Investors and retailers were instantly enamored with the idea behind Shopkick, a mobile phone app that offers shoppers reward points for future purchases when they walk into a store. In 2009 the company raised $2.5 million in a round of financing led by Kleiner Perkins Caufield & Byers and recruited Best Buy (BBY) as a client. Then it became Aaron Emigh’s job to figure out how to make it actually work.
The trick was to get smartphones to figure out whether users of the app are inside an American Eagle Outfitter (AEO) or the Old Navy (GPS) down the block. Emigh, Shopkick’s chief technology officer, started by making a list of components inside smartphones: cellular radio, GPS, Wi-Fi, Bluetooth, accelerometer, magnetometer, camera, microphone. GPS was too imprecise, especially indoors at a shopping mall. Engineers conducted a series of experiments, and one by one, Emigh crossed each item off the list until only the microphones were left. “That’s where some work I had done a very long time earlier had come in handy,” he says.
Emigh, 43, began programming computers when he was 14. He later went on to work for Typepad and other tech companies, developing a word processor, networking tools, and blogging software. It was his experience as a member of an industrywide committee that established standards for modem signals that helped him solve Shopkick’s technical riddle. He discovered that the fidelity in smartphone mics is high enough to record at frequencies that can’t be heard by humans. “We could actually do an audio modem, which humans can’t hear,” he says.
The breakthrough—that inaudible messages could be pumped through speakers at a store and picked up by a phone—brought its own challenges. Modems transmitted over a phone line and never had to account for much echo. Sound waves change as they bounce off walls and displays in a store before reaching a phone. Emigh and his team eventually found ways to deal with echo and the various characteristics of each phone’s mic. But the technology worked only if a shopper stood still near the store’s entryway. “The one that we hadn’t thought about was Doppler shift,” he says, referring to how sound changes as subjects move. To overcome this hurdle, the team wrote an algorithm to compensate for movement.
In three years, Shopkick has installed its service in more than 7,000 stores in the U.S., including all American Eagle and Old Navy stores, and now has more than 3.5 million users. Recently, Emigh’s team wrote software so the audio signal could be broadcast over a store’s PA system instead of a speaker at the entrance. The new system was implemented at Macy’s (M) stores in July. “You’re constantly finding these interconnections,” Emigh says. “If you squint your eyes and turn your head sideways to this problem, some experience you thought was likely completely irrelevant turns out to be really applicable and useful.”