【USA】

I wanted to dismiss Amazon Go as another Jeff Bezos attention play. But I was wrong. This is USAthe shopping experience we’ve all been waiting for. Amazon is the most important retailer in the country, and when it tries to transform how we shop and receive products, people listen. Ilisten.

When Amazon CEO Jeff Bezos went on 60 Minutesa few years ago to unveil his plans for drone delivery, I had a million questions and almost zero faith that they would ever make it happen. Delivery drones like Amazon's proposed Prime Air must be fairly large, they’re loud and, in most neighborhoods, will always feel invasive. And the FAA is never going to cooperate. Delivery of one product by drone isn’t even particularly efficient. If nothing else, Amazon is an incredibly efficient product delivery service. I can order five products this morning and receive some or all of them within 48 hours (and sometimes sooner).

Amazon Go, though, is a retail experience of a different color. It relies on several pretty well-understood technologies that have already migrated into public consciousness and use: computer vision, deep learning algorithms, and sensor fusion. Okay, that last one may be a bit less familiar -- Amazon describes it as similar to what you’d find in self-driving cars, where data from several different sensors is considered before the system draws a conclusion about what's actually happening.

The whole thing reminded me of a shopping technology I saw at CES 2013. Like Amazon Go, Shopperperception (yes, the name was awful), used 3D imaging (this time from PrimeSense, now owned by Apple) to observe consumers as they looked at store shelves, perused products and even picked some up. It was intended as a tool for retailers, to help them better understand and programmatically measure consumer interest in specific products. It knew when you plucked a box of Kellogg’s Corn Flakes off the shelf and then put them back.

Amazon’s solution takes this core idea much further. Amazon Go marries knowledge of store shelf activity with live information of every shopper in the store. It starts with the smartphone app, which customers use to check in as they walk in and, very likely, beacon technology to know exactly who is standing in front on which store shelves. So a sensor watches exactly which products are taken while the beacon notes who grabbed the product.

I know that some are worried about the loss of human interaction both at store counters and at checkout, but seriously -- how many of us still shop in local food stores? Our neighborhoods are filled with large supermarkets staffed with mostly pimpled teenagers. They barely glance up as they wordlessly scan item after item. If something doesn’t scan on the first pass, they keep doing it until it reads, while you look on in frustration. The last thing the cashier wants to do is call for a price.

What, exactly, is so wonderful about these human interactions?

The goal of going to the supermarket is to quickly get all the groceries you need. If it’s a cartful, you know you still have to unpack the whole cart to check out. Even if it’s self-checkout, there’s that two-step process. Amazon Go is promising just one: grab-and-go.

Someone mentioned that Amazon Go is a brick-and-mortar play. Even if the company does build 2,000 stores, that’s not the play (I'm betting we get more like a 100 in 2017). Amazon’s interest has been and always will be customer data. The more it can collect, the better it can know you and anticipate your needs. The more you shop at Amazon Go, the more Amazon will understand your weekly grocery cadence. Eventually, Amazon’s Alexa digital assistant will be telling you when you need milk, butter, sugar and even suggesting full weekly meal plans based on past purchases.

The other reason Amazon might be interested in building all those stores is that stocking and delivering food is harder than the books, back scratchers and Blu-rays Amazon typically sells. With groceries, inventory is perishable, and overstocking can lead to huge loss. Storing groceries is also more expensive (all that refrigeration). Distributing it to stores where you pick up the product and help them manage inventories in real-time makes more sense for a food-service.

For similar reasons, grocery shopping is one area where many customers prefer to go to a store instead of ordering online. As good as services like FreshDirect have gotten, you still can't view a picture of the actualpork chops (or cheese, or asparagus, etc.) you're buying, and even if you could, you can't smell it or feel the texture. These things matter to many food shoppers, which is why there will always be a significant number who will never order groceries online.

In the space of 24 hours, I’ve gone from deriding Amazon’s big grocery play to realizing that this may be Amazon’s best consumer idea yet. It’ll be a game-changer for them and a life-saver for us. Bring it on and, please, leave drone delivery behind.