Watch all the Transform 2020 sessions on-demand here.
Restaurant chain TGI Friday’s (TGIF) used AI at many levels of its marketing operations to double its off-premise sales in just the last year — adding up to about $150 million in revenue, according to VentureBeat’s estimates.
The strategy included using three main software tools, each of which leverages a different sort of AI: Amperity, Aqua: Conversable, and Hypergiant. The chain’s strategy shows that brands don’t have to hire big, expensive teams of data scientists to build AI-driven products themselves. Rather, they can buy software and advice from the outside to get it done — and the results may even be better and faster.
TGIF chief experience officer Sherif Mityas explained in detail how he achieved such impressive AI-powered growth during our livestream interview a couple of weeks ago. We decided to dig a little deeper today to document the ways each of these software tools leveraged AI for the company.
Stitching customer data together
The first product we looked at was Amperity, which has been in the market only nine months and became the base for much of TGIF’s AI-powered marketing. Amperity stitches together the customer data from TGIF’s multiple applications — from emails to the company’s app, loyalty programs, and in-store receipts. By merging receipts and loyalty program information, Amperity knows that a specific customer has ordered chicken wings in the past and tends to order at 5:30 p.m. TGIF’s bot, Conversable, can then make a personalized offer to that customer via a text notification.
June 5th: The AI Audit in NYC
Join us next week in NYC to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.
We got in touch with Amperity CEO Kabir Shahani and asked him to talk with us about exactly how his product works (livestream is below).
Amperity uses a variety of machine learning (ML) techniques, including decision-tree approaches, and neural networks, all operating under its own hood. It has been trained on “billions of pieces” of context about people, says Shahani.
Take, for example, Amperity’s reconciliation of profile information residing across TGIF’s applications. The tool uses machine learning to make sure those profiles are as complete and integrated as possible, mostly by using TGIF’s own data (Shahani goes into a lot of detail in the livestream). But it also accesses third-party public datasets to help determine things like the likelihood that two profiles with the same name are the same person, said cofounder Derek Slager. In Middle America, public data may suggest that John Smith is a popular name, for example, and Amperity’s AI algorithm could take that into consideration.
Amperity is a Customer Data Platform (CDP), a class of company that has emerged over the last few years as more and more businesses invest in personalized marketing and customer service. In fact, the number of CDP companies has doubled over just the last year.
One voice across all channels
Chatbot technology company Conversable supports all of TGIF’s messaging channels, including Facebook, Twitter, Amazon, and Alexa. It uses natural language processing (NLP) to understand what customers are asking, rather than relying on FAQs, like many chatbot companies do, and forcing a customer to select answer A, B, or C. Instead, Conversable will actually create new content (answer D) — based on dialogues it has learned from the past — that drives a guest to take one of TGIF’s desired next actions, from ordering food to making a reservation.
Virtual Bartender
Finally, Mityas said TGIF called upon another company, Hypergiant, to brainstorm and build a product that could give its guests a differentiated and engaging experience. Collectively, Hypergiant and TGIF came up with the Virtual Bartender, an AI-powered service that mixes drinks for TGIF’s bar customers based on their unique tastes, mood, and past behaviors. Dubbed “Flanagan” after Tom Cruise’s character in Cocktail, the service created 300 different taste profiles to mix ingredients for. Flanagan is just TGIF’s first step in creating a personalized concierge service that moves to the dining room and beyond, said Hypergiant CEO Ben Lamm. It also will help TGIF more accurately predict staffing, stock, and profitability.
TGIF’s chief experience officer, Sherif Mityas, will be speaking at Transform, our event in San Francisco on August 21-22 about AI and marketers. Our theme is “You can do it too!”
