Skip to content

florianbonnet14/ThePowerOfAnalytics_ClaudeSkills / cohort-analysis-specialist

cohort-analysis-specialist

Verified

Track customer behavior over time by grouping customers based on shared characteristics or experiences. Use when measuring retention rates, understanding how customers evolve over time, comparing new vs old user behavior, evaluating product changes by cohort, identifying at-risk cohorts early, or calculating LTV by cohort. Essential for understanding retention, lifecycle patterns, and how product changes affect different user vintages.

FileKindSize
SKILL.mdskill md9.5 KB
references/cohort_metrics.mdmarkdown7.4 KB
references/retention_table_guide.mdmarkdown7.1 KB

references/cohort_metrics.md

View on GitHub
# Cohort Metrics Catalog

## Retention Metrics

### Classic Retention Rate
**Definition:** % of cohort active in a given period

**Formula:** Active Users(Cohort, Period) / Cohort Size × 100%

**Example:**
- Jan 2024 cohort: 1,000 users
- Active in M3 (April): 450 users
- M3 Retention = 45%

**Use for:** Overall retention tracking, cohort comparison

**Benchmarks (vary by industry):**
- M1: 40-70% (good)
- M3: 30-50% (good)
- M6: 20-40% (good)

### D1/D7/D30 Retention
**Definition:** % active on specific day after signup

**Why these matter:**
- **D1:** Measures immediate value ("Did they come back tomorrow?")
- **D7:** Measures habit formation ("Did they form a routine?")
- **D30:** Measures stickiness ("Did they stick around?")

**Good targets:**
- D1: >40% (consumer), >60% (B2B)
- D7: >30% (consumer), >50% (B2B)
- D30: >20% (consumer), >40% (B2B)

**Use for:** Product optimization, identifying where users drop off

### Rolling Retention
**Definition:** % active in period N OR ANY period after N

**Example:**
- User signs up Jan 1
- Active Jan 15 (M0)
- Inactive in Feb (M1)
- Active in March (M2)
- Rolling M1 retention = YES (active in M1 or later)

**Use for:** Understanding "dormant but not churned" users

### Return Retention
**Definition:** % who return in period N (can be inactive in between)

**Example:**
- User signs up Jan 1
- Inactive in Feb (M1)
- Active in March (M2)
- Return M2 retention = YES

**Use for:** Resurrection campaigns, understanding re-engagement

## Engagement Metrics

### Activity Frequency
**Definition:** How often cohort members are active

**Metrics:**
- % Active Days per Period
- Sessions per Active Day
- Average Time per Session

**Example:**
- Jan 2024 cohort in M3
- Active on 12 of 30 days = 40% active days
- 3 sessions per active day
- 15 minutes per session

**Use for:** Understanding depth of engagement

### Feature Adoption by Cohort
**Definition:** % of cohort that used specific feature

**Track:**
- Which features each cohort adopts
- Speed of adoption (days to first use)
- Ongoing usage vs. one-time trial

**Use for:** Feature success evaluation, onboarding optimization

### Engagement Score Evolution
**Definition:** Composite engagement metric tracked over time

**Example:**
- Engagement Score = (Sessions × 0.3) + (Features Used × 0.5) + (Days Active × 0.2)
- Track average score per cohort per period

**Use for:** Overall engagement health, predictive modeling

## Value Metrics

### Cohort Lifetime Value (LTV)
**Definition:** Total revenue generated by cohort over lifetime

**Formula:** 
```
Actual LTV = Total Revenue from Cohort / Cohort Size

Projected LTV = (Current ARPU × Projected Lifetime) / Discount Factor
```

**Components:**
- ARPU (Average Revenue Per User)
- Lifetime (months/years as customer)
- Gross Margin

**Use for:** CAC targets, channel evaluation, prioritization

### Cumulative Revenue by Period
**Definition:** Total revenue generated by cohort up to period N

**Example:**
```
Cohort   M0    M1    M2    M3    M4    M5
Jan 24   $50K  $80K  $95K  $105K $110K $113K
```

**Use for:** Understanding revenue trajectory, LTV projection

### Time to 80% LTV
**Definition:** How long until cohort generates 80% of projected LTV

**Example:**
- Projected LTV: $100
- Revenue by M6: $80
- Time to 80% LTV = 6 months

**Use for:** Payback period calculation, cash flow planning

### Revenue Retention
**Definition:** % of revenue retained from cohort

**Formula:** Revenue(Cohort, Period N) / Revenue(Cohort, Period 0) × 100%

**Note:** Can exceed 100% with expansion revenue

**Use for:** Net revenue retention, expansion tracking

## Predictive Metrics

### Leading Indicator Score
**Definition:** Early behaviors that predict long-term retention

**Process:**
1. Identify which Week 1 behaviors correlate with M6 retention
2. Create composite score
3. Track for new cohorts

**Example Indicators:**
- 3+ sessions in Week 1 → 75% M6 retention
- Invited teammate → 82% M6 retention
- Used Feature X → 68% M6 retention

**Use for:** Early warning system, proactive intervention

### Activation Rate by Cohort
**Definition:** % of cohort that completes activation milestone

**Track:**
- Activation rate per cohort
- Time to activation
- Correlation with retention

**Use for:** Onboarding optimization, retention prediction

### Quality Score Evolution
**Definition:** Cohort quality metric tracked over time

**Components:**
- Activation rate
- D7 retention
- Feature adoption
- Any early engagement metric

**Use for:** Acquisition channel evaluation, quality trends

## Comparison Metrics

### Cohort-to-Cohort Change
**Definition:** Period retention change between consecutive cohorts

**Formula:** Cohort(N) M1 - Cohort(N-1) M1

**Example:**
- Feb M1: 68%
- Jan M1: 65%
- Change: +3pp (percentage points)

**Use for:** Identifying inflection points, trend direction

### Best vs Worst Cohort Gap
**Definition:** Difference between best and worst performing cohort

**Formula:** Max(All Cohort Retentions) - Min(All Cohort Retentions)

**Example:**
- Best cohort M3: 55%
- Worst cohort M3: 42%
- Gap: 13pp

**Use for:** Understanding variance, identifying what changed

### Cohort Convergence
**Definition:** Do cohort retention curves converge or diverge over time?

**Measure:** Standard deviation of retention rates across cohorts by period

**Use for:** Understanding if product improvements are sustainable

## Health Metrics

### Retention Curve Shape Score
**Definition:** Quantifies "health" of retention curve

**Components:**
- Initial drop (M0→M1)
- Slope after M1
- Plateau level (if any)
- Time to plateau

**Classification:**
- Healthy: Quick plateau at acceptable level
- Concerning: Continuous decline, no plateau
- Critical: Steep continuous decline

**Use for:** Executive summary, health monitoring

### Early Warning Signal
**Definition:** Leading indicators that recent cohorts underperforming

**Triggers:**
- D7 retention drops >5pp from trend
- M1 retention below historical average
- Activation rate declining 2+ cohorts in a row

**Use for:** Proactive issue detection

### Resurrection Rate
**Definition:** % of churned users who return

**Formula:** Churned Users Who Returned / Total Churned × 100%

**Benchmark:** 10-20% resurrection is common

**Use for:** Win-back campaign potential, product stickiness

## Segmented Cohort Metrics

### Cohort × Segment Retention
**Definition:** Retention broken down by both cohort and segment

**Example:**
```
         M1 Retention
         Enterprise  SMB
Jan 24   78%        65%
Feb 24   80%        68%
```

**Use for:** Understanding which segments improve/decline

### Cohort × Channel Quality
**Definition:** Retention by acquisition channel within cohort

**Use for:** Channel quality over time, CAC optimization

### Cohort × Feature Usage
**Definition:** Retention by feature adoption within cohort

**Use for:** Feature impact, onboarding focus

## Calculation Tips

**Handling Edge Cases:**
- Users who sign up on last day of month → Include in that month's cohort
- Users with no activity → Mark as inactive, not missing data
- Test/internal users → Exclude from analysis
- Users who delete account → Include as churned

**Statistical Significance:**
- Small cohorts (<100 users) → Wide confidence intervals
- Use statistical tests when comparing cohorts
- Report confidence levels for predictions

**Data Freshness:**
- Mark incomplete periods clearly
- Wait for period to complete before comparing
- Use projections for most recent cohorts
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 196kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 196kUpdated Claude CodeCodexVerifiedInstall