Edge Computing Helps Feed Taco Bell’s Digital Business

At Taco Bell, a computer server at each location ingests data from in-person and digital orders and customer loyalty accounts, as well as kitchen operations, and uses custom algorithms to make decisions about, say, when to tell employees at the fryer to sink the potatoes for an order of Nacho Fries so that they are warm when a delivery driver arrives for pickup, Mr. Parizher said.

“We took [on] our most critical workloads in order processing and menu data,” Mr. Parizher said. “If it runs it more efficiently than what you had before, even by a small percentage, for a Fortune 500 company the results are significant.”

The so-called edge isn’t a place but a computing model, said Lynda Stadtmueller, senior vice president of the research, information and communication technology practice at Frost & Sullivan, a market research firm.

Key aspects are sensored devices, connectivity, analytics and responsiveness, Ms. Stadtmueller said, speaking at the same event.

The goal is to improve the performance of applications by processing data where it is generated, such as within a local Taco Bell, and applying it at lightning speed. Energy companies, retailers and industrial manufacturers are using edge computing to take advantage of fast internet speeds, including 5G networks, and a growing range of connected devices.

General Electric Co.

and

Siemens AG

, for example, are using edge computing to optimize factory machines in real time.

“A John Deere tractor is an edge when it’s equipped with sensors that maybe monitor components. Your phone could be an edge when it’s collecting data or a kiosk reaches out to it and pings it,” Ms. Stadtmueller said.

Taco Bell has spent about five years putting in edge-computing capabilities, Mr. Parizher said. Each location has duplicate devices to serve as backup in case of an outage.

Aside from walking into a storefront or using a drive-through, Taco Bell consumers can order their Mexican Pizzas and Chalupa Supremes through the company’s website, mobile app, from food delivery services and, in some geographies, via text messages. In January, the chain rolled out a subscription service nationwide. For $10, customers can buy a Taco Lover’s Pass to get a taco a day for 30 days. Menus are customized by location.

When a repeat customer places an order, the local restaurant’s system fetches relevant information from the cloud. Understanding frequent purchases, typical order size and penchant for trying specialty items helps determine whether to offer a customized deal, Mr. Parizher said.

In addition to processing and analyzing orders from a multitude of platforms, restaurants must deal with menu tweaks, food assembly and changing personnel, as well as data coming in from several channels at once, Mr. Parizher said. “You process these events and try to optimize. We’re not different from manufacturing, in some sense,” he said.

The team had to consider how to protect personal data about customers in Taco Bell’s loyalty program. “For security purposes, we don’t want that information to be at the store,” he said, so it is stored in the cloud. “You don’t keep data at the edge for any longer than you have to.”

Taco Bell contracted with a third party for a platform to monitor the overall edge-computing activity while Mr. Parizher’s own team built the software for routing order data to various connected devices in the kitchens, he said.

With edge-computing foundations in place, he said, Taco Bell can experiment with connected robotic equipment that can fry food, warm up tortillas or pour drinks. “Now it gets a bit more exciting,” he said.

Write to Kim S. Nash at [email protected]