Verizon Wireless principal data mining engineer Ksenija Krunic pointed out that the average churn rate was 2 per cent per month. The average cost of acquiring a new customer ranges from $320-$360 as per industry analysts.
In order to manage churn, operators usually sent out pamphlets to people whose contracts were on the verge of expiry. To bring churn into control, Verizon decided to develop churn predictive models using data mining techniques. The models were used to analyse outputs to identify those sets of customers who could be looking at exiting from the company. It also used that predictive modelling to target customers with specific, relevant and timely offers.
Simultaneously, the company decided to use data mining techniques to predict the right kind of plan for any subscriber based on his/her call usage pattern.
Up until this point, data mining was basically an IT initiative in the company. In order to derive the best results, IT brought the idea to marketing team and presented it as a partnership project within the organisation. The marketing team suggested that IT should work on offering better pricing plans and recommended additional attributes that should be taken in to use in building model.
The partnership of the IT and marketing team was a learning experience for both the teams, said Ms Krunic. While those from the marketing side learned modelling processes as well as capabilities and weaknesses of modeling, those from IT got a better perspective of the business processes and direct marketing strategies.
Ms Krunic said as a part of the partnership, the IT team sat through phone calls made to subscribers in order to get a better idea about what exactly were customers concerns.
Verizon used data mining to build a model as a result of which the company could decide the score for every customer. That is the score for propensity to leave for every customer, identify precise reasons as to why the customer is leaving. This information was used by marketing while crystallising certain offers (tariff plans) for different customer segments. These results were made available to the customer service representatives, retail and telemarketing. All this was finally appended back to the data warehouse for further usage.
Verizon used data mining techniques to do predictive churn modeling (PCM) and predictive takers modeling (PTM). Simply put, out of the total targeted customers who would have high propensity to leave, Verizon had to use various techniques to arrive at a very specific small group, who would end up as actual takers of Verizons offers. While PCM was used to find a smaller group than the overall target group, PTM was used by the company to target only those customers who were likely to stay.
In the process, the company sent mailers to a relatively smaller number of people and saved on costs incurred over those customers who are less likely to stay.
Verizon was able to reduce its churn from over 2 per cent to below 1.5 per cent using the data mining tools. Ms Krunic said Verizon was able to achieve this because the company had a critical, highly visible business problem on which it focused. The fact that a multi-disciplinary team with specific area experts was used was another important reason among others for the success of the project.
We had a well defined problem churn, a defined customer segment retail. We began with one marketing campaign by sending out 40-60,000 mailers per month with highly personalised, unique offers for each customer. Now we send out over 400,000 mails per month, said Ms Krunic.
Elaborating on the multi-disciplinary team approach, Ms Krunic said that apart from the IT and marketing department, Verizon had used expertise from the management team and also brought in consultants for certain portions of the project.
Various business benefits that accrued to Verizon Wireless due to adoption of data mining techniques involved cost reduction, revenue increase and learning processes among others. The companys direct mail budget for churner mailing reduced by 60 per cent for highly targeted marketing campaigns. The revenue increase could be seen from thousands of contract renewals per month and increased usage by subscribers.
Ms Krunic summed up the keys to effectively combat churn and maximize return on investment (RoI). We proactively addressed customer needs before the customer made a decision to leave by custom tailoring marketing campaigns. We also targeted customers with high propensity to accept the offer, Ms Krunic said.