Starbucks Isn’t a Coffee Business — It’s a Data Tech Company
They are a textbook example of how to strategically use data to stay competitive
Starbucks doesn’t simply sell huge numbers of hot and cold drinks around the world — it also gathers huge amounts of data from over 100 million transactions a week. How does it use this data? And what role do A.I. and the internet of things (IoT) play in this?
The way Starbucks uses data and modern technology for competitive advantage is instructive for all businesses, regardless of size. For example, it’s a pioneer in combining loyalty systems, payment cards, and mobile apps. But that just scratches the surface.
This article highlights five of the most interesting examples of how Starbucks uses data, A.I., and IoT for competitive advantage. They suggest there might be a compelling argument that Starbucks is no longer a coffee business, but is now a data technology company in the food and beverage space.
Starbucks demonstrates the relationship between data, technology, and business better than most
Starbucks is not short on data. It has over 30,000 stores worldwide and completes close to 100 million transactions per week. This gives it a comprehensive view of what its customers consume and enjoy. But perhaps surprisingly, it’s only really focused on the value of this data for little over a decade.
It’s not that it didn’t use data before then. But, as with many big shifts in a company, a crisis caused the change. In this case, it was the 2008 financial blip and associated store closures. Then-CEO Howard Schultz’s lesson from this was that Starbucks use of data needed to be more analytical, specifically in deciding store locations.
Prior to that, Starbucks’ decisions were — like many other organizations — human-driven, based on experience and judgment. Data was obviously important, but wasn’t as systematic as it could be. There’s little written about it, but it appears to have been the conventional approach of using data to validate and inform human ideas and decisions.
What it does exceptionally well is trial all sorts of new ideas using data and technology, then use more data to figure out which ones to take forward.
As well as real estate, Starbucks’ use of data today also, of course, extends to an array of marketing and product activities. This in turn leads to intelligence in how it manages its supply chain. A core piece of this is the Starbucks Rewards loyalty program, which also started in 2008.
What’s less common is the way how Starbucks’ use of data embraces the internet of things, particularly in-store operations. This started with coffee machines, and is now extending to other in-store equipment like ovens.
Five examples of how Starbucks uses data, A.I., and IoT for competitive advantage
We could fill a book with details of how Starbucks uses data and related technologies, in common with many other large, modern corporations. What it does exceptionally well is trial all sorts of new ideas using data and technology, then use more data to figure out which ones to take forward.
Of the many great examples, I’ve chosen five highlights. I’ve picked these because they demonstrate how using data well has improved Starbucks’ business, alongside technology like A.I., IoT, and the cloud:
- Targeting customers with personalized promotions and offers
- Insight-driven product development, including across channels
- Sophisticated real estate planning
- Dynamic menu creation and adjustments
- Optimized machine maintenance
Example 1: Personalized promotions
The classic use of customer data is personalizing your offer to an individual consumer’s preferences, and Starbucks is no different. With over 16 million members in the U.S. alone, its loyalty program accounts for nearly half of all U.S. store transactions.
Knowing individual customer order preferences and buying patterns allows Starbucks to send personalized offers more likely to be relevant. Using A.I. to determine such campaigns is becoming a standard application of artificial intelligence, and Starbucks has been doing this since 2017 with its “Digital Flywheel” program.
An important focus of this kind of work is suggesting new products a consumer might enjoy, based on what else they order.
But it’s not just about personalized promotions. A large part is still delivering conventional mass campaigns, but direct to each consumer in the target segment. These might include cold drinks on hot days, product launches, or seasonal menus.
Example 2: Insight-driven products
Personalized promotions are undoubtedly effective, but equally important to Starbucks is use of customer data in developing its product range.
One powerful way Starbucks uses data arises from the buying habits across large consumer numbers. Insights from this data suggest variations and developments from existing products. For example, there was a cute idea over 15 years ago to introduce pumpkin-flavored drinks at Halloween. This has become a whole range of global pumpkin-inspired products. One result is a huge spike in footfall during autumn months.
A second type is using data across channels. The most significant example of this is probably the firm’s push into the coffee at home space in 2016. This was the mainstream launch of products into supermarkets, for customers to make coffee at home. In-store data gave it a strong basis for deciding which products to target for the home drinker. It could even road-test take-home products like instant coffee in the regular stores.
It also added products like unsweetened versions of home products. Another variation that in-store consumption data suggested was versions with and without milk.
Example 3: Sophisticated real estate planning
Planning where to open a Starbucks store is now a complex piece of data analysis. The way Starbucks uses data for this covers every conceivable factor you’d expect. And it also considers a few you probably wouldn’t.
The A.I. support for store planning models economic factors about a location. These include population, income levels, traffic, competitor presence, and so on. It uses this to forecast revenues, profits, and other aspects of economic performance.
The system also considers the location of existing Starbucks outlets. It considers the impact of a proposed new store on existing revenues in nearby areas.
The A.I. technology at the heart of this application is location-based analytics. This is also known as mapping or GIS (Geospatial Information Systems).
Example 4: Dynamic menus
One implication of the examples above is that Starbucks has the ability to continually refine and adjust its offerings. The way Starbucks uses data means it can make revisions based on customer, location, and time. This affects products, promotions, and pricing.
However, if you display your in-store offerings on printed menu boards above the counter, there’s a disconnect with the ability to continually adjust things. This is one reason lo-fi solutions like blackboards remain popular with retailers. But for Starbucks, the answer is a roll-out of digital signage in stores, with menu displays set up by computer.
This completes a chain that allows changes possible elsewhere in the customer experience to be reflected in the store.
Obviously there are lots of questions this poses, and there’s plenty of scope to over-complicate things. However, as of mid-2018, Starbucks was trialing this in a handful of stores. It focused efforts on pushing selected products based on local circumstances like weather or time of day.
Example 5: Optimizing machine maintenance
Our final example is coffee machine maintenance, and in-store machinery in general.
The typical in-store Starbucks transaction is relatively low cost and short duration. High volumes of customer throughput is key to success of a store. So if a machine breaks down, it can significantly disrupt business performance.
Starbucks doesn’t keep engineers on-site for breakdowns. Instead, they send them out to deal with repairs, and, of course, perform planned maintenance. So getting engineers to broken machines quickly makes a difference.
There are conventional approaches to this problem. This typically means collecting data about failures, machine usage, repairs required, and so on. Regular data analytics is good at finding trends and patterns. A.I. can help take this up a level, forecasting breakdowns and maintenance needs.
Where Starbucks has taken things forward a step is in developing a new coffee machine, the Clover X. This is currently only used in flagship and concept stores. As well as being cutting edge in its ability to make coffee, it’s also cloud-connected. This doesn’t just allow for a more comprehensive collection of operational data. It also allows for remote diagnostics of faults, and even remote repairs.
Similar concepts will apply to other machines. For example, stores now have a standard oven, which is also computer-controlled, for consistent preparation for hot products globally. However, the current machines need to be updated by USB drive. This happens every time there’s a change in machine configuration, for example, new products. In the future, this will no doubt become a direct cloud connection, also creating more A.I. opportunities.
Starbucks is a pretty typical example of a leading modern global business. How Starbucks uses data is an exemplar of managing data and technology to great effect. There’s nothing dramatically surprising about its use of data and A.I. Nor are there any breathtaking innovations about A.I. or analytics.
But the way Starbucks uses data is a textbook example of how to start a journey to use data strategically, executing plans systematically and thoroughly. The innovation appears, but in what you do in your core business because of A.I., not necessarily in the A.I. itself. And IoT is just a natural extension of this, along with the cloud.
Another lesson is that A.I. seems to be part of Starbucks’ journey of learning to use data. It’s not something that happened because of a burning desire to use A.I. It was just the next thing to do in each area when the time was right.
The final takeaway from examining its journey is the way it scales solutions. In this case, it’s not just about things getting bigger once a concept has been proven. The global nature of the business adds regional complexities.
Most of us don’t compare our organization to Starbucks, and don’t see much in common. But that changes if we narrow our view to how Starbucks uses data. It’s also instructive to see how this has evolved into effective applications of artificial intelligence.
Like Starbucks, most of us don’t consider ourselves in the business of A.I. or data. But that doesn’t mean these aren’t becoming core to our organizations. And it does raise questions about what business you’re actually in — is it simply about what you sell most of, or what you do best?
This article is based on a piece first published on www.aiprescience.com