MarketingX

 

TECHNEWS

MarketingX is proudly produced & published
by Technews
www.technews.co.za
Issue Date: October 2004

Catch them before they jump ship

October 2004

Knowing when a customer is going to jump ship is crucial to catching them before they do. That is according to Suben Moodley of marketing insights company, Knowledge Factory. He was speaking at the recent SAS Forum on business intelligence in Johannesburg.

"Knowing whether or not a customer is going to take his business elsewhere is critical as this allows a company to act fast in order to retain this business. This knowledge can only be achieved using sophisticated analytical software so any customer strategy has to be data-driven," says Moodley.
He cited an example of an organisation whose customer retention strategy was based on assumption. It believed that if it retained a customer for a year, that person would remain a customer for life.
The organisation was shocked to discover, following a churn modelling exercise, that in reality, customers of between one and three years had the highest propensity (32%) to move to the competition.
Using analytics to identify customers who will churn, enables companies to implement marketing strategies to change this behaviour. Called churn modelling, it is now used extensively by businesses with customer data.
"Many companies today are battling to deal with churn. Using good technology, churn modelling can lead to excellent returns," says Moodley.
The first step is to define churn properly. Voluntary churn is when a customer decides not to do business with a company any longer, and goes to a competitor. Involuntary churn happens when an organisation decides not to do business with a customer, usually because of a poor payment history. There is also expected churn, for example when customers no longer buy baby food because their babies get teeth.
"Too many organisations lump these together. Clearly defining churn, and when it has occurred, is critical to churn modelling. Measuring churn is easy - defining it is the problem," he explains.
He defines churn as the closure of one account in conjunction with the opening of another for the same product or service, usually at a reduced price or better service.
"Basically, churn is when customers jump ship. Businesses need to construct definitions that are accurate, and tailored to suit their objectives. For example, they may only want to identify high value customers that are about to churn."
Some companies should also enrich their bespoke data with alternative sources, such as geo-spatial information. This would enable them to look at churn rates compared to income groups, for example.
Text mining is also becoming very useful in churn modelling.
"Much of the data in the real world is text, but large volumes of text data go unanalysed," says Moodley.
"Information within text can be converted for use in churn models. Using solutions like SAS Text Miner, we can group similar text information, for example customer discussions recorded by the call centre. When this is filtered into churn models, it can improve predictive performance by up to 50%."
Moodley advises companies with huge numbers of records to build representative samples and mine these.
"This way they minimise processing time and get effective sampling with little or no loss of generality," he says.
He also advises companies to keep tracking population and score shifts to ensure their models remain accurate. "Customers churn because they struggle to see value in the product or service. The business therefore needs to target them to demonstrate value.
"Retention campaigns, however, are costly to roll out to market. Business cannot simply cross fingers and hope they work. It is important that companies use experimental design techniques and test retention campaigns on small samples initially."
To test retention campaigns, Knowledge Factory creates four population coordinates:
* Customers with a high propensity to churn, who receive a mailed communication.

* A random sample who receive the same communication.

* Customers with a high propensity to churn who receive no communication.

* A random sample who receive no communication.
The churn rates of all these groups are then measured to assess the effectiveness of the churn model, and the communication.
Finally, Moodley stresses that sales, marketing and finance must all collaborate to deliver solutions that provide value for the customer, and hence enhanced margins for the business.
For more information contact SAS Institute, 011 713 3400, www.sas.com/sa


Others who read this also read these articles

Search Site





Subscribe

Previous Issues