If you’ve ever entered a Starbucks store, you’d agree that it’s just like a regular coffee place. Espresso shots and latte cups are being served, coffee is being ground and customers are talking to baristas customizing their coffee orders.
Serving over 100 million customer occasions across 78 markets worldwide in rapidly changing market conditions requires the coffee giant to have perfectly orchestrated processes. Additionally, it necessitates investing in technological innovations that transform Starbucks from a beverage supplier to a data-driven tech company. Deep Brew is their successful AI-driven platform. Deep Brew drives the brand’s personalization engine, optimizes store labour allocations, and drives inventory management in stores.
But Starbucks’ data transformation started way before the launch of Deep Brew in 2019. Before we zoom in on the coffee giant’s AI-driven platform that helps them serve their data-driven coffee, we’ll explore how they have utilized data in various ways to create value for their business and customers. This approach has reinforced Starbucks as the undisputed coffee shop leader.
Digital Flywheel: Mobile App for Suburb Customer Experience
Starbucks launched their mobile app in 2011, which marked its point of entry into data and analytics. It turned out to be one of the biggest underpinnings of their digital transformation.
At the beginning, it was intended to be used as a loyalty programme. This enabled customers to collect stars for every purchase and redeem them in their next drink order. The app quickly grew into a hub where customers could get information about menus, store locations, and opening hours. Customer activities in the app provided Starbucks with useful insights on the popular store locations, drinks and times of the day.
Today, a quarter of Starbucks’ 100 million weekly transactions happen through its mobile app. And the trend accelerated further with the social distancing measures. What’s more, the app’s members account for nearly 50% of Starbucks’ revenue, reports Yahoo Finance.
As part of their digital flywheel strategy, the Starbucks mobile app allowed coffee lovers to place their orders in advance. They could then collect their orders through a store window or by walking inside. By merging the power of AI and marketing, the brand has expanded its app features. Now, Starbucks’ digital flywheel consists of four digital components — reward programme, personalization, payment, and ordering.
No doubt, Starbucks’ digital innovation is credited with driving growth. They have established themselves as experts in creating loyal customers with the help of data.
How Starbucks Creates Value out of Data and AI
Early on, the coffee brand realized that using data analytics to maximize their customer lifetime value was going to be the golden ticket for gaining unbeatable competitive advantage. This includes metrics such as the average purchase price per customer per visit, the number of visits per customer per year, and the average customer lifetime.
That is how the coffee company has used data analytics to maximize customer lifetime value. And at the same time reinvent their brand offerings:
Personalized Recommendations
In the first place, collecting and analyzing a huge amount of data on customer spending and preferences helped Starbucks personalize the customer experience. This personalization was based on each customer’s preferences and spending habits.
By analyzing the history of orders and patterns, the app can suggest food and beverage choices. And also push tailor-made recommendations according to the time of day and frequency of customer visits.
By sending real-time triggers and push notifications, Starbucks creates a deeper level of connection with the customers. Buyers notice that the brand takes their preferences into account and delights them with tailoring their experience.
Innovation and New Product Offerings
Apart from personalisation, Starbucks leverages the data gathered from their digital flywheel to create new products. Their innovative products, including non-dairy and unsweetened drinks, summer specials, and new home items, resulted from analyzing user data.
For example, Starbucks discovered that about 43% of tea drinkers don’t add sugar in their tea and about 25% don’t add milk to their iced coffee when drinking at home. These insights led to creating two unsweetened ice tea K-Cups — Mango Green Iced Tea and Peachy Black Tea, states TowardsML. Some of the other fruits of their data efforts are pumpkin spice caffe latte and iced coffee without milk or added flavours.
Opening New Store Locations
You might feel like there’s a Starbucks shop popping up on every corner. In reality, they carefully use the flywheel data to determine the optimal location for every new Starbucks place. The coffee giant leverages data and AI to make revenue projections based on variables such as income levels, traffic, or competitor presence, and help determine where the next big revenue opportunity is. Thus, this enables them to minimize the risk of cannibalization and strategically position the new store to serve a specific customer base.
Deep Brew: A Platform that Elevates Humanity, Business and Customer Experience
In an interview for Yahoo Finance Live, the former COO of Starbucks, Roz Brewer, stated that the tech-enabled future of the coffee chain is focused on growing drive-thru stores. This includes customizable menu boards that leverage AI to suggest items based on factors. Factors such as the weather, time of day, store inventory, popularity, community preferences and the customer’s purchase history.
Their AI-driven platform Deep Brew allows them to innovate with AI and ML, not only to personalize drive-thru experience but also automate time-consuming tasks. Tasks such as inventory management and preventive maintenance on its internet of things (IoT) connected espresso machines.
Usually, when people hear technology and automation being integrated into the workplace, they start to worry that their jobs are being taken over by robots and machines.
However, Kevin Johnson, Starbucks CEO, Gerri Martin-Flickinger, Starbucks’ CTO and others have a vision they are evangelizing that AI is not there to take people’s jobs. Instead, they believe that AI can be used for empowering humanity at the workplace. They believe that it can help people find ways to lean into their humanity, freeing up time for staff to connect with customers and provide a personal touch.
In Starbucks’ news blog, Martin-Flickinger stated that she believed AI is applicable to virtually every aspect of the business. Technology, finance, legal, supply chain, marketing or retail stores.
With their AI technology initiative Deep Brew, they are working on technology that helps amplify the human connection, explained Martin-Flickinger. The broad suite of AI tools is set to elevate every aspect of the business and the in-store and customer experience.
But does that mean that soon enough, we’ll be seeing apron-wearing robots asking for our order at Starbucks? Not at all. Deep Brew is more like an invisible, super-smart sidekick to the human baristas helping do the heavy lifting with inventory, supply chain logistics and replenishment orders, saving partners time, predicting staffing needs and making schedules. It can also help with predictive maintenance, giving staff a heads up before a coffee machine breaks down.
The Roadmap of Deep Brew
The Deep Brew initiative was first launched in 2019 and introduced ML into the coffee company, helping them deploy models into production in their large organization, said Brian Ames, Lead Manager Data Science, Analytics Ops at Starbucks, during his Data Innovation Summit 2020 talk DeepBrew – Machine Learning at Starbucks – Roadmap.
Starbucks initially found motivation for Deep Brew in McDonald’s strategic acquisition of Dynamic Yield, which aimed to introduce reinforcement learning and machine learning to fast food. This event drove Starbucks’ C-suite to seek a response and consider integrating machine learning into their business. They saw Deep Brew as the ideal solution to address the fast-moving market changes.
Learn more about the Data Innovation Summit
Deep Brew was especially instrumental in drive-thrus during COVID world last year. With it, they could personalise the recommendations appearing on the screen at different stores with a drive-thru. Every store in every country has its distinctive personality, on top of other factors like week day, time of day, temperature, amount of traffic, etc., explains Brian. These are all points that Starbucks was able to apply in designing their recommendation system driven by Deep Brew.
“Advanced technology, the ability to deploy quickly and some advanced brains is a really nice unlock for Starbucks,“ affirmed Brian.
Under the Hood of Deep Brew
Brian explains that all of Deep Brew’s capabilities are possible because of Starbucks’ robust data foundations. Their enterprise data analytics platform (EDAP) and data lake unify data from various sources. They extract data from the lake, load it into the Deep Brew platform, and process it through a compute layer.
The output reaches and speaks to different touchpoints such as the mobile app, digital drive-thru, website, social media.
Building a machine learning system like Deep Brew presents challenges because it is a cross-functional, complex solution. Only a small fraction of the system consists of ML code. To ensure success, teams must also manage other components. They need to handle data collection, data verification, feature extraction, process management tools, and analysis tools. Unifying all teams and making them understand the importance of excellence across all dimensions can be difficult.
This is why it’s difficult to take a concept from a person’s mind or from research and deploy it at scale in a large organisation, maintains Brian.
Today, their Deep Brew platform is growing nicely and provides highly personalised recommendations to customers. But to reach this stage, Starbucks had to navigate through all the above and learn some hard lessons – or the Seven Deadly Sins to avoid, as Brian calls them.
Some pitfalls to avoid while deploying an ML system at scale in a large organization like Deep Brew Biran:
- Having no meaningful and aligned uplift to the others in the organisation
- Having no baseline to ensure that you do no harm to existing operations
- Not having clear expectations about different jobs impact: is it going to “uplift” or take over their jobs
- No easy access to security guidelines and expertise
- Not taking into account operations
- Having people that are multi-threaded or have conflicting priorities – not having dedicated teams with regular standup meetings
- Telling people that it can’t be done, and not showing for it.
Watch the whole session of Brian Ames at the Data Innovation Summit 2020
What does the Data-Driven Future of Starbucks Look Like?
As part of their digital flywheel initiative, Deep Brew has brought in a huge success for Starbucks. It has resulted in a growth in their customer base to nearly 18 million by the end of 2019. This has lead to same-store sales growth of 6% in the USA.
Apart from the numerical gains, the AI platform has helped the coffee brand achieve its goal of becoming a data-driven company. And a self-sustaining one, for that matter. This means the more data Starbucks collects, the more it can make the right decisions to grow their business.
Boosted by the AI-driven sidekick, Starbucks staff and partners can focus on the most important part of the business: coffee and customers. Providing personalized, thoughtful product choices gives customers a warm feeling. It shows that their preferences matter and impacts their sentiment towards the beloved coffee brand.
Operating world-class technology has allowed the coffee chain to attract top talent. It has drawn candidates away from tech giants.
“Over the next 10 years, we want to be as good at AI as the tech giants,” Starbucks President and CEO Kevin Johnson quoted in Marketing Dive. Starbucks’ strategy is a human-first digital strategy. Johnson adds that this vision inspires people to contribute to an environment capable of driving positive global change.
Add comment