Score

Expansion Readiness Score (0-100):

Usage approaching limits (0-25): 80%+ of limits: 25 | 60-79%: 15 | 40-59%: 5
Team growth signals (0-25): Added members recently: 15 | Using collab features: 10
Feature interest signals (0-25): Clicked locked features: 15 | Viewed pricing: 10
Engagement level (0-25): Score >70: 25 | 50-70: 15 | 30-50: 5

Tiers:
  75-100: High readiness (proactive outreach, upgrade prompts)
  50-74: Moderate (nurture, show higher-tier value)
  25-49: Low (deepen current engagement first)
  0-24: Not ready (focus on activation and value delivery)

4.4 Churn Risk Score

Churn Risk Score (0-100):

Usage decline (0-30): 70%+ decline: 30 | 50-69%: 20 | 30-49%: 10
Recency gap (0-25): 2x typical gap: 25 | 1.5x: 15 | 1x: 5
Support signals (0-20): Negative interactions: 20 | Multiple tickets: 10
Billing signals (0-15): Failed payment: 15 | Downgraded: 10 | Viewed cancel page: 10
Engagement trend (0-10): Declining: 10 | Stable/increasing: 0

Tiers:
  70-100: High risk (immediate intervention)
  40-69: Moderate risk (proactive outreach)
  20-39: Low risk (monitor)
  0-19: Healthy

5. Segmentation for Personalization

5.1 Onboarding Paths by Segment

IF user.role == "designer":
  Flow: Design templates, design tool integrations, visual workspace
  First action: "Create your first design project"

IF user.role == "developer":
  Flow: API docs, code integration, developer workspace
  First action: "Connect your first repository"

IF user.role == "manager":
  Flow: Team setup, reporting, collaboration features
  First action: "Invite your team"

IF user.role == "unknown":
  Flow: General flow with role selection step

5.2 Feature Recommendations by Usage Pattern

IF user uses feature A frequently AND has never used feature B:
  AND feature A users who also use B have 30% higher retention:
  THEN recommend B: "Users who use [A] love [B] because [benefit]"

Implementation:
  1. Build feature affinity matrix (which features power users combine)
  2. For each user, find gaps vs similar power users
  3. Surface top 1-2 recommendations via in-product messaging

5.3 Upgrade Messaging by Engagement Level

6. Implementing Segmentation

6.1 Event-Based Rules

segment "Power Users" {
  conditions {
    events.session_count(last_7_days) >= 5
    AND events.unique_features_used(last_30_days) >= 0.7 * total_features
    AND user.days_since_signup >= 30
  }
  refresh: daily
}

segment "At Risk" {
  conditions {
    events.session_count(last_7_days) < 0.5 * events.avg_weekly_sessions(last_28_days)
    AND events.avg_weekly_sessions(last_28_days) >= 2
  }
  refresh: daily
}

segment "Expansion Ready" {
  conditions {
    scores.engagement >= 60
    AND (
      usage.percentage_of_limit >= 0.7
      OR events.pricing_page_view(last_14_days) >= 1
      OR events.locked_feature_click(last_14_days) >= 1
    )
  }
  refresh: daily
}

6.2 ML-Based vs Rule-Based

Use ML clustering when: You suspect undiscovered user groups, have 10,000+ users with rich behavioral data, have data science resources.

Use rule-based segments when: You know your segments (lifecycle, plan, role), have <10,000 users, need easy-to-explain segments.

6.3 RFM Analysis

RFM Scoring (1-5 each):

Recency: 5=Today, 4=1-3 days, 3=4-7 days, 2=8-14 days, 1=15+ days
Frequency: 5=Daily, 4=4-6x/week, 3=2-3x/week, 2=Weekly, 1=<Weekly
Monetary (plan value, seat count, or usage volume):
  5=Enterprise/highest, 4=Pro/high usage, 3=Basic paid, 2=Active free, 1=Inactive free

RFM Segments:
  555, 554, 545: Champions -- nurture, upsell, advocacy
  444, 445, 455: Loyal -- deepen engagement, expand
  334, 343, 344: Promising -- increase frequency, feature adoption
  233, 234, 244: Need attention -- re-engage, show value
  111, 112, 121: Lost -- win-back campaign or sunset

7. Segmentation in Practice

7.1 Email Campaigns by Segment

7.2 Sales Prioritization (PLS)

Tier 1 (Hot -- sales-assist immediately):
  Expansion readiness > 75, 5+ active users, pricing page 2x in last week, enterprise signals

Tier 2 (Warm -- outreach within 1 week):
  Expansion readiness 50-75, 3+ active users, using team/collab features

Tier 3 (Nurture -- marketing-led):
  Expansion readiness 25-49, individual user, no team signals

Tier 4 (No action):
  Expansion readiness < 25, not activated or disengaged

8. ICP Refinement

Step 1: Identify best customers (top 20% by retention + revenue + engagement)
Step 2: Find common attributes (firmographic, behavioral, channel)
Step 3: Compare to worst customers (bottom 20% by retention)
Step 4: Refine ICP -- current vs data-informed, document delta
Step 5: Action -- update targeting, adjust messaging, deprioritize churning segments, inform roadmap

9. Segment Analysis

9.1 Cross-Segment Comparison Dashboard

| Metric | Power Users | Regular | Casual | At-Risk | Dormant |
|---|---|---|---|---|---|
| Segment size | [N] ([%]) | [N] ([%]) | [N] ([%]) | [N] ([%]) | [N] ([%]) |
| 30-day retention | [%] | [%] | [%] | [%] | [%] |
| Paid conversion | [%] | [%] | [%] | [%] | [%] |
| Avg revenue | [$] | [$] | [$] | [$] | [$] |
| LTV | [$] | [$] | [$] | [$] | [$] |

9.2 Movement Analysis

Segment transition matrix (monthly):

From \ To    | Power | Regular | Casual | At-Risk | Dormant | Churned
Power        | 85%   | 10%     | 2%     | 2%      | 1%      | 0%
Regular      | 8%    | 70%     | 12%    | 7%      | 2%      | 1%
Casual       | 2%    | 10%     | 55%    | 15%     | 12%     | 6%
At-Risk      | 1%    | 5%      | 10%    | 30%     | 30%     | 24%
Dormant      | 0%    | 2%      | 3%     | 5%      | 60%     | 30%

10. Anti-Patterns

11. Output Format

# Segmentation Strategy

## Segmentation Goals
- Primary goal: [What decisions will segmentation inform?]
- Key questions to answer: [...]

## Data Foundation
- Available data sources: [Product events, CRM, enrichment, billing]
- Data gaps: [What's missing?]

## Segment Definitions

### Segment 1: [Name]
- Criteria: [Specific, measurable rules]
- Expected size: [N users, X% of base]
- Actions: Onboarding / Messaging / Upgrade path / Support
- Success metric: [How you measure if segment is served well]

## Scoring Models
- Engagement score: [Components and weights]
- Expansion readiness score: [Components and weights]
- Churn risk score: [Components and weights]

## Personalization Playbook
| Touchpoint | Segment 1 | Segment 2 | Segment 3 |
|---|---|---|---|
| Onboarding | [...] | [...] | [...] |
| In-product messaging | [...] | [...] | [...] |
| Email campaigns | [...] | [...] | [...] |
| Upgrade prompts | [...] | [...] | [...] |

## Implementation Plan
- Phase 1: Basic segments (lifecycle + plan)
- Phase 2: Behavioral segments (engagement scoring)
- Phase 3: Advanced (ML clustering, predictive scoring)

## Measurement
- Track: segment distribution, transitions, per-segment KPIs
- Cadence: [Weekly / Monthly]

Related skills: product-led-sales, plg-metrics, product-analytics, retention-analysis