Skip to content

florianbonnet14/ThePowerOfAnalytics_ClaudeSkills / segmentation-expert

segmentation-expert

Verified

Identify, analyze, and act on customer segments to uncover hidden patterns and understand behavior differences. Use when investigating KPI changes to understand which customers drove them, planning targeted initiatives, building customer personas, or analyzing performance differences across groups. Helps select meaningful segmentation dimensions, create actionable segments, analyze performance differences, calculate segment contributions, and generate segment-specific recommendations.

FileKindSize
SKILL.mdskill md9.7 KB
references/segmentation_dimensions.mdmarkdown3.9 KB
references/segmentation_methods.mdmarkdown7.0 KB

references/segmentation_methods.md

View on GitHub
# Segmentation Methods Comparison

## Method 1: Rules-Based Segmentation

**Definition:** Explicit, predetermined rules define segment membership

**How It Works:**
```
IF user_sessions_per_month > 20 AND features_used > 5
  THEN Power User
ELSE IF user_sessions_per_month >= 5
  THEN Regular User
ELSE
  THEN Casual User
```

**Pros:**
- Clear and explainable
- Easy to implement in code
- Deterministic (same inputs → same segment)
- Stakeholders easily understand
- Simple to maintain

**Cons:**
- May miss nuanced patterns
- Requires domain knowledge upfront
- Cutoffs can be arbitrary
- May not adapt to changing patterns

**Best For:**
- Operational segmentation (daily use)
- Clear behavioral groups
- When interpretability is critical
- When you know what matters

**Example Use Cases:**
- User engagement tiers
- Customer health scores
- Lead scoring
- Account prioritization

## Method 2: Value-Based Segmentation

**Definition:** Segment by value delivered to or generated from customers

**How It Works:**
```
Calculate: LTV, Revenue, Engagement Score, or custom value metric
Sort users by value
Divide into tiers (e.g., Top 20%, Middle 60%, Bottom 20%)
```

**Pros:**
- Directly tied to business impact
- Clear prioritization
- Simple to understand and communicate
- Easy to calculate

**Cons:**
- Doesn't explain "why" differences exist
- May miss growth potential
- Can be influenced by external factors
- Backward-looking

**Best For:**
- Prioritization and resource allocation
- Retention focus areas
- Customer success triage
- Executive reporting

**Example Use Cases:**
- Account tiering for customer success
- Marketing budget allocation
- Sales territory assignment
- Churn prevention prioritization

## Method 3: RFM (Recency, Frequency, Monetary)

**Definition:** Classic segmentation using three transaction-based dimensions

**How It Works:**
```
R (Recency): Days since last purchase/action
F (Frequency): Number of purchases/actions in period
M (Monetary): Total revenue/value generated

Score each dimension 1-5 (quintiles)
Combine into segments (e.g., 555 = Champions)
```

**Common Segments:**
- Champions (555): Best customers
- Loyal (X5X): High frequency, any recency
- Promising (51X): Recent, low frequency (growth potential)
- At-Risk (1XX): Previously good, now inactive
- Lost (111): Haven't engaged in long time

**Pros:**
- Proven framework (decades of use)
- Balances multiple factors
- Easy to visualize and communicate
- Actionable segments

**Cons:**
- Requires transaction data
- Doesn't capture intent or satisfaction
- May not work for non-transactional products
- Cutoffs can be arbitrary

**Best For:**
- E-commerce and retail
- SaaS with clear usage patterns
- Marketplaces
- Subscription businesses

**Example Use Cases:**
- Email campaign targeting
- Win-back campaigns
- Loyalty program tiers
- Personalized promotions

## Method 4: Statistical Clustering

**Definition:** Machine learning algorithms find natural groupings in data

**Common Algorithms:**

### K-Means Clustering
- Partitions data into K clusters
- Minimizes within-cluster variance
- Fast and scalable
- Requires specifying K upfront

### Hierarchical Clustering
- Builds tree of clusters
- Can visualize as dendrogram
- Doesn't require K upfront
- Computationally intensive

### DBSCAN
- Density-based clustering
- Can find arbitrary-shaped clusters
- Doesn't require K upfront
- Handles outliers well

**Process:**
1. Select features to cluster on (usage, demographics, etc.)
2. Normalize/standardize features
3. Run clustering algorithm
4. Analyze and interpret resulting clusters
5. Name/label segments based on characteristics
6. Validate segments

**Pros:**
- Finds patterns you might miss
- Data-driven, not assumption-driven
- Can handle many dimensions
- May reveal unexpected segments

**Cons:**
- Less interpretable ("Cluster 3" means what?)
- Requires statistical knowledge
- May produce unactionable segments
- Needs sufficient data
- Sensitive to feature selection

**Best For:**
- Exploratory analysis
- Persona development
- Complex behavior patterns
- When you don't know what matters

**Example Use Cases:**
- Customer persona creation
- Identifying user archetypes
- Market segmentation research
- Product-market fit exploration

## Method 5: Persona-Based Segmentation

**Definition:** Segments based on goals, use cases, jobs to be done (JTBD)

**How It Works:**
1. Conduct qualitative research (interviews, surveys)
2. Identify distinct goals, pain points, use cases
3. Create archetypal personas
4. Map users to personas based on behavior/attributes

**Example Personas - Project Management Tool:**
- Marketing Managers: Track campaigns, collaborate with agencies
- Product Managers: Roadmap planning, feature tracking
- Engineering Leads: Sprint planning, bug tracking

**Pros:**
- Deeply human-centered
- Captures "why" not just "what"
- Aligns product and marketing
- Memorable and compelling

**Cons:**
- Requires qualitative research
- Harder to measure
- Can be stereotypical
- May not be exhaustive

**Best For:**
- Product strategy and roadmap
- Marketing messaging and positioning
- Feature prioritization
- Design and UX decisions

**Example Use Cases:**
- Product feature development
- Marketing campaign creation
- Sales messaging
- Onboarding flow design

## Comparison Matrix

| Method | Ease of Implementation | Interpretability | Data Requirements | Business Impact Clarity | Flexibility |
|--------|----------------------|------------------|-------------------|----------------------|------------|
| Rules-Based | High | Very High | Low | High | Medium |
| Value-Based | Very High | Very High | Medium | Very High | Low |
| RFM | High | High | Medium | High | Medium |
| Clustering | Low | Low | High | Medium | High |
| Persona | Medium | Very High | High | Medium | High |

## Choosing the Right Method

**Start with Rules-Based if:**
- You know what behaviors matter
- Need operational segmentation
- Team has clear hypotheses
- Want simple, maintainable system

**Use Value-Based when:**
- Need to prioritize quickly
- Resources are constrained
- Executive audience
- Clear value metric exists

**Apply RFM when:**
- Transactional business model
- Have purchase history
- Marketing-driven segmentation
- Proven framework fits your needs

**Explore with Clustering when:**
- Don't know what patterns exist
- Have rich behavioral data
- Building personas
- Time for exploration

**Develop Personas when:**
- Product strategy phase
- Need qualitative depth
- Multiple stakeholders to align
- Marketing and product need shared language

## Hybrid Approaches

Often best results come from combining methods:

**Example 1: Value-Based + Rules-Based**
- First tier by value (High/Medium/Low)
- Then subdivide by engagement rules
- Result: "High-Value Active" vs "High-Value At-Risk"

**Example 2: RFM + Personas**
- RFM for quantitative segmentation
- Personas for qualitative understanding
- Result: "Champion Marketing Managers" vs "At-Risk Product Managers"

**Example 3: Clustering + Rules-Based**
- Clustering to discover natural groups
- Rules-based to operationalize findings
- Result: Data-driven discovery, operationally simple implementation
planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies.Stars: 108kUpdated Claude CodeCodexVerifiedInstall
meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要".Stars: 108kUpdated Claude CodeCodexVerifiedInstall