Generating Cellular Intelligence

Competency center integrates analytics into organizational culture, leads to low churn rates

As the cell phone market matures and subscriber numbers increase in Japan, business challenges for telecommunications providers shift from getting new customers to keeping existing customers and persuading them to increase cell phone use.

At NTT DoCoMo, the leading mobile communications provider in Japan, SAS data mining technology was introduced to fine-tune strategies that increase customer satisfaction. A Business Intelligence Competency Center (BICC) was established for the same reason.

We recently asked Akira Kubota, Director of NTT DoCoMo Information Systems Group, how effective the BICC is as an organizational structure to support the companywide use of business intelligence.

Most people use cell phones these days, and the cell phone market is seen as saturated. What steps is NTT DoCoMo taking in order to improve business performance?
AKIRA KUBOTA: It is certainly true that responding sensitively to customers' needs and increasing customer satisfaction is now vital. The best way to achieve these objectives is to implement strategies that anticipate individual customer behavior and needs, using sophisticated analytical tools. We are using SAS data mining technology and have established a BICC as the organizational structure responsible for promoting the use of business intelligence throughout the organization.

Based on the results obtained from our analyses, including data mining, we are now introducing programs that will lead to greater levels of customer satisfaction. Actions we took in fiscal year 2005 resulted in a churn rate of 0.77 percent, which is 0.24 percent lower than the previous year. This amounts to a retention of 9.9 billion yen (US$84 million) annually, assuming 0.24 percent of our 50 million users were using 6,900 yen worth of services per person per month.

That's an amazing figure. I understand it was the BICC that brought about that sort of benefit, but what was behind the actions you implemented?
KUBOTA: To use data mining – and other sophisticated analytical tools – and link the results to improvements in business performance, not only do you need IT to do the analysis, but one must also think about creating an organizational structure to support the analytical work. Knowledge of information systems, data, tools and bulk data-processing are all essential to realizing sophisticated analytical work.

The business divisions did not have enough of this knowledge when we introduced data mining in 2000, so we were unable to carry out sophisticated analysis smoothly. Furthermore, there was inadequate general support from the IT division. The BICC was established in 2003 as an organizational structure for data analysis support and recommendation purposes, as well as for promoting the effective use of business intelligence across the organization.

What is the actual form of the NTT DoCoMo BICC?
KUBOTA: Our BICC is in our Information Systems Department. Inside is the Analysis Group, which is responsible for data mining as well as other types of sophisticated analysis, and a Data Extraction Group, which extracts complex data. Head count in those two groups adds up to approximately 30 people. We also have a Data Warehouse Development Group doing development work on the data warehouse. This BICC organization effectively meets the business intelligence needs of our business divisions.

The BICC does the planning and development for the data warehouse and data mining systems and manages data analysis projects. For each issue that needs to be analyzed, such as preventing contract termination, the BICC and the related business divisions, such as the Corporate Strategy and Planning Department and the Corporate Marketing Departments, collaborate to find the best answers. Each of the business divisions thus benefits from the BICC and is able to implement projects using the intelligence derived from data.

The business divisions and the BICC start working through all the conceivable scenarios by topic. There may be more than a hundred scenarios for each topic. From several hundred scenarios, the BICC creates several hundred or several thousand variables for the project and builds a mathematical model that can be used for prediction. Using this prediction model, the business divisions can then devise measures that can be taken to increase levels of customer satisfaction, stop contract cancellations and promote the uptake of new services.

What measures are you working on?
KUBOTA: The strategies developed based on the prediction models will enable us to make business recommendations and improve customer service based on individual customer needs through DoCoMo stores and call centers. For example, when customers who are likely to cancel their contracts visit a DoCoMo shop, staff will have three or so options displayed on their terminals to help them decide what the best recommendation might be. Suggestions may include introducing a suitable payment plan or applying for a DoCoMo credit card.

As far as customer service is concerned, for members of "DoCoMo Premier Club," we are now providing an extension of the period covered for repairs and free battery replacement.

Using the predictive models, we can estimate the ROI of the steps that will be taken to prevent contract cancellation, and we can conduct timely pre-emptive measures against contract cancellation. Moreover, the validity of models is constantly verified – and repeating the verification cycle makes it possible to further increase the accuracy of the correlation models.

Is there a formula for success of the BICC? It seems that simply setting up a BICC would not necessarily mean BI would be enhanced.
KUBOTA: Because the business divisions have the needs to introduce more appropriate strategies based on deeper analyses, I think it is important for the BICC to actively support their initiatives and make suggestions.

Finally, how did you use the Information Evolution Model proposed by SAS for a BICC?
KUBOTA: The organizational structure for BI deployment and business performance improvement is described in Information Revolution  written by Jim Davis, Gloria J. Miller and Allan Russell. Using this Information Evolution Model, we can objectively explain to our top executives and related organizations the various BI issues and initiatives, such as how our company is now on Level 4 and is aiming for Level 5, etc. I also think that when it comes to exchanges of opinion concerning BI with other businesses in Japan and overseas, this Information Evolution Model will probably be useful as a practical global benchmark.

Copyright © SAS Institute Inc. All Rights Reserved.