Structural analysis outputs
The level set output is usually a google spreadsheet with multiple tabs. Each tab represents a metric like Customer Counts, Revenue, AOV, ARPU, etc and contains many cohort triangles, each of which represent a segment like Overall, By product, By Region, Demographics, etc.
See 108.10.50 Data Analysis - Step 5 outputs and presentations for the way to think about outputting/presenting the level set data. It is useless in a database, it must be viewable, understandable, in front of a human, and it needs to be reliable.
Segments and metrics
- Segmentation is important because averages are misleading. [link to the averages are misleading page when it exists]
- Metrics are important because you view information through many different and important lenses.
- 126.96.36.199.20 Data Analysis - Segments and metrics for outputs
Growth Accounting Metrics
- Gross Retention
- Quick Ratio
- Net churn
Cohort Analysis Metrics
- Customer counts
- Customer quality (LTV)
- Other metrics as appropriate for the business like AOV, ARPU, customer distribution
Reading a cohort triangle
Communication/documentation is key
Cohorts (or any analysis) should not stand on its own. They should have some kind of header/communication to explain what is important about it.
- Quantify cohort metrics. Instead of just saying "cohorts are getting better", say "cohorts are getting 18% better". Quantify it on a true weighted basis, or a simple average, or by recency.
- Export cohorts with both numbers and percentages
- 20200706 Lucky8 notes (clustered)
- Quantitative Approach to PMF: https://tribecap.co/a-quantitative-approach-to-product-market-fit/