The Technology Behind Amazon Go: How it Works and What It Means for Retailers

The Technology Behind Amazon Go: How it Works and What It Means for Retailers
Posted on | Updated on

Whether you spotted it in your social feeds, saw it in a news headline, or were one of the 7.9 million people who have viewed its debut YouTube video, it’s likely you’ve heard of  Amazon Go. On December 5th 2016, Amazon announced that it was opening an 1,800 sq ft, fully digitized grab-and-go convenience store, sending the tech world into a frenzy. With the store set to open ‘early 2017,’ the conversation continues; whether it’s to praise Amazon’s willingness to take risks and experiment with the latest technology, or to raise concerns around privacy and the supposed bleak future of the brick and mortar shopping experience, Amazon Go has grabbed the attention of techies, retailers, and consumers alike. As a tech company, we’re excited by the multiple uses of technology at play here, and curious to see how the future of retail develops with these sorts of advancements.

If you’ve not had a chance to read up on Amazon Go (or haven’t caught wind of one of the many debates it has stirred) its basic premise is that it’s “a new kind of store with no checkout required,” meaning no queues and little to no human interaction – essentially a far more autonomous shopping experience. Attractively branded as ‘Just Walk Out Shopping,’ Amazon Go is turning heads, and not just because it promises a seamless and convenient way to shop: notable failures by other brands and retail chains in the past had put a damper on the concept, but against all odds, Amazon have successfully integrated their eCommerce platform into a brick and mortar / high street setting. Chains like Tesco, who have trialled the likes of RFID readers on their trolleys, may well begin to reconsider their technical strategy in order to secure a lucrative future in retail.

So, how does Amazon Go work, and what key features place it so far ahead of the competition? Above all else, it’s their use of multifaceted and advanced technologies that can produce troves of data with unprecedented accuracy. Based on the introductory video, these smart shops will be able to detect any and everything you pick up and place in your shopping bag, and will even know if you change your mind and put something back. At face value (and given the concept’s history), a system like this seems vulnerable and prone to error. For example, will Amazon’s AI become confused when two consumers, or ‘users,’ are shopping in close proximity? What if you’re charged incorrectly for someone else’s selection? A deeper dive into the technology behind Amazon Go allows us to understand how they’ve covered these bases and managed to overcome the existing obstacles associated with integrating eCommerce into physical spaces.

The first technology Amazon Go attributes its success to is Computer Vision, which allows computers to acquire and process visual information and generate appropriate algorithms based on the data that is gathered. Unsurprisingly, Amazon has kept their patent undisclosed, however we can speculate that in the case of Amazon Go shops, multiple cameras are used to recognize customers as well as their placement or proximity to departments, shelves, items, and so on. The second is Sensor Fusion, which is essentially an amalgamation of sensor data from a number of inputs, such as weight and motion sensors; when paired with Computer Vision, Sensor Fusion helps to determine when someone has reached for an item, removed it, or placed it back on the shelf. The final and perhaps most integral factor attributing to a bright looking future for Amazon Go is the use of Deep Learning, a branch of machine learning that enables computers to learn by continuously collecting and analyzing digital data.

With this many advanced technologies working together, it’s conceivable that Amazon can learn (rather exponentially) about their customers from the moment they ‘scan’ into a store, including what they look like, their movement patterns, and their buying habits. While we can’t be totally sure of the capabilities just yet, the learning aspect also implies that with each recurring customer visit, the system becomes less prone to error, as it can begin to make educated guesses on what a customer may be doing based on past purchases and shopping behavior. This sort of consumer data is extremely valuable to retailers, particularly for marketing purposes – we’d expect that Amazon would tap into the rich pool of data collected and begin to tailor products and offers toward its users. In doing so, they can offer a more personalized, relevant, and attractive service than their competitors.

Ultimately, by employing the most cutting edge technology and heavily investing into their idea, Amazon’s ambitious vision four years ago has become a reality. They are demonstrating that the ‘Just Walk Out’ shopping concept is entirely possible, and by solving many of the technical issues retailers have previously had with this idea, Amazon have given us the first real glimpse into a future with a fully automated, self-service shopping system. No doubt there will be some hiccups, but it’s becoming more and more apparent that this (or something very similar) is what the future of retail will look like. Consequently, retailers who fail or refuse to adapt over the next decade or so will inevitably become less and less relevant to their ever-evolving customer base and risk losing out.

For retailers wondering what the next steps are, a safe bet would be to keep a close eye on what Amazon are doing, specifically around what sort of technologies they are using / investing in. Bear in mind that a reluctance to employ more advanced technologies and penny pinching have been some of the major roadblocks for retailers attempting to make their supermarkets ‘smart.’ Chains who are still skeptical or fearful of investing too much too soon may want to consider a trial store location (as Amazon Go has done in Seattle) where they can test new ideas and technologies on a smaller scale. Another viable option would be to invest in a cloud-based / location and analytics platform ike Purple’s retail WiFi solution, which allows retailers to understand exactly who their customers are, how those customers interact with shopping environments / move around their store, and where to focus their marketing efforts. Purple’s platform can also collect and report on data from the existing smart devices, sensors, and appliances in your stores, making it a powerful tool for developing / improving your IoT strategy.

Whatever retailers choose to do, the next decade undoubtedly promises some rapid modernization in the shopping industry. It will be interesting to see how it all plays out, but it’s clear that lack of a response may see them into an unstable financial future.

© 2024 Purple. All Rights Reserved.