pio3 - Fotolia
Smoking meats in a barbecue restaurant is a messy, hands-on job that doesn't lend itself to using a keyboard to look up operational data in BI dashboards. To get around that, the Dickey's barbecue chain is cooking up a new analytics recipe with voice-based querying as an added ingredient.
The company, which has almost 600 restaurants in the U.S., plans to deploy an application on top of the Amazon Alexa voice assistant to let its franchise owners get information on restaurant operations verbally while working in their barbecue pits. For example, they could ask Alexa for up-to-date data on daily sales, inventory levels, scheduled deliveries by suppliers, customer ratings and even optimal smoking times, said Laura Rea Dickey, CEO of Dallas-based Dickey's Barbecue Restaurants Inc.
BI dashboards with data on operations, marketing and customer sentiment will continue to be the heart of the analytics system for information delivery, said Dickey, who was the family-owned chain's CIO before becoming CEO at the start of 2017. She still expects franchisees to primarily access the dashboards via tablet devices to do longer-term analysis and planning when they aren't barbecuing.
But using Alexa will provide a more viable interface for real-time interactions with the data Dickey's makes available, without forcing franchisees to "unglove or step away from the pit," Dickey said. The voice technology can also be tapped to tell them when they need to do certain things in the cooking process, such as reminders on food-safety procedures and waste management controls, she added.
Restaurant owners are required to get certified as pitmasters through an internal training program, and most do work their pits, according to Dickey. In addition, the Dickey's Barbecue Pit restaurants have an open-kitchen design that doesn't give owners the opportunity to cloister themselves in a backroom, even if they might want to. "There isn't an office with an office manager looking at spreadsheets," Dickey said. "That's not barbecue."
A barrier to entry on analytics
As a result, having to use a tablet to analyze operational data in the dashboards was a barrier to expanding a cloud-based analytics platform -- launched by Dickey's in 2015 -- into the restaurants. The architecture, called Smokestack, combines the Amazon Redshift data warehouse with Syncsort's data integration tools and Yellowfin's BI and dashboard software. The first phase of the project focused mostly on pulling together siloed data sets to enable better analytics at the corporate level, Dickey explained.
After that was completed, iOLAP Inc., a consulting services company that helped build Smokestack, pitched Dickey's on using a new voice environment it had created as an interface to the analytics platform in meetings. That's a future possibility, Dickey said, but she and other executives initially saw the voice capabilities as a way to make it easier to use Smokestack "at the front of the house."
But there are some challenges involved in using Alexa there. The company began testing the Alexa application in February 2017 and is "still working out some of the nuances" in a dozen restaurants, Dickey acknowledged. That includes testing different hardware configurations -- for example, ones with and without a screen or Bluetooth connection.
Another issue, she noted, is finding an optimal location for the voice devices to protect them from smoke, grease and ambient noise while letting owners and their pit crews interact with the application in a convenient way. Dickey hopes to begin a production rollout late this year, but said it could take into 2018 depending on how the tests go and the final decision on the hardware setup.
Several other organizations, including Dallas-Fort Worth International Airport, are also working with iOLAP on voice applications using Alexa, but the one at Dickey's is closest to being production-ready, said Doug Krauss, iOLAP's vice president of advanced analytics.
Questions about questions
Despite Alexa's artificial intelligence capabilities, getting it to understand all the different ways that Dickey's users could ask a question has also been a challenge, Krauss said. The development team at iOLAP does provide some guidelines on how to phrase questions, but that only goes so far. "You can't just say everybody has to ask it the same way, because that's just not reality," he explained.
For now, the developers are manually building as many as 40 variations of the same question into the Dickey's application, according to Krauss. They're also working on a custom machine learning and natural language processing component aimed at improving Alexa's ability to determine the meaning of unfamiliar or unclear phrases; that's expected to be ready for testing sometime in 2018.
The voice application connects directly to data in Redshift, same as the BI dashboards do, Krauss said. But, he added, iOLAP is exploring add-on functionality that could let users access data visualizations in dashboards and pull information from them via voice commands. It's also working separately with another customer in the energy industry on a voice-based feature for launching dashboards and reports.
Whether it's through dashboards, voice technologies or other types of tools, organizations need to build BI and analytics systems that are a good match for how they utilize data, said Donald Farmer, principal of consultancy TreeHive Strategy. As the project at Dickey's shows, "there's no one way to do this," Farmer said during a presentation at the 2017 Pacific Northwest BI & Analytics Summit.
More on developing enterprise applications with Alexa
Tableau users point to effective dashboard design practices
AI chatbots can aid customer service when used judiciously