In-Depth

Sprint Mines Marketing Gold

Sprint's new data mining system yields rich stores of customer information.

Strong customer relationships are vital in the competitive telecommunications industry, because customers are notoriously fickle. That's why telecomm provider Sprint considers customer support and marketing programs to be of high strategic importance.

Sprint depends heavily on highly targeted marketing efforts to reach its 23 million existing customers. The company's Global Markets Group in Dallas relies on database marketing and customer relationship management (CRM) systems. In fact, the data warehouse and front-end analytical tools that currently underlie those systems are the firm's second effort in less than six years. "We outgrew the information our first system provided and we became more sophisticated as marketers and sales folks. We needed to move to the next level," says April Killion, group manager of Sprint's Marketing Information Solutions (MkIS) Group in Dallas.

"We spent a lot of time with our company's internal customers in sales and marketing, gathering their requirements and finding out what would help them meet their needs, satisfy their objectives and help them do their jobs better." In fact, she adds, the business requirements formed the foundation of the new system; without them, "you set yourself up for failure."

The success of the group's first CRM system turned out to be a fairly convincing argument for the upgrade, says Craver. "The first platform was limited. It was small and we'd outgrown it. But at the same time, we'd also shown the value of CRM in general. We told top management we were going to take the next step, and we explained what the benefits were."

Data Warehousing Heating Up

Despite the fact that data warehousing is more than a decade old, new implementations of the technology are growing at a healthy clip. According to figures cited by SAS from The Data Warehousing Institute and Deloitte Research, IT managers ranked data warehousing second on their list of important technologies, behind the Internet and ahead of e-commerce. SAS reports that sales of its core data warehousing product, Warehouse Administrator, were up 66 percent through August this year, on top of 87 percent growth in the company's fiscal 2000.

"We looked at vendors that offered out-of-the-box or semi-out-of-the-box solutions, but we didn't find one that we felt was going to be robust enough to support our business," says Craver. "We already had some experience with SAS, and when we looked at their tools, we realized they actually do offer a way to do everything. Our whole ETL process is in SAS. The front-end process, the query tools—they're all in SAS."

End-user Oriented
Start-to-finish, system development took nine months to a year, Craver says, with different parts of the team focusing on data modeling, setting up the Oracle database and creating the front-end tools. MkIS front-ended the system with multidimensional database (MDDB) cubes that let users explore warehoused information and build their own reports. Hooks built into the system will automatically snag and delete old data.

Approximately 50 different data sources feed into the data warehouse, including mainframe files from legacy applications and files from other applications on a wide variety of platforms. "We're probably the first people, from a marketing standpoint, to bring all this data together and give the marketing teams a truly integrated view of all their customer data," says Craver, "On top of that, we have written a custom front-end application that insulates users from having to know the ins and outs of the Oracle database itself.

"We did have to train our support staff and our analytical team to use the new application, however," Killion says. "We had to teach them where things were and how to use the new information."

Reaction to the system so far has been enthusiastic, according to Killion. Users are particularly pleased by their ability to pull together customer data easily. They're using the system to manage campaigns and judge their effectiveness. Coincidentally, the company's executives use the system to check on market characteristics and business processes. That's reinforced their support.

"We've certainly improved our ability to target-market. We've been a mass-marketing shop in the past, but we've become very targeted, especially in times such as these when companies are cutting back on advertising expenses.

"We've been able to go after customers and prospects who appear to be more valuable, who appear to be more worth the marketing dollars we spend to reach them," Killion continues. "I can truly say we've honed in on something that will increase our efficiency—not just from a productivity standpoint, but from an expense standpoint as well."

Details: Sprint's MarketMine

Organization: Sprint

Goal: To allow Sprint planners and marketeers to mine a rich and multi-sourced store of customer information easily and efficiently.

Business: Started as the Brown Telephone Company as a Ma Bell rival in 1899, Sprint evolved into a long-distance provider in the 1980s and made a name in digital wireless communications in the 90s. Since then it's aggressively expanded into Internet and data communications services.

The Sprint Team: Sprint's Marketing Information Solutions group (MkIS), a 13-person team charged with implementing and supporting marketing-related technology. The team came to this project with one Sprint data warehouse project—the existing warehouse they'd be replacing—under their belts.

Team Leaders: April Killion, Group Manager, MkIS; Flint Craver, Systems Development Manger

Location: Dallas, Texas

Web Site: www.sprint.com

Scope: About 50 data sources detailing the buying habits and trends of 23 million customers feed into the warehouse, including legacy mainframe files and applications on a wide variety of platforms, including data from HQ's NCT Teradata warehouse.

At 18 months the new system comprises about a terabyte of data. Plans are to keep three years' worth of customer information in the warehouse, which estimates place at about 2.5 terabytes.

Solution Requirements:

  • Single, out-of-box solution encompassing everything from extract-transformation-load (ETL) to front-end data analysis
  • Enable users to explore the data warehouse to pull customer data together easily
  • Automatic identification and deletion of old data
  • Exceptional knowledge transfer/training capabilities
  • Flexibility to add new technologies as they arrive, and to evolve existing data warehouse into new shapes as marketing needs change
  • Support a variety of front-end access points, including fat clients, Web browsers.

Solution:

  • SAS/Warehouse Administrator R2.2
  • SAS's bundled application, "Dataflux," to manage data cleansing and ETL
  • Platform Computing's LSF Job Scheduler, which ships with Warehouse Administrator and performs scheduling and load-sharing
  • Oracle database

Timeframe: Counting requirements-gathering, implementation took about 18 months, with actual system development taking less than a year.

Results: MarketMine, the finished project, has so far saved an estimated $1 million in outsourced data analysis contracts. Sprint has gained new metrics for measuring the success of its marketing dollars.

About the Author

Bob Mueller is a writer and magazine publishing consultant based in the Chicago area, covering technology and management subjects.

Must Read Articles