Yann Decre, business intelligence director and chief data officer, Bouygues Telecom

Yann Decre, business intelligence director and chief data officer, Bouygues Telecom


Case Study

A Smarter
'Data Plan'

Bouygues Telecom initiates a cost-effective governance approach without compromising agility.

In a saturated mobile phone market, competitors must fight aggressively for profitable customers while vigorously cutting costs and monitoring performance. To help meet these needs, Bouygues Telecom, the third-largest wireless operator in France, implemented an enterprise data warehouse (EDW) from Teradata.

By transforming customer information into enterprise intelligence, the EDW was designed to help the telco enhance agility, improve service and respond quickly to market change.

Yet the EDW deployment was no one-and-done effort. Converting all of the more than 300 data marts into a single enterprise view and extending tools to users throughout the organization will take nearly three years, according to the firm’s deployment schedule. To ensure that it efficiently and cost-effec­tively managed its long-term EDW roadmap, Bouygues Telecom made another shrewd move: Executives launched a budget-conscious governance program.

Executive summary

Bouygues Telecom

The company:
Bouygues Telecom, a 5.368 billion-euro (2009) subsidiary of multi-industry giant Bouygues, is the third-largest wireless telecommunications operator in France. The Paris-based company has more than 10 million mobile-phone customers and nearly 9,000 employees—2,000 of whom staff six customer-relations centers.

The challenge: To keep pace with a maturing market, the telco needed to acquire and retain profitable customers while reducing costs. Yet the effort needed to gather data from more than 300 data marts was compromising the speed and accuracy of enterprise decision making.

The solution:Bouygues Telecom deployed an enter­prise data warehouse from Teradata to leverage customer information to make smarter decisions. It then developed a comprehensive, yet affordable, governance method to ensure that the most valuable busi­ness intelligence (BI) services were deployed.

The results: The company realized a 33% cost savings on BI services while creating 88% of the services requested by users and decommissioning all of the old data warehouse systems. The new Data Lab architecture allows users to perform analytics within private sandboxes while sharing the resulting BI, which enhances enterprise decision making.

“When we began construction of the data warehouse, we realized we needed to arbitrate needs and prioritize business requirements, because there is never enough money to do everything that everyone wants,” says Yann Decre, business intelligence (BI) director and chief data officer for Bouygues Telecom. “So we needed a plan that would help us combine agility with strong governance for our data ware­house environment.”

Call for Customer Insight

For the first decade after Bouygues Telecom’s 1994 founding, the wireless market grew expo­nentially. With the subsequent saturation of the market, however, the telco’s focus necessar­ily shifted away from gaining new customers. “The priority has become retaining valuable customers,” Decre says. “To do that, we need inside customer knowledge to make good offers and create value for consumers.”

Succeeding in the maturing market required other new capabilities. To reduce costs and closely monitor performance, the company needed to be able to share analyt­ics, BI and key performance indicators.

"We achieved a 33 percent cost savings, but we were able to create 88 percent of the value that the business units requested. And we did this while decommissioning 100 percent of the old systems. So it was a successful arbitration."

Additionally, Bouygues Telecom required a centralized source of enterprise data to enable rapid response to market change. By replacing hundreds of data marts with an EDW, the company could end all-too-frequent debates over whose data was correct and allow managers to proactively tackle problems and opportunities.

Together, these business drivers convinced executives that an EDW was essential. “Cost reduction alone is not reason enough to build an EDW, because it’s a very big investment,” Decre says. “We needed to have these multiple triggers to gain consensus.”

In late 2008, Bouygues Telecom selected Teradata to deploy an EDW supporting marketing, sales, finance, fraud detection and insurance revenue processes. The data warehouse was designed to provide high-level business information for strategic decision making as well as real-time information to support daily operations. Combined with strong executive commitment to the project, Decre says the power and rich functionality of the Teradata solution paved the way for the company’s EDW success.

Arbitrating Value

The first phase of the governance process began in concert with the construction of the EDW. Supported by top management, Decre’s team collected requests for BI services from business users.

The company tried an innovative approach: Each business unit was assigned 100 BI “chips” and asked to use them to “purchase” the services it needed.

“The list of needs was huge,” he says. “Accommodating all of those requests would have put us 33 percent over budget for BI services. So our next step was to establish priorities, so we could structure activities and set time lines.”

To prioritize the list, the team needed to understand the value of each BI service to the business. However, performing return on invest­ment (ROI) calculations for each service was too time-consuming and costly. “It’s not a good investment of time to make those calculations if it means slowing down your time-to-market,” Decre notes.

Instead, the company tried an innovative approach: Each business unit was assigned 100 BI “chips” and asked to use them to “purchase” the services it needed.

Some services were requested by multiple business units. Those services marked by multiple chips were deemed more valuable to the company.

Furthermore, recognizing that the desires of all business units are not equally critical, the BI services team sought executive input. A short consultation with the telco’s president resulted in a ranking of the business units’ BI needs.

Using the chips and the rankings, the team crafted a single, weighted list of BI services for the company. “This was a simple way to prioritize the services,” Decre says. “Even better, the exercise reduced the number of services on the original list by 20 percent, almost without any discussion.”

Next, the team worked with business units to understand what data was needed to sup­port their projects. “To get maximum value from the warehouse, we wanted to make sure we loaded the data carefully,” Decre says. “By prioritizing the data in this way, we created a tool to arbitrate the value of each block of data. We also considered the contribution of each block of data to our ability to decommis­sion some of the original data marts, which was another goal.”

After a one-day workshop with repre­sentatives of the business units, the team consolidated vast amounts of customer data from hundreds of data marts into the central data warehouse. The results were impressive.

“We achieved a 33 percent cost savings, but we were able to create 88 percent of the value that the business units requested,” Decre explains. “And we did this while decommis­sioning 100 percent of the old systems. So it was a successful arbitration. ”

Shared Data, Better Decisions

Having effectively managed governance during construction of the EDW, the team began to consider the need for an ongoing governance program. “BI is a moving target,” Decre says. “Every month, the needs are dif­ferent. Governance starts to become difficult when people begin using the BI.”

Behind the Solution:
Bouygues Telecom

Database: Teradata 12

Platform: 16-node Teradata Active Enterprise Data Warehouse and Data Lab production system; 2-node disaster recovery; 3-node test; 2-node development

Users: 3,000 plus (30 business units)

DBAs: 2

Data model: Customized Logical Data Model built upon Teradata Communications LDM 10.0

Operating system: SUSE Linux 10

Storage: 367TB total

Teradata utilities: Teradata Tools & Utilities 12.0: Enterprise Warehouse Edition including Teradata Fastload, MultiLoad, Fast Export, TPump, Teradata Parallel Transporter, Teradata Manager, Utility Pack, Teradata Active System Management

Tools/applications: Products from Informatica, Microsoft Analysis Services and Oracle (ODI, OBI, Siebel Analytics)

He identified two primary goals for the second phase of the governance program: shared data and oversight. To accomplish both, Bouygues Telecom created an archi­tecture called the Data Lab, which provides each business unit with a private, controlled sandbox for conducting analytics. Any BI developed in the sandbox is funneled back into the EDW, where it can be shared.

Although a few business units initially resisted the Data Lab approach because they wanted to keep their data private, the increased power of the new EDW has con­vinced even doubters. “Very quickly, our users have gained much more data on which to base their analytics,” Decre says. “The service is much improved, and the agility gains have been huge.”

To help maintain control, IT uses surveil­lance tools to monitor sandbox activity. Keeping a close watch on user activities prevents the submission of overly large jobs, which might cause performance problems, for example. In addition, some members of Decre’s team work with users to help them embrace more efficient data-loading, analysis and reporting processes.

Moving Forward

After a year of using the Data Lab architec­ture, Decre and his team are taking steps to industrialize its BI operations—without compromising agility. In 2010, the BI services team is incorporating such enhancements as service level agreements, logical data models, and additional supervision and guidance.

Together, these enhanced management approaches will optimize performance and provide Bouygues Telecom with richer data. “It’s natural that some individu­als and business units may want to work independently,” Decre says, “but this isn’t for the best. When we share the data, we are able to make better business decisions as an enterprise.”

Your Comment:
Your Rating:

E arrived here from Boumalnes de Dades, and after a long bus ride it was a welcome retreat... no hassle, no pressure to buy anything, definitely a good way to escape the desert heat. While here we went on a 3day/2 night camel trek, staying in a berber tent, and then with a nomadic family on the 2nd night.... this was an amazing experience, and will be one of our best memories of morocco. Returning to the auberge after this was bliss, warm showers proper beds and toilets. We really had a great stay here at chez Isabelle et Rachid 00212 0667250328 Leo and Becky

11/30/2011 8:49:55 AM
— Anonymous

11/11/2010 6:39:35 AM
— Anonymous
The idea of using chips to prioritize services is quite innovative and could be applied to many other situations.

8/26/2010 3:20:28 PM
— Anonymous
Fuzzy Logix