Know­ing Cus­tomers Be­fore They Know You

The tech­nol­ogy that lets you iden­tify them— lit­er­ally—is here


your fa­vorite ice cream: “Nick, Bosco Café has fresh cho­co­late ice cream wait­ing for you. Show your screen for a free sec­ond scoop.” Sound like a hor­rific in­va­sion of pri­vacy? Per­haps. But con­sider how much of our per­sonal data we’ve al­ready given out in the name of con­ve­nience: We now will­ingly share our fin­ger­prints with our de­vices.

Other emerg­ing recog­ni­tion tech­nolo­gies will help you hear what no one else can, par­tic­i­pate in a con­ver­sa­tion when you’re not in the room, and know who’s come in your front door. An early ex­am­ple was SceneTap, which in 2010 of­fered an app that helped bar hop­pers track, in real time, how many peo­ple were at a fa­vorite hang­out, their av­er­age age, and the male-to-fe­male ra­tio in that room. SceneTap mor­phed into DoorS­tat, a ser­vice for re­tail­ers, which col­lects and an­a­lyzes cus­tomer data, in­clud­ing gen­der, age, eth­nic­ity, and even mood, in real time. It lets re­tail­ers ob­serve shop­pers’ be­hav­ior in their stores and, for in­stance, move mer­chan­dise to bet­ter lo­ca­tions, or de­ploy staff who have more (or less) out­go­ing personalities.

SceneTap pre­ceded sim­i­lar star­tups, like Den­sity, whose sen­sors mea­sure cus­tomer move­ment and trig­ger pre­pro­grammed com­mands. It’s used by some restau­rants in Sacra­mento. When Den­sity senses that foot traf­fic has been slow for a spec­i­fied pe­riod, dis­counted menu prices are trig­gered and cus­tomers are no­ti­fied of the tem­po­rary changes. Another ser­vice—Placeme­ter—quan­ti­fies the vol­ume and move­ments of pedes­tri­ans, cars, and bi­cy­cles, giv­ing an in­stant snap­shot of how many peo­ple pass a store­front and how many walk into the store.

Some new soft­ware can even iden­tify peo­ple in the dark. Ger­man re­searchers have dis­cov­ered how to rec­og­nize faces by us­ing in­frared tech­nol­ogy and pat­tern recog­ni­tion—and in less than 35 mil­lisec­onds, re­gard­less of light­ing or fa­cial ex­pres­sions.

And if your cus­tomer wears a heat-sig­na­ture-block­ing hel­met? (Hey, it could hap­pen.) KnuEdge has built a plat­form that rec­og­nizes in­di­vid­ual voices, even in noisy en­vi­ron­ments. Founded by a for­mer ad­min­is­tra­tor at NASA, KnuEdge re­cently hired world-class voice im­per­son­ators to see if they could fool the sys­tem. The tech­nol­ogy pre­vailed ev­ery time.

Soon, such tech­nolo­gies will meld with ar­ti­fi­cial in­tel­li­gence and neu­ral nets—those huge com­puter net­works built and trained to “think” like a hu­man. With them, you’ll not only rec­og­nize Nick and know how much he likes cho­co­late ice cream; you’ll also know when there’s a 90 per­cent chance that he’ll walk by your store, and how likely he is to want ice cream when he does. When that day comes, it will be up to you to use your su­per­pow­ers wisely. F YOU HAD JUST ONE SU­PER­POWER, would you rather fly or be in­vis­i­ble? If you chose the lat­ter, start think­ing of a su­per­hero name.

De­vel­op­ers in Rus­sia re­cently launched an app called FindFace, which lets users scan a stranger’s face in a crowd and iden­tify her. FindFace re­lies on VKon­takte—Rus­sia’s Face­book—to re­veal, re­port­edly with 70 per­cent ac­cu­racy, who some­one is, as well as per­sonal de­tails scraped from that so­cial net­work: who her friends are, who she’s mar­ried to, what sports teams she fol­lows, what she likes to eat. FindFace’s founders plan to mar­ket their recog­ni­tion tech­nol­ogy to highly so­phis­ti­cated pro­fes­sion­als who need to iden­tify and track peo­ple: re­tail­ers.

Here’s how it could work. Imag­ine walk­ing past Bosco Café, on the edge of Red Square in Moscow. As you ap­proach, a dis­creetly mounted cam­era would rec­og­nize your face, ping a database, and learn that you just chat­ted with a friend about dessert. You’d hear a beep and look down at your mo­bile phone to find a no­ti­fi­ca­tion about

Newspapers in English

Newspapers from USA

© PressReader. All rights reserved.