I love building cohort analyses and find them super valuable for measuring how a business is running. But a lot of companies either don’t do them, do them incompletely, or struggle to build them correctly in standard BI tools.

For clarity: a cohort analysis groups customers and examines how they perform over a common period of time relative to some starting point (like signup date), rather than calendar time.

Example: a gym could look at check-ins by month and compare April to June. That’s a calendar analysis. A cohort analysis instead compares how often members visit in month 6 of their membership versus month 1, regardless of when they joined. This lets you see the natural decay curve of gym attendance and whether members who joined during a promotion retain differently than those who joined at full price.

The tool would let people upload CSVs (or connect directly to a data source), handle the data transformation automatically, and spit out the visuals. Users could pivot the data to group customers by whatever factors they want: signup month, acquisition channel, plan type, geography.

Cohort analyses are one of those things that’s conceptually simple but annoying to build. This would make it easy.