What is churn?
Churn, (also called lapsing or defecting behaviour) occurs when a customer stops buying from you. The difficult part is identifying the point at which the customer stops buying from you – most organisations define an arbitrary period based on the last order. For example, "all customers who have not purchased for 12 months have churned".
The problem with such an inflexible rule is that each customer is different, so setting an arbitrary period is inaccurate.
A customer who buys every month may have lapsed if they don't order after 2 months, whereas one who orders every 9 months may still be active after 12 months.
"It costs less to retain a customer with a well targeted and and well timed offer that shows an understanding of their relationship with you, than to reactive the same customer once they have lapsed". Churn modelling is the technique of predicting the likelihood to lapse, on an individual basis.
Identify customers before they lapse
Imagine that you are offered a special discount or offer by a company after you have lapsed. You might view this as a bribe, as a sales promotion to buy your business and hence your "loyalty" is not increased.
Now imagine being given the same offer before you lapse. You may see this as a reward for using the company and your loyalty could increase.
The example below shows a streaming model built for a client. At various stages in their purchasing history, each customer's LTV and churn probability is calculated and they are "streamed" into LTV segments (high, medium & low).
Based on a combination of LTV and churn probability, each customer is then allocated to a segment in the churn matrix and sent a range of different communications and promotions, based on price, service and product.
The major benefit is that you can then proactively provide anti-churn strategies to stop the customer from lapsing in the first place.
Qbase have acheived significant success in devising customer retention strategies to reduce churn for large charities and companies in the communications sector.
The major steps in our proven approach are detailed below, to find out more contact Mark Robinson.
- Churn investigation – Compare activity between retained & lapsed customers to identify "triggers" that cause customers to defect
- Multivariate modelling – Use propensity modelling to weigh the significance of and rank each churn trigger
- Churn monitoring – Define & implement measures to highlight any trigger activity likely to cause a customer to churn or defect
- Anti churn offer strategy – Test and roll-out anti churn offers based on service enhancement, product enhancement or price promotion