Identification of nonactive customers that can be reactivated into active forex traders.
We defined the properties to identify the group of non-active trading customers.
- Past active customers who have not traded for over a defined period.
- A customer who in the past traded with a profitable volume of activity.
During the analysis phase we tried to identify other properties that could impact the customers probability to return and trade for instance his country of origin and the first time he made a deposit.
We segmented the customers and created specific offerings we tried to match to the customers profiles such as: financial support/Personal guidance/close-up/waiver of deposit fees and so on.
We decided to contact the customers personally with the companies call center. Two initial attempts will be made to contact the customer by phone and if that channel fails, two additional attempts will be made to contact the customers by email. The whole process will be managed and registered in the CRM systems that then after can be used to increase the accuracy of the statistical model.
We built a statistical model to identify the probability of each customer to return and become an active customer. The process of building the model included a test of a number of algorithm alternatives, when the algorithm with the best prediction ability was chosen. The most accurate algorithm was selected after comparing all the algorithms results with statistical tools.
Due to the high cost of maintaining a call center and the need to support many languages, it was decided only to only make offers to customers with a high probability of reactivation potential. This would reduce cost of process and maximize profitability.
The potential customers, the attempts to contact them and the responses of the customers where all documented in the CRM system.
The information accumulated in CRM was used afterword’s to increase the accuracy of the model.