This is a very well-written and detailed article. Thank you for this.

I have a question though:

- Is there any reason why the route that you have chosen is the `cumulative_revenue`/`no_of_customers`? The reason I ask that is the `average_revenue_per_user` in each of the cohorts would be a lot higher and this could make it seem like the customers are getting more profitable which if done differently would result in significantly lower numbers.

I tried this with the `revenue`/`no_of_customers` since we were working on the `cohort_level` and got a much smaller value.