Predicting Adoption of MVAS

Project Details


2012 (Nov)


Anand Prasad, Charanpreet Singh Arora, Dhruv Gandhi, Gagan Oberoi, Nikesh Lamba





The Indian Telecom industry is amongst the most fiercely fought services market in India with more than 10 large-scale operators providing voice and data services at highly competitive prices. The Industry has witnessed significant reduction in profit margins in recent years, with the average revenues per user from voice telephony being amongst the lowest in the world. On the other hand, margins from value added services are largely seen as the next source of growth for mobile operators. The telecom companies therefore need to be innovative to come up with more relevant and value-adding services to gain a competitive edge in this cut-throat environment.

The client in this case is an advisory board to a consortium of Indian telecom operators consulting on the issue of declining revenues. The consortium wants to increase the subscription of MVAS services (services such as ring tones, sports updates, weather updates etc.) and identify potential customers within their existing bases to be targeted. They also want to understand key factors that would govern the adoption and usage of mobile value-added services.

Demand for MVAS in India has been on the rise in recent years, with growth in mobile penetration and customer purchasing power. According to a report published by Deloitte Consulting, the demand for MVAS is being driven by the following services:
 Information based services such as news updates, health-related information, stock details etc.
 Application based services that need the user to play an active role such as checking the status of payments, GPS
 Enablement services which are a substitute to those provided by physical infrastructure such as a bank or a school e.g. person-to-person payments, travel reservations etc

These services determine consumer attitude towards MVAS and thus need to be considered for any predictive model developed to measure the likelihood of adoption. Other important factors that could be considered are access to technology (smartphone ownership, Internet enabled phone), demographics (age, gender, education etc.), customer satisfaction levels with mobile operators (call charges, network coverage etc.) and current usage levels (monthly expenditure on mobile services, Internet usage).

There are various classification techniques that could be employed to build such a predictive model, the choice of which would depend on the type of data available – categorical or continuous. We also need to keep in mind if the data will be available on future dates to help predict consumer behavior. It is typically considered useful to combine predictions from different techniques in order to reduce inaccuracies and obtain more sound results. This ‘combination-based’ methodology called ensembling was employed by our team to reach final predictions. End results indicate valuable information on traits defining MVAS adoption with the younger age groups and the male gender dominating trends. Moreover, services such as social networking, news updates and GPS outscore other examples of MVAS as triggers of customer adoption.

Application Area: