It’s a fact of life that customers sometimes leave. This is called ‘churn’ and there are two well defined variants, either: you have a contract with the customer, like a streaming service; or you don’t, like a shop.
Both of these variants have well understood analytical approaches, but it is not common to include the value of the customer in the calculations even though core customers have different behaviour to the average and therefore different signals for an intention to leave from other customers. Also they typically stay a lot longer than average customers.
As a simple example if a customer has visited your site at least once a week for six months and then stops for a month that is a stronger signal than someone visiting twice in the last six months not visiting for a month.
Not only is the churn behaviour easier to spot but the economics of remedial action are also better. In many cases the suggested remedy for churn is a special offer to encourage retention . Unfortunately, the accuracy of the churn prediction is so poor that the remedy is destined to go to a large proportion of people unlikely to churn making the cost of retention this way simply too high. However with the very high value of core customers the value equation becomes rebalanced in the business favour. These are customers really worth keeping.