This Technology Makes It Possible to A/B Test Brick-and-Mortar StoresBy Kieran Dahl November 14th, 2014
Imagine how many video cameras there are in retail stores around the world. Now imagine the usefulness of turning those cameras’ millions of hours of video—of shoppers’ in-store interactions and activity—into simple, optimizable, visual data.
Steve Russell has long seen the potential to turn that retailers’ dream into a reality. After studying computer science and economics at Stanford, Russell founded a company, eScene Networks, that was effectively YouTube for businesses—six years before YouTube even existed. Later, in the wake of 9/11, he founded 3VR, an intelligence company that utilizes video search technology to, in his words, “catch bad guys.” 3VR is now used by law enforcement across the country.
“I’d just become fascinated with video as a source of information,” Russell explained. “I realized that there are tens of millions of video cameras in the world, and if we could use computer vision and search technologies to suss out the interesting information held by those cameras, it would be beneficial for many different businesses.”
With that in mind, and with his background in video-based analytics, Russell founded Prism Skylabs in September 2011—an attempt, he said, “to build a cloud-based SaaS business around unique physical infrastructures already out there.” The most crucial of those unique physical infrastructures: retailers’ video cameras.
As the San Francisco-based company explains on its website:
Prism helps make sense of a visual world. Our unique cloud service transforms any video camera into a business intelligence tool that can be accessed from any device. Retailers large and small, as well as other customers, use Prism’s platform to remotely audit, manage, and optimize their real-world businesses.
I spoke to Russell about the Internet of Things, how Prism is addressing privacy concerns, and the future of real-world optimization.
Negating privacy concerns
The Internet of Things (IoT) is the practice of bringing everyday, otherwise inanimate objects to life, as it were, by connecting them to the Internet. Examples include a smart thermostat that learns your temperature preferences; a fitness tracking device with built-in sensors, paradigmatic of the “quantified self” movement; and, in Prism’s case, a technology that optimizes offline commerce by putting visual data online. While the IoT’s evolution means increased convenience—Cisco says there will be 50 billion IoT devices by 2015—it also means more pervasive privacy concerns.
Some of Prism’s competitors, whose goals are likewise to help retailers analyze customers’ in-store activity, “deliver information about shopper behavior by identifying phones’ unique MAC addresses and [using] them to track movement,” writes Gigaom. Because this tracking is “often done surreptitiously and on an opt-out basis,” it has been met with widespread criticism, including from senator Al Franken.
Apple attempted to boost customers’ privacy and negate companies’ abilities to track people by including a feature in iOS 8 that sends random, fake MAC addresses to Wi-Fi networks, though its effectiveness is questionable. Even some pockets of the government—the very entity notorious for its invasion of privacy—are attempting to get on board, Russell said, with pending privacy regulations coming from Congress and the FTC.
Russell said he anticipated these concerns. “I thought it would be important to get ahead of that and invent the technology that might become standardized in years to come,” he said. He has thus positioned Prism as an analytics company with privacy built into its technology engine.
“All our data is anonymized in the aggregate, so you end up with something in the form of ‘500 people walked into your storefront today,’” he said. Unlike Google, which only blurs people’s faces in Street View, Prism’s technology can remove people completely from the visual landscape.
“We have a more privacy-centric product that retailers and others can confidently employ without fear of a law coming down the pipeline that will make it illegal,” Russell said. “We’re an answer to privacy concerns—a good actor, a white knight with a real privacy solution and a way to get businesses to take this massive investment they’ve made in video infrastructure in their stores and put it to productive business use.”
It would be imprudent, however, to completely discount Prism’s competition. Some startups are finding that opt-in tracking, in which consumers fork over their personal data in exchange for a small sum of money, is feasible. Still others are aiming to be personal data brokers between consumers and data-hungry companies.
Prism vs. the competition
Beyond its privacy-protection software, Prism is different in other ways. For one, Prism doesn’t require retailers retailers to install expensive new devices to generate optimizable data. Prism’s data—not hard numbers but visually comprehensible heat maps—is overlaid on existing video-camera imagery of actual stores.
Prism’s heat-map technology allow retailers to see, for example, which products, and how their placement and layout, are engaging customers. If Prism’s heat map shows that a certain product display is consistently red—meaning it’s getting a lot of attention—a retailer would know to put that display in a more prominent, accessible part of the store. Conversely, a retailer might want to scale back its in-store advertising of a product that is blue or green, which signals that consumers aren’t giving it much consideration. It’s A/B split-testing real life.
What’s more, Prism’s high-quality imagery is created at a very low bandwidth, “about one percent of what a competitor with streaming video, like a Dropcam, might require,” explained Russell. That allows for an easier implementation of Prism across retail markets.
Since Prism’s inception, Russell has had no trouble finding retail clients, most of whom have issues “managing hundreds of stores around the globe. They began using us to look at their stores,” he said, “and effectively turn all the cameras into sensors that can analyze customer behavior.” Prism—whose team includes former employees of NASA, Google, Apple, and Microsoft—now has north of 300 customers, is deployed in 63 countries on five continents, and has analyzed more than 300 million customer movements and customer–product interactions.
A launch into the future
“Since the beginning of Prism,” Russell told me, “we’ve been conscious of the ton of information that’s hidden within cameras and certain sensors all around us. The big challenge is finding ways to use technology to unlock that and make it useful.”
Now, it’s Prism’s technology that’s being unlocked: On November 3, the company launched Prism Connect, opening up Prism’s software platform to device manufacturers, who will be able to embed the technology “on a host of next-generation devices, from cameras to routers to sensors,” Russell said. By opening its technology, Prism is smartly making its native integration into consumer devices even easier, with out-of-the-box cloud connectivity. “There have been a ton of entrants to the IoT video analytics space in the past year, from Dropcam to companies like Samsung, so it just seemed sensible to open up the software to these devices,” Russell said. “It’ll mean a greater payoff for our customers to have a more turnkey value-driven solution.”
Connect is launching with more than 10 brand partners, including Samsung, Sony, Intel, and Cisco, with whom Prism’s platform will have easy integration.
At the same time, Prism announced the forthcoming release of the first camera powered by Prism. The device will be manufactured by a boutique company called ISD. “It’ll be the first device of its class with Prism on board,” Russell said, “with a tiny, low-cost sensor that does everything Prism is known for, and with the kind of ease of installation you’d expect from a consumer IoT device.” Prism also announced its updated mobile app, “the fastest, sleekest, and most searchable way to interact with cameras and sensor networks,” the company said a press release.
Prism Connect is noteworthy on multiple levels, Russell boasted. “It’s news for Prism’s core retailer customer base, which is getting a better, faster, cheaper, easier device. It’s news for many other customer-facing verticals, from restaurants to hotels to college campuses, that will now have a better video-based analytics solution. And it’s just relevant to the ongoing disco of the IoT space more generally.”
Time will tell where that “ongoing disco”—a corporate dance of innovation and takeovers, such as Google-owned Nest’s $555 million acquisition of Dropcam—goes. In the next year or so, Russell predicted, Prism and its ilk will be “embedded or installed in more cameras natively, and you’ll see price points come down on those devices, and retailers will be able to deploy the service more broadly and more quickly at lower costs.”
Retailers may soon be able to optimize the real world like never before.Image by Prism Skylabs