Starbucks — Grinding Data

Coffee giant Starbucks has mounds of data, and they're putting it to good use marketing to customers and locating new stores.

An Opportunity Brewing:

What happens when you conduct 90 million transactions a week (1), spread around almost 25,000 stores worldwide (2)? You accumulate a lot of data. Over the last few years, coffee giant Starbucks has been putting that data to good use, incorporating data analytics into its marketing and sales efforts.

Not surprisingly, mobile and Starbucks Rewards have become critical factors in those efforts. Starbucks mobile app has over 17 million active users (1), and mobile order and pay records over 7 million transactions a month (1). In the US, mobile order and pay alone makes up an astounding 27% of all transactions (3). Starbucks Rewards has 13 million active members (3). This extraordinary level of engagement on the mobile app and rewards program greatly increases Starbucks’s ability to collect data, and test and roll out targeted data-driven initiatives.

 

Pressing on with data analytics:

Personalizing orders (and experiences):

During last year’s annual shareholder meeting, CTO Gerri Martin-Flickinger emphasized the goal of personalizing the customer experience: “with our 90 million transactions a week we know a lot about what people are buying, where they’re buying, how they’re buying, and if we combine this information with other data, like weather, promotions, inventory, insights into local events, we can actually deliver better personalized service to other customers.” (1)

That can happen even when shopping at a Starbucks you’ve never been to before. Say you usually make the same Starbucks order most days, at around the same time. A store’s point-of-sales system can detect the proximity of your smartphone, and provide the barista with that information (1).

Martin-Flickinger went on to explain how data could be further used to up-sell customers in a more targeted and efficient manner: “We also show you your favorite treat in a picture at same time. Does that sound crazy? No actually, not really. In the coming months and years you will see us continue to deliver on a basic aspiration to deliver technology that enhances human connection.” (1)

 

New store locations:

Patrick O’Hagan, director of market planning, talked about how data drives decisions on where to open new stores: “Through a system called Atlas, Starbucks links to as many external and internal APIs as possible, connecting the data with R to build cannibalization models that can determine impact to existing stores if a new store enters the area.” (4)

The platform O’Hagan references, Atlas, is a mapping and business intelligence tool developed by GIS company, Esri (5). The system takes into account a multitude of factors, including traffic patterns, population density, demographics, and proximity to other Starbucks stores to evaluate candidate locations for new stores (5).

 

Targeted marketing:

Another application for Starbucks’s abundant data is targeted marketing. A recent mobile app update started targeting customers with discounts and rewards on certain items based on their purchase history. Additionally, Starbucks sends out emails to re-engage dormant customers. The content of those emails is targeted towards each customer, based on their purchase history as well. That could be particularly powerful, considering that Starbucks offers 87,000 unique drink combinations [5].

One interesting component in Starbucks’s data analytics efforts is weather. The company has been trying to understand the effects of not only seasonality, but also day-to-day weather on customer order patterns. That allows a higher level of customization and more effective targeted marketing to drive up revenue.

 

Sources:

 

Previous:

Constraints on Data-driven Systems in Public Education

Leave a comment