Startups usually fall into two main categories. They either manufacture and/or distribute products or they provide services. Some startups may have a combination of both. Depending on the business model, customer churn could be calculated either using number of customers the business lost, the nominal value of recurring revenue lost, or the percentage value of the recurring revenue lost.
For example, a business which has 100,000 customers at the beginning of a period, be it a month, quarter or fiscal year, losing 2,500 existing customers by the end of the year results in customer churn of 2.5%. This method is particularly useful for businesses which have a paid subscription service. A startup which delivers lunch to offices on business days, entertainment, internet or cellphone packages, magazines or online publishing, and accounting services are all examples of businesses where a customer base typically pays a similar subscription for the services provided. In such a case, number of customers is a very strong predictor of revenue and earnings so calculating churn in terms of number of customers lost makes sense.
A more complicated model would be a business with two tiers of customers. Take Spotify, for example. One could pay the subscription fees and remove adverts on the music streaming app, or one could listen for free while being exposed to advertisement. For such a business, it would not make sense to calculate churn in terms of end-users who stop utilizing Spotify as their streaming app. Advertisers also bring in significant revenue for the company. Perhaps Spotify could calculate churn in terms of paid subscribers lost per period but that model would not be efficient for the advertiser side of the model. Instead, churn should be calculated in terms of recurring revenue lost.
Recurring revenue may be calculated as monthly recurring revenue (MRR) or annual recurring revenue (ARR). A local bank may enter into an advertising contract which allows them to run different campaigns on Spotify over a two year period, with renewals available 3 months prior to expiration of the service agreement. That would count as recurring revenue, as would monthly subscriptions paid by users who are expected to continue using the Spotify streaming service indefinitely. If an artist is going on tour in Georgia, for example, and decides to run ads promoting that tour on Spotify for a month, that would not be considered recurring revenue. Churn for Spotify can then be calculated in terms of the nominal or relative value of revenue lost. This would apply to many service provider startups as well as product manufactures whose clients are retailers.
To further clarify why this is significantly different from merely measuring number of customers, think of a startup which delivers lunch packages to 10 companies in a central business district. Two of their biggest clients, OmniCorp and Goliath National Bank, contribute nearly 50% of their revenue. However, the tech startup next door, the Bluth Company, decides that KFC is a better option than the home-cooked meals offered by the delivery business. Calculating churn in terms of number of customers would result in a churn of 10% even though only 3% of their revenue was lost due to the change made by the Bluths. When dealing with hundreds or thousands of customers whose contributions to the business vary significantly, churn in terms of revenue reflects a clearer picture than customer numbers. Regardless, customer numbers are still useful in certain contexts.