In the second of GDR’s exclusive conversations with Intel we explore the future potential of the Internet of Things. GDR’s SVP Global Innovation Alex Sbardella gets the thoughts of Intel’s VP of Visual Retail Jose Avalos and Director of Retail Sensors Dan Gutwein.
As the Internet of Things in retail matures and develops, retailers today have an abundance of options when it comes to using technology to improve their businesses – from in-store touchpoints like mobile point-of-sale, kiosks, digital signage, digitally-enabled store associates and the customer’s device; to enabling-infrastructure like wifi, smart cameras, 4G, RFID and electronic shelf edge; and applied artificial intelligence through computer vision, natural language interfaces, machine learning and prescriptive analytics. But, with such a bewildering array of options available, and the field developing at breakneck pace, it can be difficult (if not impossible) to know which technologies to implement, and crucially, when the time is right to implement them. Do you wait and see, but risk falling behind consumer expectations? Or be an early-adopter, but risk investing in the retail equivalent of Betamax?
Luckily, it looks like the question might be getting easier. “In the last ten years, the cost of bandwidth has fallen 40 times, the cost of computing has dropped 60 times, and the cost of sensors has halved,” says Jose Avalos, VP of Visual Retail at Intel. “So right now we’re at the early stages of IoT, which is about getting devices connected and making them “smart”. But these falling costs, combined with advances in both analytics and artificial intelligence, means we’re right on the edge of the next wave of IoT, which is about making those devices autonomous – sensing, understanding, adapting and reacting to the world.” This means there’s never been a better time to start experimenting, testing, and building the components of the future store piece by piece; unlike the “big bang” approaches seen with retail technology in the past, delivering what Intel calls “Autonomous IoT” is, by its definition, a more modular and fluid process, characterised by lots of small investments and initiatives rather than large-scale digital transformation.
In this new approach, the use of artificial intelligence is critical, as the amount of data being generated by the plethora of smart devices that are being packed into stores is already too much for humans to work with. “We’ve seen a lot of devices hit the store, but not all of them create value. Having the data is not enough – you have to respond, and act immediately,” explains Dan Gutwein, Intel’s Director of Retail Sensors. And this is where Autonomous IoT starts to differentiate from its predecessor: as the “things” inside stores (and thus the stores themselves) start to get smarter, they can handle more real-time data processing in place. “The quality of your decisions improves, but crucially, the timing of those decisions improves as well – it’s win-win,” says Gutwein.
The challenge for retailers of the future, then, will not be whether or not to use analytics in their store – that will be a given – but instead which analytical decisions they need in real-time, and what additional localised contextual data can improve that decision making. Lacking this store-level context can lead to retail environments that feel disconnected from their local conditions, and missing out on data-driven insight where it’s primed to make the biggest difference, so it seems likely the long term ideal will be to move away from centralised control and towards more localised and real-time decision making – whether through empowering store managers with better actionable insights, or by building more autonomous local store systems. So whilst Target were considered revolutionary when they moved to regionalised merchandising and marketing decision making in the 1990’s, imagine what Autonomous IoT can offer towards hyper-localised decision making for every single store in the estate.
The other benefit of the near ubiquitous connectivity IoT offers is that every device in the store of the future will be able to take advantage of the data and context generated by all of its neighbours in the local network – as well as add to the network effect with the sensors of its own. In the short term, the most obvious example of this is the latest generation of digital signage and kiosks that can sense and communicate with nearby customer devices – creating a more valuable one-to-one relationship from what was previously a one-to-many device, and instantly creating the opportunity for personalised digital merchandising in-store. eBay took this approach to extremes in November 2017 by using EEG devices to offer “subconscious shopping”, but we have also seen it used in the last few months to help customers with wayfinding, in-store entertainment or payments.
Advances in RFID technology are also helping to solve one of physical retail’s largest headaches: inventory management. “A lot of governments, the Japanese in particular, are investing heavily to bring the price of RFID technology down to a penny or so per tag,” explains Gutwein. But, more importantly, only recently has the technology to extract and interpret the data from these tags in real-time been feasible to implement. Now, the store itself has visibility of what products are available, and it can adjust itself accordingly. For as much as retailers pride themselves on inventory management as a core skill, decisions often still rely on out of date data, or even worse, plain old gut feel. “We’re really trying to take the guesswork out of inventory,” explains Gutwein.
For example, if a certain product is overstocked, a smart store can adjust merchandising via digital signage, alter the cross-sell recommendations shown to store associates on their tablets, or trigger a flash sale using digital shelf edge technology in order to shift the excess stock. Or, should an item sell out, it can be removed from merchandising rotations to prevent customer disappointment, and a restock order automatically placed – all without human intervention. Scanners placed by the fitting rooms deduce which items a customer has brought in with them, and shows related content or complementary accessories on a touch screen. The store network can even deduce what items frequently enter the changing rooms but are never purchased, and feedback important data to the buying teams that might indicate a sizing problem.
Computer vision also represents a huge opportunity, especially in categories like grocery where RFID may not be price effective. Intel is investing heavily in this area. From detecting which products customers have placed in their basket, to mapping customer movements throughout the store, to recognising repeat customers and letting them pay with just their face, video solutions are attractive because they impose minimal disruption on existing store processes or designs, and require less additional hardware. Whilst Amazon Go and Tao Café are some of the most well known examples, companies such as Standard Cognition offer an off-the-shelf solution to achieve the same sort of frictionless checkout.
But in a store bristling with smart devices that are monitoring their every move, should customers be concerned about privacy? And with regulations like GDPR, should retailers be concerned about the cost of compliance as the data they are collecting increases exponentially? Although it seems counterintuitive, Gutwein says that Autonomous IoT can actually alleviate these concerns. “By making the communication bilateral between devices, it’s easier for customers to opt in or out of data collection, or gain control over exactly what data gets shared.” And advances in predictive analytics means retailers can achieve more and more effective results with anonymous data, or ringfence specific data within specific stores rather than in long-term central storage.
For retailers looking to act on these developments, Avalos has some advice: “All of this technology is really just about enabling a richer and deeper ongoing dialogue between brands and customers. Retailers need to embrace this dialogue, regardless of their model, and understand how it applies to their products and services.” In the next article in this series, we’ll explore exactly what that means. Click here to read it.