OR/MS Today - October 1997 - Practice Lesson

OR/MS Today, October 1997

Taco Bell Finds Recipe for Success

By Nancy Bistritz

Everyday, the quest begins for millions of men and women around the world — to find the cheapest and most economical meal as fast as possible. The selections are endless, and with those unlimited choices stems a new brand of competition among fast food companies that attempt to lure the public into their establishments by offering everything from toys to millions of dollars in cash prizes.

But aside from the special attempts to attract new customers is the daily strategy it takes to keep faithful customers coming back for more. That's why in 1988, 26 years after Glen Bell opened Bell's Drive-In stand in San Bernardino, Calif., Taco Bell Corporation introduced its "value" concept — what corporate executives say was one of the most ingenious marketing strategies the industry has ever seen. Among other things, the value concept slashed prices, provided free drink refills and created a need for Taco Bell to reevaluate the way it was handling its customers and its labor management system.

Approximately two years after the birth of the value strategy, Taco Bell Corporation assembled a task force that was forced to examine ways the company could better handle the new influx of customers while still keeping its labor costs down. In order to do so, the company needed to predict customer arrival times to more accurately schedule for demand. That's where the concepts of operations research and management science came into play.

According to Jacqueline Hueter, manager of Restaurant Systems Engineering at Taco Bell, it was the idea of being able to schedule employees to meet customer demands that convinced the task force that operations research methods would be a viable solution. "Forecasting scheduling to meet your capacity needs requires something more than a two-dimensional spreadsheet — which was very common in the industry at the time," Hueter said. What the company needed was a system that took into account the entire capacity requirement element.

With no labor management system that could make these new predictions, the company was forced to start at ground zero with nothing more than a simple objective: to manage the critical balance between labor and customer sales/service — the key to surviving in the fast food industry.

"Everyone's had that experience at a fast food restaurant where you sit there and you sit there and you sit there. You go in thinking, 'OK, fast food. I don't have a lot of time. Should take five minutes or less,'" Hueter said. "Then it winds up taking 15 minutes because the restaurant's production capacity can't keep up with demand. Eventually, the consensus was reached by the task force that these capabilities could only be provided through the development of a sort of integrated, model-based labor management solution to address larger strategic issues."

It was at this time that William Swart joined Taco Bell. Currently Dean of the College of Engineering and Technology at Old Dominion University in Norfolk, Va., Swart had extensive experience in the application of operations research methods in the fast food service industry. His claim to fast food fame include being one of the first operations researchers/industrial engineers at Burger King in the late 1970s. His success stories at Burger King — for which he was an Edelman Award finalist in 1981 — included developing "simulated" restaurants that could predict how menu changes and more sophisticated cash registers would affect customer service, and developing a linear programming model for purchasing beef that immediately saved the company $1 million a year.

This experience would help Swart and others at Taco Bell to develop a system that was both reliable and predictive. "In order to guarantee that customers will not have to wait in line longer than three to five minutes, we had to plan ahead of time," Swart said.

The company created the SMART (Scheduling Management and Restaurant Tool) Labor Management System consisting of three models: a forecasting model, a simulation model and an integer programming model — each one dependent on the other in order to achieve success.

SMART Ingredients
The first model in the SMART LMS is the forecasting model which is used to "predict" customer arrival times (sales) so that managers can accurately schedule crew members. "You can't staff the restaurant for yesterday," Swart said. "You have to staff it for tomorrow. And if you staff it for tomorrow, then you have to have an idea of what your sales are going to be tomorrow." Swart and the Taco Bell team needed to know how many dollars worth of business each store would bring in throughout each day.

However, unlike a manufacturing plant which make predictions in terms of years, months or weeks, Taco Bell was making predictions based on 15-minute intervals. Fifteen minutes became the "magic number" because of "spiking" arrival times, particularly during the hours of 11:30 a.m. until 12:30 p.m.

But simply knowing how much business a store would have the following day was not enough. Managers needed to know how many employees would be needed in the store and where they should be positioned throughout the facility. This is where the discrete event simulation model assisted managers in developing labor tables or staffing tables.

Because there are various types of Taco Bell facilities (those with drive-thru capabilities vs. those without, older facilities vs. modern facilities) as well as different menu items, developing store-specific labor tables is crucial to a restaurant's success. According to Hueter, even a menu item can play a significant role in how a restaurant's staffing table should be designed.

"We don't sell french fries at Taco Bell in the United States, but we do in Canada," Hueter said. "French fries are more labor intensive, and they are a very different system. A store that runs french fries gets a very different labor table and has a very different labor demand — for the same number of transactions — than a store that does not run french fries."

Additionally, the model had to take into account the unique method of food preparation that is used by Taco Bell. Unlike other fast food restaurants that prepare the food in advance, Taco Bell is an "assemble-to-order" facility meaning that food is prepared once the customer has ordered. "They (other fast food restaurants) can serve you in five minutes or less because they start building your food before you come in," Hueter said. "We build to order which makes the challenge that much more important to us. That link to customer arrival is even more critical at Taco Bell than it is to other companies."

Another key variable the simulation model takes into consideration is the randomness of the business. While it may be ideal to think that a store with 60 transactions meant 60 customers walking in every minute, that is rarely the case. "What the simulation model allows us to do is say, 'Okay, for so many transactions in a certain type of environment with certain types of equipment and a certain menu, here's how many crew members you need to have, Mr. Manager, and here's where you should place them,' " Hueter said.

Once a prediction about the next day's sales has been made and a labor table has been established, managers need to know what the schedule should look like. The last model in the LMS — the integer programming model — answers the questions: how many people do I need on the schedule for a particular day and what should their shifts look like?

"What the integer programming model does is allows the manager to meet a store's customer service demand while simultaneously minimizing payroll and managing that staff-up, staff-down challenge," Hueter said, adding that the integer programming model also takes into account all of the customer service-related responsibilities from preparing the food to checking the order to making change. Furthermore, the integer programming model enables the manager to accurately schedule other tasks such as cleaning and maintenance and even the managerial responsibilities such as money drops and drawer counts.

Managers are able to access this information using a computer in the store which displays a graphical color representation of what the total labor will look like throughout the day. In addition, it will display each crew member's schedule (including when they are scheduled to take a break) and where they should be positioned throughout the day.

Making the Sale
When Swart first began working with Burger King in the late 1970s, operations research was not widely heard of or used in the industry. As one of the first operations researchers at the company, he quickly learned that selling is often the toughest part of the job. He often took his sales approach to extremes — such was the case when he built a simulated restaurant in an abandoned warehouse to prove that what the computer was predicting would actually happen in a restaurant environment.

"You're constantly dealing with individuals who need to be convinced that what you're doing is the right way to do it, and some of them are never convinced," Swart said, adding that being confident in OR methodologies and techniques helped him both while working with Burger King and Taco Bell.

Similar to Burger King, the Taco Bell team required much convincing, but was definitely receptive to new ideas. "The value idea was followed by an enormous growth spurt," Hueter said. "I think the whole company woke up and said, 'What we're using today isn't working. We're probably losing customers.' I think Taco Bell recognized it couldn't continue doing business the way it was doing business. It was not a good long-term strategy. That's why they deferred to the experts." The expert in this case was the engineering services department that led the labor management system task force. The engineers were able to derive fact-based solutions to issues the company was rapidly facing.

While convincing Taco Bell that OR was the answer did not prove to be an obstacle, implementing the system in all company stores and then training personnel proved to be a tremendous challenge. Currently, the system is being used in all company-owned store, and the system has steadily won the approval of the franchised-based stores. "Franchisees, being entrepreneurs and wanting to use any system they see fit to best manage their business, have the option to choose the SMART LMS, but are not forced to use it," Hueter said adding that from 1993 to 1996, the percentage of franchises that invested in the system went from 0 percent to 70 percent.

Perhaps the biggest obstacle faced was the availability of data. Prior to the SMART LMS, no other labor management system of its kind had existed. In addition to knowing how many dollars a particular store would bring in on a particular day, the Taco Bell team needed to know much more specific details such as how long it took to prepare a particular menu item — data that is rarely collected. Researchers literally went into restaurants and timed how long it took to prepare a meal, much like Swart did while working with Burger King.

In addition to a massive savings in labor costs (in excess of $40.34 million from 1993-1996), the company also saw its operations research department grow as well. Formerly a two-person group, the industrial engineering team is now a Restaurant Systems Engineering Department. There, four individuals with training in operations research and industrial engineering devote half of their time to using the LMS to support a number of activities designed to improve efficiency within the company.

Hueter said feedback is positive from employees and amazement from other fast food companies. "Competitors say, 'Wow, how did you do that?' I see what we have as a major advantage over the other players out there."

Nancy Bistritz is Managing Editor of OR/MS Today.

For more information, put the number 1 in the appropriate space on the Reader Service Form

E-mail to the Editorial Department of OR/MS Today: orms@lionhrtpub.com

Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com