Cultivating business growth only looks easy from the perspective of spectacular success. Consider the case of Cabela's Inc. Branded as the "World's Foremost Outfitter" of hunting, fishing and outdoor gear, Cabela's began nearly 50 years ago at the kitchen table of founder Dick Cabela. He and his wife sold fishing flies through classified ads and mimeographed catalogs, developing the mail-order business into a retail and e-commerce phenomenon.

The Cabela's team includes, from left, Ryan Coldwell, Marketing statistician; Corey Bergstrom, director of Market Research and Analysis; and Dean Wynkoop, manager of Data Management for Market Research and Analysis.

With $2.34 billion in 2007 sales, the company is now the largest direct marketer in its industry and a leading specialty retailer. Its operations include nearly 30 destination retail stores, a bank, a travel agency, magazines and a television show—all focused on helping people enjoy the outdoor lifestyle.

Yet the steady growth of Cabela's was no sure shot. Helmed by Cabela, now chairman, and his brother Jim, vice chairman, the company carefully expanded its market presence from catalog to retail and the Internet. In doing so, the enterprise developed new channels and learned how to communicate most effectively with customers in each one.

By using powerful analytics tools to sort through a huge collection of customer data, Cabela's has taken new and innovative steps to understand and market to consumers around the world. At the heart of this technology are the enterprise data warehouse (EDW) from Teradata and SAS analytics tools. "[Our] current partnership with Teradata has positioned Cabela's to provide more dynamic information to more areas of the company," says Corey Bergstrom, director of Market Research and Analysis. "We feel this expanded partnership will provide even more flexibility in terms of data availability and, ultimately, insights that can be leveraged to better take care of our customers."

Tracking down data
Collecting this data and using it to support the company's growth has been a learning process. Not long ago, Cabela's was still primarily a catalog company with a few retail showrooms to spotlight featured products. Relevant consumer data was collected and stored in homegrown customer master data management and order capture systems that supported call center operations.

To enable Internet sales that began in the late 1990s, the company built a Web site that utilized logic processes and data from those systems exposed as XML-based Web services that assisted both channels. As more stores began to sprout, IT installed a point-of-sale (POS) system to track retail data. Unlike the call center/Internet systems, the retail system collected relatively little data from customers at the POS.

Yet Cabela's knew that building the business could be done only with savvy marketing, which required enhanced insight into customer preferences and buying patterns. Because its operational systems were incapable of supporting sophisticated analytics processing, in the mid-1990s the Cabela's team built a DB2 warehouse and used SAS tools to perform analytics.

Over time, this solution became unproductive. Among other issues, the statisticians needed access through SAS to the DB2 environment, but it was not scaled to support such extensive processing. So Cabela's deployed a SAS-based data mart to handle analytics. The company augmented the sales and customer data sets with demographic and psychographic data in that SAS data mart independent from the DB2 data warehouse.

"This gave us three versions of the truth," recalls Dean Wynkoop, manager of Data Management for Cabela's Market Research and Analysis. "We had the source systems (one version of the truth) that fed into the DB2 warehouse (a second version of the truth), which was incomplete because the retail system was never properly tied to it. Then we had the replicated data mart that supported the SAS analysis (the third version of the truth)."

Users seeking insight from the data struggled. "We spent a lot of time building the data instead of actually working with the data," says Ryan Coldwell, a Cabela's Marketing statistician. The extract, transform and load process into the SAS data mart was run biweekly, but loading often took as long as two weeks. "By the time it was done building, the data could be up to four weeks old, and we'd have to start again," he adds.

In addition to the availability and latency problems, the old system simply could not provide the insight that Cabela's needed to support its growth. "With the old data mart, you had to know the problem before you could use the IT resources to construct data that would help you develop an answer," explains Wynkoop.

Seek new capabilities
In 2005, Cabela's began searching for a new data warehouse solution and quickly chose Teradata. "The Teradata mantra of 'any data, any time' rang true for us," says Wynkoop. "We're in a world where we don't know what all of the problems are. If we could get the full range of the data, we could answer the business questions."

The new data warehouse was launched in June 2007 and fully rolled out by October to the Marketing team. Cabela's used the Teradata Retail Logical Data Model as a foundation, then worked with consultants from Teradata Professional Services to modify the model to meet the company's unique needs.

The data warehouse provides insight throughout the organization. Statisticians use advanced statistical analyses to help Cabela's understand customer trends and preferences, retail performance, and even product affinity.

Because so many groups at Cabela's use the data warehouse, it was critical that no one compromise its performance by submitting overly large or problematic queries. Cabela's IT staff received training from Teradata Education Services, then taught champions such as Wynkoop how to use the system. Informal in-house training was provided to show statisticians how to properly use SAS analytical tools to perform in-database processing that would collect data efficiently without degrading processing speed.

"For data residing in the sandbox, we had to learn how to create tables with proper primary indexes to avoid skewing. We had to learn when to use implicit versus explicit SQL [structured query language] statements. We also had to learn how to collect statistics and use column compression—things normally reserved for DBAs [database administrators]," says Wynkoop. "Much of this was learned through trial and error. Each member of the team contributed to our understanding. Sharing what we learned amongst ourselves helped us to most effectively use Teradata analytics with the SAS tools."

Building in the sandbox
Members of the IT team went a step further. From Cabela's EDW, the team created a 250GB analytic sandbox for the Marketing group. Segregated within a separate database, the sandbox is a single collection of data put in place by one or more informal load events, where Cabela's Marketing analysts and statisticians can collect data and perform in-depth analyses without compromising the performance and data quality of the EDW.

Each analyst and statistician has an individual account and rights to the sandbox as well as the ability to view and join to production data to reduce replication while enhancing analytics. SAS users can place work tables in the sandbox and use them to manipulate or transform data. "With the sandbox, we do all of the heavy lifting with the Teradata system," says Coldwell. "Then we can bring just what we need into the SAS environment."

Members of the Data Management team frequently use the sandbox as a "pre-production data warehouse," where they can work with augmented data that is not ready to be added to the main EDW. They also use the sandbox to perform preliminary data preparation work, defining data requirements and determining how to use the data. The sandbox is administered by the Data Management team, which notifies Marketing users when it becomes full and requires cleanup. IT also uses workload and system management tools that signal when the sandbox consumes an inordinate amount of resources.

Unearth valuable insights
Using the sandbox and the data warehouse, Marketing analysts and statisticians have been quick to find innovative ways to leverage the technology. To better understand the effectiveness of marketing efforts, for example, the Data Management team built a small database of retail fliers in the sandbox. By linking sales data to flier data, the Marketing analysts have begun to understand which blocks of content perform well, leading to improvements in overall flier performance.

The team is also using the sandbox to assess the value of the company's advertising efforts. Conducting a simultaneous campaign with e-mail, catalogs and retail fliers, analysts are measuring the impact of each communication vehicle and investigating how each medium interrelates with other media. Being able to perform most of the data preparations without the help of IT—which is simultaneously working on other important data warehouse initiatives—saves time for IT and helps the Marketing team gain the benefits of the data that much faster.

The technology is utilized to make Cabela's more agile. With the data warehouse, the company receives each day's sales data by the following morning and can assess conditions and proposed steps—such as launching an e-mail campaign—to mitigate problems and boost sales. With the old system, producing the same information would have taken days.

"With the new data warehouse, we're able to respond to questions in a very timely manner," says Coldwell. "Let's say we need to have information out the door this afternoon. We can do that because we have one source of data, and we can access it very quickly using the Teradata environment."

Going forward, the Data Management team expects the sandbox to grow to meet its needs. In addition, Cabela's has recently expanded its overall Teradata solution to enable more extensive analysis of customer behavior and provide even more precise and personal service to its millions of customers.

"With the Teradata system and SAS, we are asking and answering questions today about our business strategies such as retail expansion that we never would have anticipated two years ago, when we started on this road," says Wynkoop. "We're able to give the business information about where we should put stores, and how we can improve retail performance. In this way, the Teradata system and SAS analytical tools have helped the company to be more responsive and agile." T

The sandbox from IT's perspective To provide business users with a safe analytics processing area, Cabela's Inc. created a 250GB sandbox for the Marketing group. Initially only 100GB, the popularity of this space for analytics and data manipulation quickly warranted several expansions. Using the sandbox, users can create, drop and delete tables, views and procedures within the database. IT segmented the sandbox, using authorities to create role-based access. To prevent conflicts, they worked with the analyst community to ensure that long-running processes are not submitted during backups of the data warehouse. The IT department uses Teradata Priority Scheduler to manage workloads. Jobs that require extensive processing are automatically shifted to lower-priority groups if they run too long. Using Teradata Manager, the IT team can monitor performance of the sandbox processing and ensure that the group is not overloading the sandbox. Analysts can build solutions that meet their needs without waiting for the team, which is working to build new areas, such as inventory, within the data warehouse. "With the sandbox, users can satisfy their requirements while we concentrate on getting more data available for the whole company," explains Craig Bruner, Cabela's enterprise data warehouse architect. "We can help productionalize those views at a later date and put appropriate controls around them. It's a great way for analysts to get the data they need while freeing us up to focus on the bigger picture of providing value to the organization." —C.D.K.

Behind the solution: Cabela's Inc. Database: Teradata Database V2R6.2 Server: 4-node Teradata 5500H Server with a "virtual" sandbox Users: 350 (200 concurrent) Data model: Logical—Teradata Retail Logical Data Model (RLDM) Physical—Third normal form Operating system: Linux Storage: Total for all systems: 32TB Teradata Utilities: Teradata Tools and Utilities 8.2, Teradata Manager and Priority Scheduler Tools/applications: Teradata Warehouse Miner and products from SAS

Cheryl D. Krivda writes about the intersection of high technology and business practices.

Photography by Curt Door/Cabela's

Teradata Magazine-September 2008