Brinker restaurants actively analyze information to continuously improve the dining experience.
by Len Lewis
It's 6:30 on a Friday evening and the tables at Chili's in Westbury, N.Y., are filling up quickly with the first wave of diners. The waitstaff
is already in overdrive and dishes are flying out of the kitchen at a frantic pace, despite a back-of-the-house bottleneck that is starting
to slow down service.
 |
|
Kenny Sullivan, director of business intelligence (BI) at Brinker, says several factors made Teradata the logical
choice for his company.
|
|
The next day, senior executives at Dallas-based Brinker International Inc., which has become one of the country's leading casual-dining chains
over the past three decades, pore over detailed analyses of the previous night's activities at this and the company's more than 1,300 other
restaurants it operates domestically.
Kenny Sullivan, director of business intelligence (BI) at Brinker, explains how the new active data warehousing solution from Teradata makes a
difference for the company: "You can't just run a restaurant on anecdotal stories anymore—you need hard facts. And having the information at
their fingertips gives our brand presidents and CEO a comfort level with the decisions they are making and with new strategic initiatives they
want to undertake. The system has generated a lot of excitement around here—much of it centered around the immediacy of information."
This immediacy, bordering on real-time data, has become increasingly important to Brinker as its portfolio of restaurants—Chili's Grill &
Bar, Maggiano's Little Italy, On the Border Mexican Grill & Cantina and Romano's Macaroni Grill—expand in existing and new market areas and
its customer base increases exponentially. In fact, more than 1 million people dine at a Brinker restaurant each day. This activity produced
sales in excess of $4.1 billion in 2006 with operating income of $327 million.
Hungry for change
Before Teradata, Brinker's system focused on operational data, according to Sullivan. "We have a custom POS [point of sale] restaurant system
in each unit, which collects a lot of data including sales transactions, cook times and time cards. We pulled that data from restaurants every
night and loaded it into our warehouse," he says, noting that the chain's SQL Server 2000 database has been online for about five years. "It
was fairly successful in capturing data but didn't meet our needs for actually analyzing and reporting it."
And despite having several terabytes of capacity, the company was having difficulty keeping up with the volume of information and providing
direct access to all who needed it. In fact, a separate application called "EatSuite" was used to pull summary data from each restaurant, but
only managers and some employees had access to this aggregated information.
"We were getting some detailed data that we put together in spreadsheets when we had special requests. But we actually had to split some of
our big tables into a separate database for every fiscal quarter. There was no way to integrate the databases because the loads would take too
long," Sullivan says. "If someone wanted information that spanned several quarters we had to build special intermediate tables and extract
the data into those."
Consequently, it was becoming increasingly difficult for corporate to do its job, according to Sullivan. "Brinker has become more strategic in
its thinking over the past several years. But the information wasn't there to support strategic analysis," he says. "We needed a way to
provide better insight from the management perspective and operational facts—not just go with a gut feel."
The "limited and frustrating" system was having an impact on the decision-making process in several areas including remodels, limited-time
offers and new product initiatives. "Anything we wanted to do that the EatSuite's application didn't support—which was mostly everything
strategic—we had to go through a very cumbersome process," Sullivan says.
Menu of choices
The company began discussing the need for a better approach to BI shortly after Sullivan came on board in September 2005. "We wanted a stable,
scalable platform that could be the core of our warehouse, could provide more open access support and had the ability to satisfy ad hoc query
requirements against large volumes of data," he explains. "The idea took awhile to gain momentum. But by April 2006 we made the decision to go
through a vendor selection process which included Microsoft, Oracle, Netezza and Teradata."
The evaluations took about two-and-a-half months. The search quickly boiled down to Netezza and Teradata. However, several factors made Teradata
the logical choice. Brinker knew it needed a solution that would give it a head start, one that it could grow into over time and would be
supported by people who were passionate about the business. "We run a relatively small IT shop and didn't have a lot of resources to throw at
setting up a new system," Sullivan notes. "So one of the things we looked at was ease of maintenance, support and stability of the platform.
In this regard, Teradata has proven itself time and again in the retail and food service industries. "Another benefit was Teradata's logical
data model, which helped Brinker speed up the development cycle. "This model has been proven throughout the industry and only requires minor
customization," Sullivan adds.
This record of success enabled Sullivan and his team to convince the chain's top management that Teradata was the right answer. "We talked
about several of the more painful analysis cycles we had gone through and about how we would apply analytical detail to all decisions in the
company. We actually sold it based on a strategic ROI [return on investment], rather than the more traditional financial ROI. We didn't factor
in dollars for what the true benefit of those business decisions would be since we couldn't quantify them. But top management knew we needed
more information and we told them we couldn't support all the strategic initiatives they wanted without a more robust platform," says Sullivan,
noting that Netezza was seen as more of a secondary platform. "We also discussed results from elsewhere in the food service industry. Our CEO
and other leaders actually met via phone with the people at Harrah's and were impressed with what they heard."
Order up
Implementation began after Brinker obtained the hardware in August 2006, and, as Sullivan expected, startup was relatively easy. The first
phase of the project was operational in about four-and-a-half months. This included replacement of extract, transform and load (ETL) processes
that were loading all the data into SQL Server because that was written in Microsoft packages. Subsequently, they were replaced with
Informatica.
"We had a couple of little bumps here and there. But it had to do with our understanding of the data. What we thought we knew—we didn't," says
Sullivan. "Overall, the data wasn't as clean as it should have been, and we found some problems in data that we never analyzed in detail
before."
One of these was cook-time information, which gives an indication of kitchen efficiency. "We have it for every item that goes through our
kitchens. We look at efficiency by station as well as overall cook time and how it impacts the guest experience," he explains. "We're trying
to turn tables during peak times and a bottleneck anywhere can be a problem. We were also looking for efficiencies in the 'to go' process.
The new system enabled us to pinpoint whether the problem was in the kitchen."
Although Sullivan knew Teradata's history with other clients, he was still surprised at the level of support. "My sales rep was fantastic,"
he notes. "He always made sure we had the right resources and staffed our project with the best people. This made implementation as smooth as
it could possibly be."
During this period, Brinker also gained some knowledge about what not to do. "One thing we learned was to be more rigorous about what goes
into the data warehouse," admits Sullivan. "We found a lot of stuff that people don't use heavily was where we had a lot of issues, and going
forward there will have to be a solid business case for what we add to the data warehouse."
Because the system has only been live since January, it is too early to gauge results. However, Sullivan notes that performance and results
have exceeded expectations in terms of capacity and loads. "Costs are in line with what we expected, and we're set up to see a very positive
return on investment. For end-user applications we simply can't afford not to have the Teradata system, and from an IS perspective we're doing
things we could never have done before," he says. "The biggest benefits lie in our ability to analyze everything, aggregate all the data and
look at it in different ways.
"We use WebFOCUS from Information Builders as our front-end tool. Additionally, we had a team from all our different brands and finance design
some initial applications and set up ad hoc reporting objects that allow them to build their own reports and queries. We don't have the analytical
apps yet, but we do have open access to information. We're kind of learning to crawl before we walk."
However, the system is going to make walking and running easier and more productive. "We're looking at a lot of ways to apply the technology.
We can't talk about them yet, but these are for strategic initiatives that will expand the data warehouse and provide better service to guests.
We are all very excited about it," shares Sullivan.
Meanwhile, training is picking up steam. "We trained a core set of 25 to 30 initial users in the ad hoc capabilities of the new platform, and
we have some slow adopters. We've started to change people's mind-sets and the entire culture within our organization. The users have learned
they can go directly to the application to get what they need faster rather than funneling every request through IS. Furthermore, the user
base is expanding rapidly as more people want to be trained," Sullivan adds. At present, the primary users are in-house brand, financial and
operational analysts who are testing new products or monitoring the health of the business.
Serving up data
What does all this mean to Brinker's future? "I think it will definitely help in restaurant expansion as well as new customer growth. If
you've read about the restaurant industry lately, the casual dining segment has taken a bit of a hit. We're working on ways to keep our
current guests excited about the Brinker brands as well as to encourage new guests to give us a try. A lot of that has to do with introducing
and testing new products to see what resonates with them. Teradata gives us immediate insight into what's working well and what needs to be
reworked in test," says Sullivan.
However, success depends on gaining operational efficiency and, as Sullivan puts it, "squeezing all we can in margins, labor and food cost
savings, without negatively impacting the guest experience. We are focused on guests more than ever before, and improving operational
efficiency is one of those things we couldn't afford not to do." T
| Behind the solution: Brinker International Inc. |
|
Database:
|
Teradata Database V2R6.1.0
|
|
Server:
|
2-node Teradata 5450 production system and a Teradata 540S Server test/development system
|
|
Users:
|
30 (20 concurrent)
|
|
DBAs:
|
1
|
|
Data Model:
|
Third Normal Form (based on Teradata's Retail Logical Data Model)
|
|
Operating System:
|
Windows 2003 Server
|
|
Teradata Utilities:
|
FastExport, FastLoad, MultiLoad, Teradata Dynamic Query Manager, Teradata Manager, Teradata TPump, Teradata Priority
Scheduler, Teradata Utility Pak - ODBC Driver, SQL Assistant, SQL Assistant Web Edition, Teradata Administrator, BETQ,
CLI, and Teradata Analyst Pak - Visual Explain, Index Wizard, Statistics Wizard
|
|
Tools/Applications:
|
Products from Informatica, Information Builders and Microsoft
|
|
Len Lewis is president of Lewis Communications Inc., a New York-based editorial planning, research and consulting firm.
Photograph by Lisa Means
Teradata Magazine-September 2007
|