Andrew Chen has a guest post up from Christoph Janz regarding his spreadsheet for churn, MRR, and cohort analysis. Christoph is the author of the awesome SaaS metrics dashboard that I adapted to work with startups that have an inside sales team.
Cohort analysis is looking at groups of customers over time as opposed to all customers at a given point in time. As an example, on any given month 3% of all customers might churn (they leave and no longer pay for the service). Upon further inspection, after grouping customers based on the month they signed up, one might find that customers within 90 days of signing up are churning at a rate of 10% per month, but once they get past 90 days, they churn at a rate of 2% per month. This cohort analysis would lead to a variety of recommendations.
Here are a few thoughts on cohort analysis:
- Consider sample size and timeframe when evaluating usefulness (e.g. a startup with a small number of customers doesn’t need to spend time on it)
- Break customers into meaningful cohorts based on different factors (e.g. some startups should measure customer cohorts by the week whereas others should do it by the month)
- Monitor multiple customer data points beyond churn including average revenue per user, engagement, logins, up-sells, etc
- Look for anomalies that might influence the data including things like weather, seasonality, etc
Cohort analysis is an important part of the recurring revenue business model and should be incorporated into the standard startup metrics.
What else? What are your thoughts on Christoph’s spreadsheet for churn, MRR, and cohort analysis?