• Intervention to reduce churnIntervention to reduce churn
  1. Product Improvement
  
  Changing product features or content
  Increase stickiness
  Increase cost for a customer to switch to an alternative (providing valuable features that are hard to reprodu...
• Churn is hard to fight
  • Churn is hard to prevent
  • Churn is hard to predictChurn is hard to predict
    Predicting churn doesn't work (well)
    
    Predicting churn should be targeted with the actionable insight, not just binary results (churn or not)
    Churn is hard to predict with high accuracy
    
    ...
  • Reducing churn is a team effort
• Great Customer Metrics
  • Customers who use the product features more than five times a month churn at half the rate of customers who use it only once a month or less.
• Products with recurring user interactionsProducts with recurring user interactions
  Payments Type
  
  Recurring payments - Fixed payments of the same amount for each period of service
  Usage-based payments - payments for the amount of service used, based on some unit of measure
  ...

Chapter 2

• How to calculate churn/churn rate/retention rate
• What is the subscription database look like
• Net retention
  • = recurring revenues
  • = MRRretained account/MRRstart
  • MRR = Monthly reucrring revenues
  • Net retention combines the effects of churns, upsells and down sells*
  • Negative churn – when the increase in revenue from upsells is greater than the combined negative effect of down sells and churns
  • Impressive to report to investors
• Standard account-based churn
  • uneffected by upsells, down sells and expiration of discounts
  • outer join
  • Use Standard account-based churn when every users pay with the same amount.
• Activity (event-based) churn for non-subscription products
  • Define of churn
    • Do not active for a time period
    • Active =
      • Do some certain specific events
      • Has minimum number of events
      • Purchases or has revenue generated from ads at some minimum amount
• Advanced churn: Monthly recurring revenue (MRR) churn
  • Use MRR churn when users has multiple pay variations
  • MRR Churn = (MRRchurned accounts + MRRdown sell) / MRRstart
• Standard Churn > MRR Churn > Net retention Churn
  • Accounts that pay more churn less.
• Churn rate conversion
  • Month to Year
    • Churn = 100% - (1 - churn_month) ^12
  • Year to Month
    • Churn = 100% - (1 - churn_year) ^1/12
  • Any p days period
    • Churn = 100% - (1 - c)^365/p
  • Measuring churn over a time window that is different from the typical subscription length can result in errors in the churn rate
• Seasonality
  • Handle seasonality using time-series analysis

Chapter 3

• Metric QA
  • How Metrics change over time
    • Missing data in some days
    • Extream value in some fields
    • % account that has metrics
• Event QA
  • How events change over time
    • Missing/Extream - by counts per day
    • event per accounts
• Anomaly detection
• 3.9 - Selecting measurekent period
  • If an event is rare, use a longer period
  • Short-time period are more responsive to changes, less sensitive at picking up accounts with low event levels
  • For subscription with a fixed term like a month or a year, behavioral measurements should be similar in time scale to the term of the subscription