How Starbucks Employed AI for a Better Customer Experience

April 20, 2020

In 2017, Starbucks launched what would become its most advanced AI-driven initiative yet. “Deep Brew,” the brand’s custom-made recommendation platform was built to reach customers across multiple channels, including the Starbucks ordering app. 

Today, Deep Brew is driving growth, providing deeper customer understanding, and allowing Starbucks to seamlessly adapt to changing customer preferences with little effort. With this foundation in place, the company is already planning several other AI-driven projects to enhance the customer experience.

Unprecedented Growth

Source

Part of the reason Starbucks executives have pushed for more AI technology is the brand’s recent growth spurt. The industry-leading Starbucks Rewards Program has continued to flourish since its introduction in 2007.According to their website, “Membership has grown more than 25% over the past two years alone, climbing to 16 million active members as of December 2018, a 14% increase over the prior year. Starbucks Rewards transactions accounted for 40% of tender in U.S. company-operated stores in the same time frame.”

Central to the rewards program is a focus on nurturing customer loyalty. A new tiered structure was created, along with new rewards options. Customers can earn points to received customized drinks, food options, or Starbucks merchandise – sometimes as soon as 2-3 visits after becoming a rewards member.

And while stocks have slipped in recent months, there’s an overall uptick in value since the beginning of the year.

“Deep Brew will increasingly power our personalization engine, optimize store labor allocations, and drive inventory management in our stores,” CEO Kevin Johnson told reporters. “We plan to leverage Deep Brew in ways that free up our partners, so that they can spend more time connecting with customers.”

How It Works

So what exactly is Deep Brew and why is it so valuable? True to its name, Deep Brew utilizes deep learning technology to gather information from unstructured data. To understand the innerworkings of Starbucks’ new AI deployment, we need a lesson in reinforced learning capabilities.

Reinforced learning is a type of machine learning that allows organizations to determine the best possible course of action in a specific situation. Simply put, it’s a type of AI that answers the question: What option should be chosen based on what is currently known? The algorithm learns through trial and error, using each piece of feedback as evidence supporting its next decision. Reinforced learning works sequentially and continuously collects customer data to create a unique profile around their tastes.

Source

Using Microsoft’s Azure cloud infrastructure, factors like location, weather, inventory, and price are taken into account for each person’s order. If a customer frequently buys non-dairy drinks, it will automatically be remembered. This technology is how Starbucks is able to recognize and adapt to the desires of its millions of customers each week. Starbucks Analytics and Market Research VP, Jon Francis sees these tailored recommendations as an extension of the in-person care a customer would receive from a barista. It further strengthens that connection and sense of trust with each buyer.

Along with personalized recommendations, Starbucks has another AI trick up its sleeve. Their Mastrena machines, originally debuted in 2008 to release shots of coffee faster, contribute to the customer experience in a more covert way:

“Those machines have Internet of Things sensors built into them. And so we get telemetry data that comes into our support center. We can see every shot of espresso that’s being pulled and we can see centrally if there is a machine that’s out there that needs tuning or maintenance. And that allows us to improve the quality of the shots that we’re pulling…with the Deep Brew and our predictive analytics, we’re going to be able to determine if a machine needs preventative maintenance on it before it breaks.” – CEO, Johnson

All store equipment is synced up to Azure Sphere, and over 5mb of data can be collected in a single 8-hour shift. Rather than shops being slowed down by sudden equipment issues, they’ll be prepared and aware of any potential problems.

Starbucks VP and CTO, Gerri Martin-Flickinger says their next project will be leveraging data for a better drive-thru experience. Because personalization is more difficult in a drive-thru line than in a mobile app, store sales history and other criteria will be used. Drivers will be greeted by a customized drive-thru screen that displays what is most likely to interest them.

The Future of Coffee

In an industry where convenience, speed, and customization are so pivotal to the customer experience, there’s plenty of room for optimization.

Starbucks self-serve kiosks are popping up around the world, providing more evidence that the brand intends to fully embrace automation and AI as a key customer service feature.

Aside from Starbucks, a company called Cafe X is gaining fame for its ‘robot barista,’ which can make 120 cups of coffee per hour. Such inventions have many people asking, “Can a robot make a better coffee than a human barista?” And if so, will it eventually mean a radical shift in how coffee shops operate? Café X’s machine is said to eliminate margin of error and is expected to become more popular in upscale commercial buildings in the next decade.

It’s safe to say the coffee industry is in the early stages of an innovation boom as AI becomes more accessible and affordable. We’re just beginning to see how these developments will reshape the coffee-drinking experience in years to come.

If you liked what you’ve just read, we have more…