|
Page 2 of 3
A Full Plate
Brinker began the process of consuming and analyzing more data about three years ago when it brought Sullivan on board to head-up a revamp of its business intelligence systems. Prior to joining Brinker in September of 2005, Sullivan was a consultant for ThinkFast Consulting (now Palladium), a firm specializing in business intelligence systems. Sullivan was actually visiting Brinker on a sales call when he heard about their plans to expand their BI capabilities and hire a manager to run it internally.
At the time Brinker was collecting data from its restaurants from its custom-developed POS system, and the kitchen display system, referred to internally as the KDS. The KDS captures such information as meal cook times. The data is sent daily to Brinker's head office and loaded into a Microsoft SQL server database. While Sullivan says the system was fairly good a collecting data, it wasn't so good at serving it up for analysis.
The company had developed an application called EatSuite, which allowed managers to view summary data from restaurant operations, such as daily sales, cook times, and sales of certain products, but digging deeper was more difficult. If a manager wanted to view information that spanned several quarters, for example, the IT department had to build special intermediate tables and extract information for analysis.
After Sullivan came on board, he began the process of looking for a new business intelligence platform. The company's goals were well defined: it needed a reliable platform that could grow with the business, that could serve up data in close to realtime, support ad hoc analysis, and that didn't require a lot of administrative support. "We are a very lean IT shop," he says. "We don't have a lot of resources for DBAs [database administrators] or administrative tasks."
Over a tight, two-month evaluation process, Sullivan's team looked at systems from Microsoft, Oracle, Netezza and Teradata. The choices were eventually boiled down to Netezza and Teradata, with Teradata getting the final nod. Sullivan says one of the key factors in Teradata's favor is that it has a proven record in the food and retail service industries. Teradata customers include Applebee's International, Lund Food Holdings, and Wal-Mart. Sullivan also spoke with peers at Harrah's, which has a large Teradata installation powering its analytics, and felt confident it would meet Brinker's needs.
"Harrah's reaffirmed the model and that the application would work for us," he says. "They were using it in more of a real-time mode than we would ever use it."
When the business case was put forward to senior executives, Sullivan says it was based more on soft return on investment projections than hard ROI. "At the time we couldn't accurately predict what decisions could be made using the system, or the financial impact that would have," he says. "We were able to walk through some scenarios, however, and basically say here's what we have to do now to get the information and here's what we would be able to do.
"Our executives were pretty insightful—they understood the value that improved business intelligence could provide."
Implementation began in August of 2006, with the first phase of the project ready about five months later. Some bumps were encountered along the way, most involving processes for extracting, transforming and loading data (ETL). "We found that we didn't pay a lot of attention in the past to how data was being captured. How, for example, do you accurately capture cook time?" he says. The team also decided to replace the tools it was using to extract, transform and load data into the Microsoft SQL database with tools from Informatica, because the Microsoft tools were more tuned to work with SQL.
The Teradata data warehouse has been live since January 2007, and uses WebFocus from Information Builders as the primary front-end tool to run queries against the Teradata warehouse.
|
