Prism Skylabs has been selling a video-based analytics service to retailers since 2011, and the company thinks its business is about to get a big boost thanks to a new privacy feature coming out in the next version of Apple’s mobile operating system, iOS 8. The new feature — sending random, phony MAC addresses to Wi-Fi networks — could make it difficult for some of Prism’s competitors to deliver on their promises of being able to track shoppers as they move about stores.
Presently, some companies deliver information about shopper behavior by identifying phones’ unique MAC addresses and them to track movement. Such systems can detect when people enter a store (or its general vicinity) and when they leave, and with multiple routers or other hardware in place can also detect where those customers are moving throughout the store. Because MAC addresses don’t change, the companies tracking them can also detect how frequently people return to a store and, if they’re deployed in enough stores, could start to build profiles of MAC addresses based on the businesses they patronize.
This type of tracking is often done surreptitiously and on an opt-out basis, a practice that has led to a fair amount of criticism. Privacy-hawk senator Al Franken, for example, called out a startup called Euclid Analytics last spring, going so far as to send a letter directly to Euclid’s CEO inquiring about his company’s business. Responding to concern from customers, popular Bay Area coffee shop chain Philz recently stopped using Euclid’s technology.
Steve Russell at Structure Data 2014. (c) Jakub Mosur
Prism Founder and CEO Steve Russell is eager to point out all of these concerns, as well as Apple’s decision to help hamstring them, as proof that his company is on the right track. Prism, which announced a revamped version of its cloud-based service on Thursday, uses stores’ existing cameras (often already in place for surveillance purposes) in order to to do many of the same things that MAC-tracking systems can do. It can count the number of people who enter a store and track where they go, and it provides an easy way to check, from anywhere with an internet connection, whether a store is clean or its displays are properly arranged.
What Prism doesn’t do, however, is track individual people. In fact, it uses advanced computer vision techniques so that the video feeds it presents back to its customers don’t show any people at all. That’s not ideal if you’re hoping to target specific customers or types of customers in some way based on their behavior, but Russell claims it’s a very good, and privacy-respecting, way to get a sense of what’s happening in a store — or across a network of stores — without having to be there physically counting customers or examining shelves. What “used to take a whole army of people and trucks” now takes a computer monitor.
A sample of the types of data and views Prism provides.
Despite Russell’s optimism, though, it’s probably premature to count out Prism’s competition. Its technology might be great for certain things, but for companies there’s nothing quite like being able to track customers across the physical world the same way they do across the digital world. Conventional wisdom might suggest consumers won’t opt in to being tracked, but mounting evidence from startups such as Placed — and even from a recent study by a group of Italian researchers – show that some of us are perfectly willing to sell our location data if the price (which usually isn’t that high) is right.
If retailers are able to dangle tempting-enough carrots in front of consumers, companies like Euclid might find a lucrative market in opt-in tracking. Another startup, called Nomi, is experimenting with just such a model. And other, larger companies such as RetailNext already consider a litany of sources in order to track in-store activity, so it’s possible neither video- nor WiFi-tracking are make-or-break capabilities on their own.
And it’s not as if video doesn’t have the power to invade privacy, at least in theory. There are other retail-focused startups using video to provide data about store visitors, some of them attempting to identify the age and sex of visitors, or where their eyes are pointing. Given the pace of advancement in facial recognition research, it wouldn’t be inconceivable that some company will try to build a system that recognizes when the same person enters a store.
A marketing photo for Intellio’s VisiScanner product.
Russell acknowledged there might be some value in trying to discern demographic data about visitors, or even doing things like training a system to alert users when a light is on or a display is in disarray. But he also thinks a lot of those capabilities are overkill right now, if only based on how big a jump it would be for most retailers to go from simply counting customers to doing advanced analytics across dozens of variables. “It's eye opening to discover how little analysis they're doing [right now],” he said.
The great thing about video, Russell added, is that there’s so much of it and so much infrastructure in place to support it already that it’s prime for innovation when the time is right. “You could very easily see how we layer in interesting analytics to get at that data,” he said. “… There are lots of directions we can take it.”
Related research and analysis from Gigaom Research:
Subscriber content. Sign up for a free trial.
