by Karen D. Schwartz

In a world filled with fast-food eateries and fancy sit-down restaurants, Applebee's International Inc. of Overland Park, Kan., figures it has a wide-open playing field. Executives at the world's largest casual dining chain, with more than 1,400 restaurants in 49 states and eight countries, aim to fill a void by positioning the chain's restaurants as comfortable establishments that patrons can visit for good food, good service and a warm atmosphere. But continually changing customer preferences and expectations, coupled with rising food, labor and administrative costs, make that goal more challenging than it sounds.

To win over customers while reining in costs and creating economies of scale, Applebee's has married state-of-the-art data warehouse technology with a host of custom-developed and off-the-shelf systems.By positioning a Teradata data warehousing system running on a two-node NCR WorldMark 4850 as its core strategic tool, Applebee's IT and business staff can slice and dice important data, helping them better examine the impact of a variety of factors such as customer food preferences, differences in staffing between restaurants, identification of promotional opportunities and menu item pricing. By analyzing the data carefully, they can better conclude what customers like to eat, how they want it served, what should be included on the menu and what should be eliminated, and which items could be improved. Analysts also can use the data to better customize menus and labor at the store and regional level.

Teradata Magazine spoke with Applebee's International Inc.'s Seth Jensen, senior manager of business analysis, and Dan Harkness, IT project manager, to find out how the restaurant chain is using technology to refine and reach its goals.

Q: What has business been like in the wake of a slower economy?
Jensen: Casual dining in general and Applebee's in particular have faired very well. Our strong price-value relationship and being close to home "in the neighborhood" have proven to be advantages in this environment.

Q: Is technology the answer to retaining that boost?
Jensen: In many ways, yes. The best way to retain that edge is by making operations more efficient and getting more people in the door. We can do that by looking at how labor is allocated, if we are using hours in an efficient manner, how well we are controlling food costs, and whether we are offering menu items that people want and that give us a good return. We can use technology in general—and data warehousing technology specifically—to help us reach many of our goals. But we also have to keep in mind that an emphasis on good service—making sure the food is hot when it arrives at the table and gets to diners quickly, as well as making sure servers are giving passionate, attentive service to our guests—is about good, old-fashioned hard work.

Dan Harkness Title: IT project manager, joined Applebee's in 1996 Education: Degree in computer science First computer: Commodore 64

Seth Jenson Title: Senior manager of business analysis, joined Applebee's in 2000 Education: Degree in accounting and finance Passion outside of work: Getting his pilot's license

Q: What are your main business goals?
Jensen: We're always trying to put the right products in front of customers at the right price. By doing so, we ultimately increase frequency and customer satisfaction. We also want to know more about regional product preferences and how that affects product mix, and we want to know more about the disparity of what it costs to staff each restaurant. We can use Teradata's data warehousing system to address all of these goals by developing financial models, which we can use to determine which products are strong performers and which should be replaced or enhanced.

Q: Before moving to Teradata, were you able to adequately obtain and analyze this type of information?
Harkness: Before 1998, when we implemented a single-node NCR WorldMark 4700-series system—since replaced with a two-node WorldMark 4850 system with 360 gigabytes of storage—we did what we could with paper-based, fixed, hard-coded product mix reports. We would print out about two feet of paper each week and distribute it to analysts, who would each look for the one line they wanted and re-key it into a spreadsheet. It was very time-consuming to run the reports, and access to those reports was very limited. We knew we had to replace that process with a flexible, drill-down analytical system so we could more easily and thoroughly analyze our data.

Q: From what systems does the data warehouse pull data?
Harkness: We have several different systems that communicate with the data warehouse, including our custom-developed Point-of-Sale systems installed at each restaurant, which provide a lot of valuable information such as what people are buying and average sale price of those products. It basically establishes a store-by-store local customer preference that's fed into the data warehouse. It also pulls information from a food cost management system from Eatec Corp. and a custom-developed labor management system. The data warehouse also draws data from our financial and customer service systems.

Q: How do you use the current system to determine the best product mix?
Jensen: If we're testing a porterhouse steak in Texas, we could determine how that product would perform if we introduced it system-wide. We could divide the number of times the item was sold by the guest counts in the testing timeframe and get a percentage of guests who ordered that product. If the answer was 10%, logic would tell us that because we tested the product in Texas, there might be a higher preference for that product in Texas than there might be across the system. Now we can use the data warehouse system to analyze what steak products guests ordered in Texas over the last two years versus what guests ordered in other markets. Then we can build an index using that same metric to determine whether the item will clear the threshold of profitability and contribution to our overall check system-wide. Having the data readily available makes it much easier to do these types of comparisons.

Q: Can you take it a step further and use the system to determine if a new item should replace an existing menu item?
Jensen: We have a fixed amount of space on the menu, so that's a real concern. It's all about price point. We have to figure out whether customers will spend $14.99 on a porterhouse versus $9.99 on a house sirloin. To do that, we multiply the average sales price by the incidence rate, which gives us a number that represents contribution to the average check. Using the porterhouse example, when we multiply the 10% incidence rate by the $14.99 price point, we get $1.49. That means on an average check of $10, we get $1.49 out of this particular product. In short, it helps us analyze what it all means to the bottom line. If everybody loves an item but it has a 50% food cost, that might not be the right move for Applebee's. It's about finding that valuable middle ground, where it's complementary to your food costs but preserves your bottom line.

Table talk

* Applebee's uses 152,000 cases of tomatoes every year in its restaurants around the world. If stacked one on top of the other, the cases would reach the first layer of outer space. * Applebee's customers gobble up 50 million pounds of french fries each year—enough to circle the globe 11 times. * Applebee's customers drink enough Coca-Cola beverages each year to fill a backyard swimming pool 900 times.

Q: In what way do you use the current data warehouse system to analyze labor issues?
Harkness: It helps us monitor how many people should be on staff for a restaurant during a particular time period during the day, because we can project what the traffic will look like through the data warehousing system. We can actually forecast staffing volumes in 15-minute increments, which means we can help the general manager at the restaurant level control his or her labor costs. They are now running their labor at a cost most efficient to them without hurting the guest experience or spilling costs. It also means we can predict from year to year what a particular Monday should look like in terms of volume, so the restaurant can be staffed accordingly and run at maximum efficiency from a labor standpoint.

Q: How does the data warehouse help you achieve your goals in the area of sales and marketing?
Jensen: It helps us understand and measure the campaigns we have been running, like limited-time offers and menu items we couple together to create a theme. Before using the data warehouse, we were taking information from campaign tests within a certain region and projecting it onto our system. By doing that, we could miss dramatically with our purchasing projections, because that region-specific preference didn't exist across the nation. We had no conflicting data to warn us that it wasn't the right forecast to give to purchasing. Now we can compare indexes by region and purchase and distribute more intelligently, avoiding product shortages and overruns. The purchasing department loves what we're doing. They're telling us they aren't missing the mark at all. They are telling us they are out of product just as the campaign runs out. It gives us the flexibility to go on to something new rather than worry about how we're going to burn off obsolete or excess inventory.

Q: Are you currently using the system to learn more about customers, or is the focus mainly on product mix, labor and sales issues?
Jensen: We're looking into how we could use the system to learn more about our customers, but we're treading very carefully because of privacy issues. We certainly could devise a loyalty program and use Teradata's capabilities to slice and dice the data to learn all kinds of things, but before we take that step, we have to assess if we have put additional value in our customers' mind by using information we have to analyze their spending habits, without necessarily interacting with them on a one-on-one basis. Then we have to determine if we want to be able to identify them when they come in the door, and if we want to be able to use that data about their preferences to suggestive-sell them specific items for which they have already shown a preference. It's about making sure the value proposition is good for the customer and not violating their privacy. We're still assessing whether it gives enough value to the customer to more than offset the cost that could arise from them perceiving it as intrusive.

Q: What type of analysis can you do now that you couldn't do before?
Jensen: We've done some fun stuff that supports our R&D culinary team. For example, we've put together a sheet that uses consumer satisfaction indexes (CSI), which are pulled through a different system. In this scenario, customers are financially rewarded through a small incentive discount to give us feedback on what they ate, what they liked about it and what they didn't like, using standardized measures. It calculates an aggregate CSI score on specific products, which we measure against the financial performance of those exact products to give us both quantitative and qualitative analysis. We end up with four quadrants, where one represents signature items that people consistently rate as above average, another represents items with high promotional potential, a third consists of items with potential for replacement, and the last quadrant represents items that could be reworked to offer better value. It gives a good roadmap to our R&D team as to what they should improve, what they should replace and what's doing well.

Q: How hard would it be to replicate these analyses without Teradata?
Jensen: I believe it would be possible, but it would be very labor-intensive and cumbersome, and the turnaround time would be significantly slower. By the time we got the information we needed, it would probably be too late to make a decision or react in a way that would be a cost avoidance approach to the business.

Q: Why Teradata instead of IBM's DB2 or Oracle's data warehouse solution?
Harkness: The price versus performance numbers were good, and the ability to scale also came into play. It has good linear scalability, so we felt we could grow without having to encounter a lot of growing pains. But what really stood out was the user-friendliness and the lower maintenance levels. We felt that once it was in place, it wouldn't take constant hands-on support for us to use it as a good modeling tool. Another factor was administration costs. We looked at the incremental database administrator (DBA) head count needed to maintain the size systems we were talking about, and Teradata was well below the others. This organization is very sensitive about head count, so that was important. The ability to have fewer DBAs on staff and maintain these larger systems was a huge benefit to us.

Digesting the data Behind every data warehouse is a great team. From upper left: Randy Parman, database architect; David Wilson, finance project manager; Donna Lamano, query developer. Applebee's Overland Park, Kan., headquarters is the nerve center of the organization, housing an Ethernet-based local area network (LAN) that connects the company's main computing systems and platforms. The main analytical tool is a two-node NCR WorldMark 4850 with 360 gigabytes of storage running Unix and housing their Teradata data warehouse. MicroStrategy 7.0, an online analytical processing tool running in a Windows-based four-tier environment, acts as the system's query agent and reporting tool. The Teradata and MicroStrategy systems connect via the LAN to a panoply of custom-developed and over-the-counter software systems, drawing valuable data that allows the technology staff to track and analyze sales, preferences and trends. Point-of-sale information from individual Applebee's restaurants is made available to the Teradata data warehouse via a frame relay, wide area network. Food-cost management information is also loaded into the data warehouse on a daily basis. Teradata draws data from several homegrown applications, including systems for sales auditing and labor management, housed on a variety of SQL Server systems. In addition, the data warehouse receives data from a J.D. Edwards financial system residing on an IBM AS/400. By next year, the company will roll out enterprise resources planning functions into a suite of Peoplesoft tools, making it easier to extract valuable data, according to IT project manager Dan Harkness.

Q: Do you think Applebee's size played a major role in the decision to build a data warehouse?
Harkness: Because we are the largest casual dining chain in the world, a tool like this becomes even more important. If we were a 10-chain store, four people could work on 2.5 stores apiece and figure out what that means and get it done quickly. But trying to compile that data for more than 300 restaurants just on the company side, and eventually leveraging franchises to more than 1,400 restaurants, would mean we'd have to hire a lot more analysts without a system like this.

Q: What kind of speed-to-value or return on investment have you seen from the system?
Jensen: We've experienced labor savings and significant dollar savings in cost avoidance on the food side, because we run limited-time offer products. Our ability to forecast demand curves for those untested, unproven products was a dangerous proposition before this system was implemented, but through analytics and the availability of data through the data warehouse, we've been able to put some reasonable expectations around it. We've tightened the parameters around obsolete inventory exposures or exit strategies on those products dramatically just by using these metrics. By using the availability of that data, we have mitigated that exposure dramatically and actually have people embracing that projection process instead of avoiding it.

Q: Is this system being used at all of your stores currently?
Harkness: Right now, only the company-owned restaurants are being loaded into the data warehouse. One of the challenges is that many franchisees have different platforms.

Q: How has data warehousing changed the way you do business?
Jensen: Without this type of system, you can still understand the importance of marrying the metrics to the data, but if you don't have the data readily available, you end up having executives making decisions about product mix based on gut feelings.

Q: Now that the system has been in place for a while, what's next?
Jensen: We'd like to add more information from our other systems. The more data we add, the more we can exponentially grow our ability to analyze, understand and improve our performance on several different measures of our Profit & Loss Statement.

Q: What are the biggest benefits for Applebee's?
Jensen: By using a data warehouse for product mix selection and campaigns, our guests' preferences speak loud and clear. It's really about making sure we are giving our customer something more and better than what our competitors can provide, and about issuing the right offer to the right customer at the right time. But the greatest feature, in my opinion, is the ability it gives the finance department to give executives what they want. If we're asked to provide certain data, the answer is typically 'yes.' Even if that means I have to go back to the drawing board and figure out what I want to query or how I want to do it, I know I can deliver a product that helps them make stronger business
decisions. T