The Telecommunication services industry is growing rapidly and is extremely competitive. Business profitability is contingent to the ability to provide customized packages based on a customer's preferences, channelize resources towards the more profitable customer/service segments and the ability to retain customers. A key advantage for the telecommunications industry is the availability of significant amount of data that can be used for guiding various CRM initiatives and increasing business profitability.

Analytics can be used to support the business at different levels:

1. Business planning
2. Marketing and customer acquisition
3. Account management and collections

A brief overview of our services in each of these areas is given below:

1. Business planning

Analytics can provide valuable insights into the profitability of existing business through an analysis of customer profitability in terms of AARPU, contract types (prepaid vs. postpaid) and usage of higher-margin value-added services. The profitability analysis can serve as a guideline for

• Identifying growth opportunities to increase AARPU through restructured tariffs and value-added services
• Develop a view on the profitability and risks attached to the existing customer and service portfolio

2. Marketing and customer acquisition

A profitability analysis of the existing tariff plans can guide efforts in developing an optimal portfolio of tariff plans. As the next step:

• Analytics can provide a view of the optimal service bundles , both from a marketing and profitability point of view.
• Response and propensity models can be developed to understand customer behavior across different distribution channels.

3. Account management and collections

For existing accounts, analytics can be used to provide the appropriate value-added services based on a customer's preferences and activity.

• Analytics also provides the basis for developing up-sell and cross-sell propensity models while also providing the basis for predicting customer churn . Information about likely churns can be combined with customer preferences to optimally allocate resources for customer retention and collections.
• An “early-warning system” for fraud detection can be developed on the basis of regressive analysis of historic fraudulent cases.

For example, for an Asian mobile phone operator, we extracted, analyzed and segmented a large dataset of account attribute and performance data to build an attrition propensity model that identified those accounts that would cancel their contracts in the next month. This propensity model, tailored to the specific client, provided a 5-fold increase in predictive power over the current in-house approach and was shown to be nearly twice as effective as the leading generic model. We built a simple tool, which provided the client with monthly lists of likely churners, which allowed the client to effectively target their finite retention budget at these customers.