Mark Elfenbein’s pitch is simple enough: Shazam, but for stuff.
The venerable mobile app, now used by more than 400 million people, recognizes any song that’s playing, so you can buy it on the spot. Elfenbein, president of the Toronto-based start-up Slyce, hopes his technology will become just as ubiquitous, or even more so, since it works with just about anything. Take a picture of an object, and Slyce’s image-recognition software will figure out what it’s looking at and take you to a website where you can buy something just like it. “Slyce allows you to capture an image at the moment of inspiration, and purchase that item upon inspiration,” says Elfenbein.
It certainly has sci-fi appeal. Technology that links real objects to online purchases has retailers excited–especially those that are competing against online stores and watching helplessly as their bricks-and-mortar shops turn into little more than showrooms for phone-wielding comparison-shoppers. Amazon, to which many of those comparison shoppers turn online, has already embedded image-recognition technology in its mobile apps, and snap-and-buy functionality is expected to come to smartphones everywhere in the next year. Investors are certainly excited about the possibilities: They’re pouring tens of millions of dollars into image-recognition technology, and Slyce is in the process of going public as part of a $60-million deal.
Finding a way to link real-world objects with online shopping has been an elusive marketing dream over the decades. The ridiculous glory days of the first dot-com boom produced something called a “CueCat,” a device that would scan special codes embedded in magazines ads and link them to web pages. It was actually shaped like a cat. In the end, customers proved inexplicably reluctant to spend their time waving a cat-shaped wand around in the hopes of being fed more advertising. So, despite being funded to the tune of $185-million, the enterprise went bust.
Its successors didn’t fare much better. Barcode-scanning apps came and went, with little effect. A future awash in RFID tags was presaged, but if it ever arrived, consumers didn’t notice. Marketers spent years flailing around with QR codes, which required users to scan a little square full of squares in order to be directed to a website corresponding to the physical object they were looking at. But the process was fiddly and the rewards often underwhelming.
Now, technologists are back for another crack at the idea, albeit in a more streamlined form. Smartphones are powerful enough, and image-recognition software is smart enough, that we can forget the QR codes and just take pictures of the items themselves. Say you see a shirt you like, but you don’t know what make it is or where to get it. If you take a picture of that shirt, Slyce’s technology will examine its features, picking out, for instance, how many buttons it has, what style the collar is, whether it’s a striped or plaid pattern.
If a given merchant has this exact item in stock, then whoosh–off you go to the online store to buy it (or you’re directed to a nearby bricks-and-mortar shop that has it in stock). If not, the image-recognition app can recommend similar items, acting as a recommendation engine that might lead to different purchases altogether. And, throughout all this, the software is taking notes on what you’re interested in, and whether or not you buy. That data can be worth more to the retailer than the purchase itself.
There are two ways these technologies are finding their way onto the market, aside from Amazon. Apps like the fashion-focused Asap54 can be downloaded for free and will direct curious viewers to offerings from various fashion houses. Slyce, on the other hand, is working with retailers to build its technology into their own proprietary apps, hoping to draw customers deeper into the retailers’ catalogues. (Elfenbein isn’t naming names yet, but says he’s working with six of the top 20 retailers in America and is aiming for a late-summer launch of the first Slyce-powered apps.)
The concept could extend beyond individual products. Elfenbein says there’s interest in the idea of using image recognition to, for instance, take note of a hairdo and suggest hair-care products a customer could use to achieve said coif. The same trick could work for home-renovation firms. “If I took a picture of a hole in the wall,” says Elfenbein, “it could say, ‘Here’s the paint, and here’s the three items you need to fix this.’”
This might be optimistic. At times, image recognition can sound like a technology in search of a use. There are a couple of big hurdles to be cleared before any of this will catch fire with consumers. For one thing, we have a low tolerance for image recognition that’s overly fiddly, needs just-so lighting–or returns lousy results. “People will walk away from the app if they’re not finding 80 per cent recognition,” says Nigel Wallis, a research director at IDC Canada.
And the technology poses as much of a threat to retailers as it does an opportunity. “Showrooming”–the act of wandering into a store, doing a price comparison, then buying the actual item elsewhere–is a consumer behaviour that has caught on. In a new IDC study of 600 Canadian smartphone users, 46 per cent of them had compared prices with other retailers while physically in someone else’s store within the last month. “That’s a killer for any kind of commodity product,” says Wallis. “That ranges beyond consumer electronics, and into generic clothing types, like when you’re looking for a pair of jeans.”
So, on one hand, retailers want image-recognition apps that will draw customers into their own catalogues. On the other, consumers are looking for the best price across stores. Image recognition might be the stuff of the future, but retailers won’t make it work for them until they give shoppers a reason to get the picture.