Sometime this shopping season, you’ll find yourself stuck in a big-box store agonizing over an attractively priced television, DVD player, high-end blender, or some other amazing thing. Can I get it somewhere else cheaper, you’ll wonder—or should I snap it up now? And anyway, is this thing all it’s cracked up to be? Do other people like it? Do I even really need it?
There was a time when you couldn’t really ask such questions of products; every purchase was a gamble, a leap of blind faith that you wouldn’t get too terribly ripped off. Then came the Internet: During the past decade, many of us have adopted a mode of shopping that smacks of private detective work—before we commit to buying anything of consequence, we embark on epic research missions to seek out the best prices and the most thorough reviews. You consult the experts at CNET before choosing a TV, then you stop by your local Best Buy to see how it looks, and you order it from a discount store on the Web, tax-free. This process has liberated us from the tyranny of shifty salesmen and gouging retailers; it’s also changed the way businesses operate, prompting many to lower their prices and boost their customer service in an effort to compete with the Web.
And yet for all the ways that the Internet has transformed shopping, we still make most of our purchases in retail stores, just out of reach of the tide of consumer advice available online. Several new pieces of mobile phone software, however, are starting to transform the way we navigate through the aisles—and with a little improvement may revolutionize both how we shop and how we manage the many things we buy. If I sound a bit overheated, it’s because I’ve been blown away by the best of these new apps: Amazon Mobile, the amazing program that Amazon.com put out for the iPhone last week. Just take a photo of any item you come across in a store—a book, a CD, a cereal box, an oven thermometer, tennis shoes, anything. The app sends the picture to Mechanical Turk, Amazon’s freelance service, where anonymous hordes stand ready to identify the item. (Amazon pays people 10 cents for each ID; the service is free for you to use.) A few minutes later, your phone displays Amazon’s listing for the product, including the site’s invaluable user reviews. If you like what you see, you can buy the item straight from Amazon over your iPhone—even while you’re still browsing the aisles of Best Buy or Target.
As a piece of technology, Amazon Mobile is a marvel. The company’s Mechanical Turk service has always seemed like a lark—an interesting concept looking for a practical use. The system allows businesses to put up requests for small, repetitive tasks that only human beings can handle—writing up descriptions for photos, doing simple Web searches, etc.; anyone with an Amazon account can take a stab at doing these tasks, earning a few cents per fulfilled request. The presence of a large, anonymous workforce willing to do menial tasks has made Mechanical Turk a hideout for vaguely scammy businesses. One current request, for instance, offers you a nickel to write a five-star review of Golden Memories 1, an album of piano music; three people seem to have taken the bait, with one writing, “Note after note, key after key, the music keeps on flowing.”
This store-browsing app, though, is perfect for Mechanical Turk. Other companies have tried to build automated iPhone shopping programs, but they’ve been hindered by a key limitation of the device—unlike the T-Mobile G1, Google’s first phone, the iPhone camera’s lens can’t focus close enough to get a good read on product barcodes. (You can buy a lens attachment that improves its focus.) So how do you identify a product without its barcode? A company called SnapTell is trying to use computerized image recognition: With its iPhone app, you snap a photo of a product and, within a few seconds, the company’s servers will analyze the picture and send back a product link. SnapTell is more limited than Amazon’s app—it is only meant to identify books, CDs, DVDs, and video games—but it did identify most of the books and some of the CDs I threw at it nearly instantly. *
Amazon’s service was slower—I had to wait between two and 10 minutes before I got a listing for my product—but because there were humans on the other end, I got much more accurate results. Amazon identified the Cheez-Its, the Arm & Hammer detergent, a copy of the Collins Gem English Dictionary, a Sylvania portable LED light, a Waterpik, a package of Oreos, a copy of my book, and a copy of A Briefer History of Time. It failed only when Amazon didn’t carry the product I’d photographed—when I snapped a shot of Fre nonalcoholic red wine at the supermarket, a helpful Mechanical Turker directed me to an Amazon listing for sparkling grape juice.
Intrigued by the photo-identifying elves on the other end, I logged on to Mechanical Turk to offer my help in finding listings for people. My first test: a slightly blurry, dim photo of a black Bluetooth headset. It’s a good thing I’ve got a job that requires me to try out a lot of Bluetooth headsets—I knew immediately that this was the Aliph Jawbone. Within 30 seconds, I found the Jawbone’s Amazon listing and sent it in, hopefully surprising the fellow on the other end with my accuracy. (Alas, Mechanical Turk offers no way for searchers and searchees to talk.) The next photo would also have stumped a computer. It was an upside-down shot of one-thong sandal, slightly blurry and pretty small. You could just barely make out a logo on the insole—a red drawing of what looked to me like an ocean wave. For about 20 seconds I wracked my brain trying to identify it. I Googled “Billabong logo“—wrong. “Hang Ten logo“—nope. “Quiksilver logo“—yup, that was it! Next I searched Amazon for men’s Quiksilver sandals and found a pair that seemed to be a pretty close match to the ones pictured. I’d done about a minute’s work, which seemed a lot more than was merited for the dime Amazon was paying me. It was quitting time.
Because phone shopping apps give shoppers power in a place where they’ve long had none, they’re bound to create some tensions with local retailers. One shopper in Michigan was recently admonished for using ShopSavvy, a barcode-scanning app for Google’s Android mobile phones, at his local Target store. (Target later said that the employees acted in error and that the company has no policies on barcode scanning.) Last week, Nate Anderson of Ars Technica wrote a thoughtful piece worrying about the ethics of Amazon’s app: “For Amazon to explicitly suggest that shoppers take advantage of bricks-and-mortar stores—an expensive investment that Amazon has purposefully not made—and then use the benefit derived from those stores to order the product cheaply online, well, that’s a pretty straightforward declaration of war.” Some consultants have even suggested that retailers fight back by installing cell phone jammers or banning iPhones from their stores.
But I’d caution the bricks-and-mortar crowd to hold their fire—it’s possible that Amazon’s app might even help them. I was at Wal-Mart this weekend when I saw that Sylvania portable LED light. I wondered if it would be bright enough to light up some dark areas in my closets. The packaging offered no help—so I took a photo and, after getting back a positive ID, consulted Amazon’s listing. For one thing, I found that Wal-Mart was much cheaper than Amazon (no surprise). But Amazon’s reviews also convinced me that the light worked pretty well—in other words, Amazon pushed me to buy the item from Wal-Mart.
The same thing could work elsewhere, of course. Next time you want a good book, try this: Go to Barnes & Noble, and find something that looks interesting. Not sure it’s as good as the blurb says? Check Amazon for reviews. If you like what you see and the book’s cheaper online, go to the store manager, show her your iPhone, and ask her to match Amazon’s price. If she says yes, buy the book there—otherwise, click Buy right in front of her. Remember: He who has the phone makes the rules.
Correction: Dec. 10, 2008: This piece originally criticized SnapTell’s iPhone application for failing to recognize products like a bag of Cheez-Its and a bottle of Arm & Hammer laundry detergent. The SnapTell application is only designed to identify books, CDs, DVDs, and video games. (Return to the corrected sentence.)
