Every so often Amazon makes announcement that looks like a Ryanair publicity stunt; something that sounds remotely plausible but in reality is a non-starter. The difference is that these are very much starters. We have seen the drone, the bookstore and even Amazon Dash and now we have the payless store. No more queues, no more coupons, no more loyalty cards, no more neural network-like calculations on the shortest line to join; you walk in, pick up your stuff and walk out again, simple as that. The problem is that to enable this simplicity requires a technology that doesn’t seem to exist right now.
Tracking customers throughout a store is not a problem today. There are a variety of technologies that can solve this and it looks as though Amazon is using a combination of real-time camera analytics and smartphone sensor fusion (possibly combined with Wi-Fi and/or BLE beacons). This is not a difficult thing, and the smartphone scan on entry/exit links the person to an Amazon on-line account.
The problem is recognising/tracking the items that a person selects. There are multiple item-level in-store inventory management technologies emerging. Passive RFiD has been around for some time, with Impinj and others working on ways to support real-time item level location. The problem here is the need to tag every single item and the inevitable difficulties and inaccuracies of RFiD- it is great, but it is not a 100% technology, which is kind of a problem in this scenario. Also, even as cheap as passive RfiD tags are, I am unconvinced they are the answer in a high-density, margin sensitive vertical like grocery.
In 2016 Tyco also launched it TruVUE solution that combines camera analytics and item-level RFiD data to provide new levels of analytics, loss prevention and inventory management. While the Amazon approach is a clear long term aim for this platform in the future, it is hard to know if it is quite there yet. But what is clear is the Tyco has the same joined up thinking as Amazon which is one of the reasons why it topped ABI’s in-store analytics competitive assessment earlier this year.
Another option is watermarking, such as Digimarc’s technology. By printing their technology on every item a decent digital camera should be able to recognise each item as it is selected. This is a very elegant, low cost technology, which works well for own-brand items. The problem is getting all other FMCG groups on board. T-ink has a great printed sensor mat solution that can measure when an item has been removed from a shelf. Linking this technology with accurate inventory location data and real-time high accuracy customer location information is actually a realistic way of doing this, but we still need to see what errors this technology may be prone to in the real world e.g. people putting the wrong thing back on the wrong shelf.
Its early days but I believe that 3D scanning is playing a big part here. This is something ABI Research has been considering for some time in the retail space and I believe it is going to be a hugely disruptive technology. You might not know it, but you have probably already used this technology in the form of Microsoft Kinect or if you are very lucky Google Project Tango (I nearly got to touch a Tango tablet at a conference recently!). The scenario I envision is placing 3D sensor cameras along every aisle in a store. Not only will this gather highly detailed information on customers movements through the store, but for the first time it will also provide item level information, such as what products were touched, what items did customers look at and what are the current real-time inventory levels. This solves immediate problems around stock-outs, inventory management and planogram compliance but more importantly, retailers have information on product interactions for the first time. Now, stop and think for a second just how much money a major CPG brand like Mondelez or Coca-Cola would pay a retailer get access to this type of information?
Today, Shopperception and AVA Retail are two start-ups blazing a trail in this area. With machine vision innovation happening across a number of verticals today, ABI Research expects to see the cost of this technology falling rapidly, to the point that 3D becomes the de-facto retail camera analytics technology in the near future. We are already aware of some leading vendors in the camera analytics space expecting to make announcements in 2017.
So, to circle back, there doesn’t appear to be one single technology that I am aware of that can solve this problem (with potentially the exception of Tyco). However, there are a few viable solutions out there today that in combination could be the answer. Either way, this is now becoming a realistic opportunity for retailers, while ABI’s prediction that future retail stores will be free of queues is already starting to come through.
Note 1: one thing I really love of Amazon’s implementation is that it has turned a potential privacy crisis into a key differentiator that customers will desire. Instead of headlines about tracking customers throughout the stores, Amazon has turned it into a utility. This is exactly the model the retailers should look for when implementing in-store analytics technology. Not only should it gather necessary data, but it should use that data to create utility and better service to the customer.
Note 2: I am already hearing the “it’s not for everyone” argument. Well guess what, nothing is for everyone, especially your long queues and ineffective marketing strategy. I am pretty sure that if Amazon does indeed get serious on bricks and mortar grocery it will also offer checkouts, delivery and click & collect. Look at something like Starbuck’s pre-ordering service. A personal straw poll last week puts it at about 20% of total traffic. That 20% of people that will nearly always use Starbucks, have a great customer experience and will become brand advocates, but you know, “it’s not for everyone”. Furthermore, I spoke to the company that helped Starbucks decide on launching this initiative, and want to know how they got there? They gathered data on their customers, much like the data in-store analytics technology enables.