Cohort Analysis and Customer Segmentation: Understanding Your Customer Base

Understanding Cohort Analysis

Grouping and tracking customers over time. **Cohort Definition** Cohort: Group of customers acquired in same time period (usually month). Example: January 2024 cohort: - First customer acquired: Jan 1 - Last customer acquired: Jan 31 - Total customers: 50 February 2024 cohort: - First customer acquired: Feb 1 - Last customer acquired: Feb 29 - Total customers: 55 **Retention Cohort Table** Track each cohort's retention % over time. | Cohort | Cohort Size | Month 0 | Month 1 | Month 3 | Month 6 | Month 12 | |--------|-------------|--------|---------|---------|---------|----------| | Jan | 50 | 100% | 98% | 94% | 90% | 80% | | Feb | 55 | 100% | 97% | 91% | 85% | 75% | | Mar | 48 | 100% | 95% | 88% | 80% | 68% | | Apr | 52 | 100% | 93% | 82% | 72% | 55% | Interpretation: - Jan cohort strongest (80% after 12 months) - Apr cohort weakest (55% after 12 months) - Trend: Declining retention over time (products getting worse?) **Why Cohort Analysis Matters** Aggregate data hides trends: - Total retention: 70% (looks okay) - But cohorts: Jan 80%, Apr 55% (shows decline) Aggregate view misses: - Onboarding improvements (didn't help Apr cohort) - Product regression (Apr cohort worse) - Acquisition quality changes (paid ads bringing worse customers) Cohort view reveals: - When problems started (April) - Root cause (acquisition, product, or both) - Trends (improving or declining) **Cohort Patterns** Expected pattern: - Month 0: 100% (all customers) - Month 1: 80-90% (onboarding drop, normal) - Month 3: 70-80% (early churn) - Month 6: 60-70% (stabilizing) - Month 12: 50-70% (mature) Red flags: - Month 1 churn >15% (bad onboarding) - Month 1-6 steep decline (fundamental issues) - Month 6+ continued decline (customers still leaving, not getting value) Good signs: - Steep drop month 0-1 (expected onboarding) - Then flattens and stable (customers stick) - Month 12 retention 60%+ (strong)

Retention Curves and Patterns

Visualizing cohort retention over time. **The S-Curve of Retention** Most healthy cohorts follow S-curve pattern: ``` 100% |● | ● 80% | ●● | ●● 60% | ●● | ●●● 40% | ●●●●● |______________________ 0 3 6 9 12 15 18 Months ``` Pattern: - Steep drop month 0-3 (onboarding, early issues) - Curve flattens month 3+ (stable cohort) - Flattens out at ~50-70% (mature retention) Interpretation: Customers who survive month 3 tend to stick. **Unhealthy Pattern 1: Continued Decline** ``` 100% |● | ● 80% | ● | ● 60% | ● | ● 40% | ● | ● |__________ 0 3 6 9 12 Months ``` Indicates: Fundamental issue - Product not delivering value - Customers eventually leave - Even long-term customers churning Action: Fix product issues (not acquisition). **Unhealthy Pattern 2: Steep Initial Drop** ``` 100% |● | ● 60% | ● | ●●●●● 40% | ●●● | 20% | |__________ 0 3 6 9 12 Months ``` Indicates: Onboarding issues - Customers struggle to get value - High early churn (month 0-1) - Those who survive stay (if they get past onboarding) Action: Fix onboarding (reduce month 1 drop). **Comparing Cohorts** Strong cohort (Jan): ``` 100% |● | ●● 80% | ●●● | ●●●● |_____________ 0 3 6 9 12 ``` Weak cohort (Apr): ``` 100% |● | ● 60% |● | ●● 40% | ●●●● |_____________ 0 3 6 9 12 ``` Difference: Jan cohort retains 80% at month 6, Apr only 60% (major gap). Questions: - What changed between Jan and Apr? - Onboarding? (Jan onboarding better) - Product? (Jan acquired before bug? Apr after?) - Acquisition? (Jan higher quality customers?) Action: Identify and fix the difference.

Free — no card needed

See this in action for your business

AskBiz tracks these metrics automatically — just connect your data and start asking questions.

Segmentation and Analysis

Breaking down cohorts by customer characteristics. **Segmentation Dimensions** Segment by acquisition channel: | Channel | Cohort Size | Month 6 Retention | Quality | |---------|-------------|---|---| | Organic | 20 | 80% | High | | Paid ads | 30 | 50% | Lower | | Sales | 5 | 70% | Medium | Insight: Organic customers most valuable (best retention). Action: Double down on organic (content, referrals, partnerships). Segment by customer size: | Segment | Cohort Size | Month 6 Retention | LTV | |---------|-------------|---|---| | SMB (<5 people) | 30 | 60% | £2K | | Mid-market (5-50) | 15 | 75% | £8K | | Enterprise (50+) | 5 | 90% | £30K | Insight: Enterprise most sticky (quality accounts). SMB high churn. Action: Focus sales on mid-market and enterprise (better economics). Segment by use case: | Use case | Cohort Size | Month 6 Retention | |----------|-------------|---| | Core use case (primary) | 35 | 75% | | Secondary use case | 12 | 50% | | Peripheral use case | 3 | 25% | Insight: Primary use case customers stick, peripheral leave. Action: Better onboarding for primary use case (lock in early). De-prioritize peripheral. **Cohort Economics** Calculate LTV/CAC by cohort: | Cohort | CAC | 6-Month Retention | LTV | LTV/CAC | |--------|-----|---|-----|-------| | Jan (organic) | £1K | 80% | £8K | 8x | | Feb (organic) | £1K | 78% | £7.8K | 7.8x | | Mar (paid ads) | £3K | 60% | £6K | 2x | | Apr (paid ads) | £3K | 55% | £5.5K | 1.8x | Insight: - Organic cohorts: Excellent economics (8x) - Paid ads cohorts: Poor economics (2x) Decision: Stop paid ads (unprofitable), invest in organic. **Identifying Cohort Issues** Problem 1: One bad cohort - April cohort 55% retention vs Jan 80% - What happened in April? - New product launch? (might have broken something) - Price increase? (lost price-sensitive customers) - Sales team change? (different ICP?) - Competitive pressure? (lost deals) - Fix: Investigate root cause, adjust April acquisition if possible Problem 2: Declining trend - Jan 80%, Feb 78%, Mar 76%, Apr 55% (all declining) - Indicates: Systematic degradation - Causes: Product getting worse, onboarding declining, market change - Fix: Deep investigation, product/onboarding improvement Problem 3: Segment-specific issue - Organic cohorts fine (75% retention) - Paid ads cohorts bad (50% retention) - Indicates: Acquisition channel bringing wrong customers - Fix: Improve targeting, pause paid ads, focus on organic

Using Cohort Analysis for Decisions

Taking action based on cohort insights. **Onboarding Optimization** Data: Jan cohort 95% month-1 retention, Apr cohort 80% Question: Why 15% difference? Hypothesis: Onboarding changed between Jan and Apr Investigation: - Jan onboarding: 1-hour call, in-app walkthrough, check-in day 2 - Apr onboarding: Email template, self-serve docs Test: - Run Jan-style onboarding with May cohort - Measure month-1 retention Result: May cohort 92% (matches Jan, validates hypothesis) Action: Revert to Jan-style onboarding, improve hiring calls **Acquisition Channel Shift** Data: - Organic cohorts: 75% retention, 8x LTV/CAC - Paid ads cohorts: 50% retention, 2x LTV/CAC Decision: Shift budget from paid to organic Implementation: - Pause paid ads (unprofitable) - Double down on content marketing (lower CAC, higher retention) - Hire partnerships manager (more organic channels) Expected impact: - Slower growth month 1-2 (less paid volume) - Better quality customers month 2+ (organic ramp) - Better retention overall (fewer low-quality paid customers) **Pricing Tier Optimization** Segment cohorts by pricing tier: | Tier | Cohort Size | Month 6 Retention | |------|-------------|---| | Starter (£50/mo) | 25 | 50% | | Pro (£200/mo) | 15 | 75% | | Enterprise (£2K+/mo) | 5 | 90% | Insight: Higher tier = better retention (more committed customers). Decision: Focus sales on Pro and Enterprise (better economics). Action: - Improve Starter onboarding (it's weak) - Or: Increase Starter price to £100 (upsell to Pro instead) - Or: Discontinue Starter (focus on Pro/Enterprise) **Predicting Churn by Cohort** Use cohort patterns to predict future churn: Jan cohort: 80% at month 6, 70% at month 12 - Monthly churn: 2% in months 6-12 (stable) Apr cohort: 55% at month 6 - If follows Jan pattern: Will churn 2% in months 6-12 - Projected 12-month: 55% × 0.88 = 48% - vs Jan's 70% (22 point gap) Action: Analyze Apr cohort churn drivers, fix before month 12.