Recency for Customer Retention on Subscription Based Services

Recency is a technique used to predict future Customer behaviour, based on past Customer behaviour. It aims to capture & analyse how recently each Customer has interacted actively with a Business. If recency is measured by an appropriately selected metric, it can be a very powerful predictor of future customer behaviour. Actually, it has been proved repeatedly in practise, that recency is the strongest predictor of future customer behaviour, among the three RFM analysis factors: recency, frequency, monetary. Therefore recency analysis can yield substantial business value, if carried out successfully.

The generic technique of recency analysis, has to be adopted to the different context of every Business: unique Customer lifecycle, product type. In this article we examine the important case of subscription based (or continuity) services. Such services involve the continuous usage by a Customer, often based on a contract. Common examples are: bank accounts, credit cards, fixed & mobile telecommunication services. These services are offered in highly competitive markets, characterised by high customer switching (attrition or churn) rates, which erode the profitability of those who suffer it.

The selection of suitable measurements for recency is not obvious, in the case of continuity services. A bank account or a telephone subscription is potentially used every day, therefore ‘last time used’ cannot normally be the basis for a recency measurement. The fact that the service is used, does not reflect an active choice of the Customer to interact with the Business. Therefore one should try to identify those events which reflect ‘an active choice of the Customer to interact with the Business’. Moreover, these events should be divided into events reflecting a positive attitude of the Customer towards the business and a predisposition to strengthen the relationship with the Business (e.g. a service upgrade) and events reflecting an increasingly negative attitude of the Customer towards the business (increased Customer friction) and a predisposition to terminate the subscription.

Events signalling a positive attitude are:

o An order in general

o An additional order, building up a subscription portfolio (studies in the banking sector have shown that customers with a larger product portfolio, tend to be more loyal than those with a smaller one)

o A service upgrade (e.g. moving to a higher fixed monthly fee contract in mobile telephony)

o An order for a service enhancement, taking advantage of a service feature not used before

o Enrolment to the web channel, offered by the service

o The acceptance of a campaign offer

Events signalling a negative attitude are:

o A complaint

o A product cancellation, reducing the subscription portfolio size

o A subscription portfolio termination

These two event categories of ‘reduced’ and ‘increased’ Customer ‘friction’, should be considered separately, since they have opposite effect. They should not be used in the same metric, since they may be cancelling each other.

Moreover, the selection of the appropriate division of time into recency time periods, can affect the effectiveness of the prediction. These time periods relate to the Customer lifecycle of each product type.

Alternative recency metrics can be tested on their effectiveness, with a test campaign. The best measurement is the one that produces the best response rate prediction. In addition, this metric should yield substantial differentiation in response rates between quintiles, especially in the highest recency quintiles.