Late last year, Amazon premiered a system that may well be the future of shopping. Nicknamed Amazon Go, it looks just like a regular brick and mortar store, except there are no lines, no self-checking machines, and no cashiers. The items you buy are checked by sensors, your account is charged through your mobile Amazon Go app, and you can just walk out of the store whenever you please.
Amazon Go is a revolutionary spin on retail, commerce, and the experience of going to a store. What's really special about Amazon Go, however, is what it represents in terms of data.
All across the retail universe, the rapidly widening Internet of Things is becoming equipped for high-frequency event analytics. Across the board, that means faster decision-making, more helpful data, and smarter, more cost-efficient businesses.
Retail and event-driven analytics
The “event-driven” company, according to VC Tom Tunguz, is one that consumes events as they occur, in real-time, from whatever data sources are available.
Rather than record data manually—making all your data liable to corruption—event-driven companies have set up the pipelines they need to always be collecting up-to-date, quality information.
The first stage in this process—“events occur”—is the most important one to consider in the retail context.
On a website, those events are fairly easy to understand. They might be clicks, button-presses, or scrolling behavior. We've been trained to think about the web in terms of events—not so with brick and mortar. And yet, the amount of events that could conceivably be collected as data from a single retail experience is tremendous.
When people enter the store, what items they pick up, which they take with them and which they put down, what order they shop in, even how they navigate the store down to the most infinitesimal of details—all of this is information that could help companies increase revenues, lower costs, and build more efficient businesses. That's also just the front-end of the retail experience.
The new retail Nervous System
The Internet of Things has spread rapidly up and down the production supply chain, laying the foundation for the future of retail.
RFID chips on products allow companies to track their inventory with an unprecedented degree of precision, even as their shipments rattle around in shipping containers, cargo ships move in and out of port, and trucks travel across the country.
Companies like Flexport make it possible to manage and visualize those complex supply chains, many of which were barely even digitized years ago. Others help optimize last-mile delivery, manage the capacities of warehouses, and plan out routes for truck drivers bringing goods to market.
In stores, the same tags that help track goods as they move around the world can be used to optimize pricing given alterations in local conditions or sudden surges of demand.
This network of physical/digital infrastructure is just the substratum, however, of the true analytics-enabled future of retail.
When data analytics meets retail
Event data is the foundation of all behavioral analytics.
When you're tracking every discrete click, scroll, or other web action, you can start to look for patterns in the data that you're collecting. You can see which pieces of content on your blog engage the most users, which version of your checkout flow is the best for conversions, and so on.
There's already technology out there to help investors like those at CircleUp analyze data around small businesses and predict those that will succeed based on a large corpus of historical data.
With the infrastructure of the Internet of Things in place, the same kind of analysis becomes possible on a physical scale. You can start to find patterns in what people buy, when people order, and how to build a more efficient goods-delivery system.
The possibilities are extensive and powerful. In Amazon's concept store, you can easily imagine sensors that take notice whenever your gaze rests on a particular item for longer than usual, or when you pick something up only to put it back down afterwards.
The decision to not purchase an item would be just as important for Amazon's recommendation engine as a confirmed sale—that data could even be fed back to the supplier for their marketing team to analyze the lost sale. Visual recognition systems could be used to show you an ad in the evening for that dress you were eyeing at the store in the afternoon.
That's just scratching the surface of an extensive universe of possibilities. Already today, IoT-enabled retail is allowing companies to:
- identify fraud before anyone from Loss Prevention even notices it's happening
- systematically reduce shrinkage by analyzing exactly where it's coming from
- give estimated delivery times in as small as 10-minute windows
A few years ago, Amazon patented the idea of “anticipatory shipping”—moving goods around based on their predictive analysis of likely consumer behavior. Because of your history, in other words, Amazon could predict that you were about to order a pack of toilet paper—and make sure it was in stock at the closest distributor well before you even clicked on the order button.
In the retail world of the future, innovations like these won't be cutting-edge. In the age of data analytics, they'll be little more than table stakes.
The data analytics long tail
The free flow of event data in retail depends on the proliferation of data sources. The more sources of data that can be cross-referenced, the more patterns that can be found and the more intelligence that can be produced.
Fortunately, the retail space is in a great position for data sources. There are not only a massive number of in-store sources of data, from sensors to registers to RFID tags, but there are complementary online sources as well.
For businesses that exist only as brick and mortar, the proliferation of IoT components and data analysis will mean a massive step forward in terms of business intelligence.
For those that are both brick and mortar stores and online, the confluence of the IoT and traditional behavioral analytics will mean an unprecedented wealth of data and an unprecedented set of options for customer engagement.
For those of us who have thrown in our lot with data, it is an exciting and fascinating time to be around.