Add Starbucks to the cadre of companies trying to boost sales with artificial intelligent based assistants. The coffee company this year will let you to place orders via voice command or messaging to a virtual barista, the next step on its journey to better personalize customer service
Starbucks CTO Gerri Martin-Flickinger introduced the My Starbucks Barista, a new feature in the Starbucks mobile app, in a video demonstration during the company’s December investor meeting. In the video, a woman spoke to her phone to order a highly customized espresso beverage. The virtual barista responded to each voice command with text messaging prompts. The technology will roll out in selected regions first on iOS in a beta in early 2017 and be made available to more iOS and Android users on a rolling basis.
Starbucks joins Domino’s Pizza and Taco Bell in the crowded field of quick-service companies providing virtual assistant technology via voice and messaging. Other enterprises such as FedEx and Capital One are trying to facilitate a better customer experience by embracing Amazon.com’s Alexa technology. Forrester Research says that percent of the Fortune 500 is using chatbots of some form is climbing, with 31 percent planning to deploy one this year.
Steep learning curves for virtual barista
Starbucks has its work cut out for it in ensuring a solid user experience, given how much consumers challenge human baristas with beverage order customization. That may require consumers to learn to curb their ordering vocabulary, says Forrester Research analyst Julie Ask.
“Think about the number of times you’ve been in a Starbucks… let alone that it is a noisy environment,” Ask tells CIO.com. “The human being listens, repeats, ensures he/she has the right order based on corrections from the consumer. The human asks for missing information: What kind of milk? How many shots of hazelnut syrup? What size? It’s still quite hard to give an open-ended order to a machine.”
Ask adds that if Starbucks allows consumers to place open-ended orders that don’t require six follow-up questions to clarify they’ll have “done something really impressive.”
Starbucks’ assistant builds on an AI approach the company is using to jumpstart personalization for a rewards program that, while successful, had been surprisingly low-tech. Martin-Flickinger, who joined the company from Adobe in 2015, says that the company was hand-crafting about 30 variants of rewards offers in spreadsheets until early 2016.
To support the company’s move away from transaction-based rewards to spend-based rewards, Starbucks in June implemented an AI-based personalization engine that generated 400,000 variants of offers per week in real-time. In October, the company introduced one-to-one offers tailored to each user based on his or her purchase history directly within the Starbucks mobile app. “There are no predetermined algorithms around specific offers. It is uniquely generated based on every individual customer’s behaviors and anticipated behaviors,” Martin-Flickinger says.
Such dynamic customization will enable Starbucks to upsell customers at the point of purchase. For example, when you purchases a latte via the mobile app, Starbucks will recommend you pair it with a food item from its bakery menu. The food item suggestion will be pulled from items you have purchased from the mobile app in the past.
Starbucks’ digital personalization efforts are paying off as the personalized email offers have more than doubled customer response rates over previous segmented email campaigns, according to Matt Ryan, global chief strategy officer. Rewards program members has grown to 12 million in 2016, up 18 percent from a year ago, and retention rate for the program is now 94 percent, which Ryan attributed to the switch to spend-based rewards.
Giving the digital flywheel a spin
The efforts mark an expansion of what Starbucks calls its “digital flywheel,” an ecosystem that leans heavily on algorithms and automation to drive sales. The flywheel includes rewards and personalized offers as well as simplified payment and efficient order processes. If the flywheel operates as planned, Martin-Flickinger says that customers traveling on an intercontinental flight will be able to order coffee from their smartphones when they land at an airport tarmac, and pick it up on their way to baggage claim.
“You don’t have to worry about loading a different app, you don’t have to worry about a different rewards program or a different ID or different currency, it’s just one seamless experience, regardless of ownership model or country,” Martin-Flickinger says.
To do this, Starbucks will consolidate on one commerce cloud its various technology platforms, including point-of-sale (POS) and back-office accounting applications, 22 iterations of its mobile app and its mobile order and payment systems.
“What the magic here is that some of the logic that up to this point of time has live in the POS or back-office systems is being migrated to a cloud agnostic system, which will allow us to move it globally and to scale and to have elasticity and stability that we wouldn’t be able to have in our own data centers,” Martin-Flickinger says. By 2019, Starbucks expects to have 80 percent of worldwide stores operating on the digital flywheel model.
Starbucks is making other changes to get stronger on the digital front. COO Kevin Johnson, who previously worked as CEO of Juniper Networks and held several leadership roles during his 16 years at Microsoft, will replace Howard Schultz as CEO in April. To accelerate the construction of its digital flywheel, Starbucks last month hired Tal Saraf, previously of Cisco, Microsoft and Amazon Web Services, as its senior vice president of engineering and architecture.