Shopkeepers aren’t so unlike us online writers. We do experiments: try different stories, headlines, and pictures to see what draws in readers and keeps you coming back. So do retailers: Put the blue shirt up front, now the red; see what sells.
The difference is, online experiments produce denser data. We know how many of you are reading this now, how far down the page you scroll, and how long you’re engaged. If you like what you read and “like” or tweet it, we’ll know; ditto if you “bounce” as soon as you visit the site. So we pick buzzier headlines, more engaging content, winning Fast Company “shoppers” by offering what you want. Physical stores do the same: feature the newest gadget or the sexiest bra model; book covers relevant to recent news or movies; water sounds or peaceful music at the massage parlor; incense at the New Age health store; hip-hop or indie pop at the teen clothing outlet. Whatever is being sold, the seller and staff test baits to hook shoppers, to keep them engaged with products and convert them to loyal customers. The problem is, with little data measured, it’s hard to say what works.
If you treat your retail store like a website, though, you’ll sell more stuff: that’s the bet that Tokyo-based tech startup Locarise is making. Their shopper analytics service offers to merge the worlds of on- and offline shopping. But whether or not this will creep out customers remains to be seen: When physical stores track shoppers’ behavior, as Nordstrom did lately, customers often freak out, spooked by reminders of NSA spying. That’s the challenge facing the latest developers to enter the fray.
The goal of the startup, led by three French coders based in Japan, is to give traffic measurements like those used online–“dwell time,” “engagement,” “bounce rate”–to brick-and-mortar retailers, where the bulk of money still moves from shoppers to sellers. The Top 10 sellers on Earth are still physical, after all; Amazon is still only #11. So while the bar to winning over skeptical shoppers may be high, the prize of telling retailers how to win more buyers is potentially huge.
Locarise is supported by Tokyo-based incubator Open Network Lab. Created in October 2009 by the Japanese director of MIT’s MediaLab, Joi Ito, ONLab hopes to be Japan’s version of Y Combinator: a seeding ground for globally minded startups to rival Silicon Valley giants like Google, Facebook, Twitter, and YouTube. And Locarise, among the incubator’s seventh batch, may be a promising contender for breaking out.
Locarise is designed to give retailers feedback they can act on, says CEO Sébastien Béal. Since the sensors pick up on weak signals from phones outside a store, they are sensitive to “window conversion”: what proportion of people who pass by the storefront decide to come in. The sensors can also tell how long customers spend in front of different displays; what proportion of shoppers are first-time visitors, returning customers, or “bounced” visitors, those who enter and leave without buying anything; weekly traffic patterns by day; and the effectiveness of marketing campaigns, either offline or on.
“When we discuss with retailers,” says Locarise CEO Sébastien Béal, “they often tell us they already count people, manually. The service people accumulate piles of papers telling them when the customer entered and when he left. So they already have this data and they know it could be useful, they just haven’t found the use yet. That’s what we are trying to do: Not only bring the raw analytics, but also the tools that let you visualize and manipulate this data.”
Store owners can respond to the data from the online Locarise dashboard in concrete ways: If Monday and Tuesday mornings get light traffic, hire less staff for those times; if people are “bouncing” from a particular display, do something to make it more engaging; tailor Web ads to in-store behavior.
“The online to offline topic is quite hot these days for retailers,” Béal explains. “They try to understand how [a brand’s online presence affects offline sales]. Our solution tries to bridge the gap a bit between the two.”
Robots drew the three young French guys from Paris to Tokyo back in 2009, as computer science interns from France’s most elite universities. They worked in separate Japanese cities– Béal in Tokyo, Fabian Dubois in Fukuoka, and Victor Perron in Kyoto–but they shared a common goal: building autonomous, self-aware robots to be integrated into everyday human life. Health care, retail, traffic, policing, public safety, commerce: All were problems, they thought, robots could be involved in fixing.
“What is surprising when you come to Japan,” Béal recalls in an email, “is that you expect to see robots everywhere, but you don’t see them. Except they are actually there! You are surrounded by more than 6 million vending machines, from the train ticketing to the ones distributing iPods. All these machines can do a lot more than you would expect. Did you know that in case of disaster, they unlock themselves to give access to their contents for free, providing drinking water? That’s how I realized you don’t necessarily need an interface like an all-in-one self-sufficient system to build a robot: You can have the sensors disseminated in your surrounding environment and your smartphone as the human-machine interface.”
Béal and his business partners came to this realization after a frustrating experience with traditional robotics.
They met when they worked in Tokyo for Orange Group, the global telecom corporation based in France. Their project was developing robots to care for old people–a major problem for Japan, with its aging population and dwindling birthrate, among the lowest in the world. But the “ambitious project,” Béal says, quickly shrunk in scope.
Robots are often like an animal with no body, whose eyes, ears and tongue are scattered about in space, sending signals to the cloud. More hivemind than motherboard. That’s what the French robot makers started to realize in Japan. “Robotics,” often as dependent on engineered space as on smart machines, is driven by distributed AI: less about bots than sensors.
The old people challenge also boiled down to measuring position in space. Since the retirement home’s layout was formulaic, components could be assumed: Regulation doors, kitchens, bathtubs, could be preprogrammed in the bot. But the robots were useless for navigating entirely new situations–like keeping track of a person’s unpredictable movements.
“In the end we narrowed the problem to: Identify the pose of the elderly from a robot’s point of view, to detect if he’s laying down or standing,” Béal recalls. “By trying to solve this problem, we decided we’d better put something on the elderly people, like a sensor to detect the fall, rather than trying do it with robots.”
Why have a robot at all, he reasoned, when the “intelligence” we’re designing is in the sensors? That was the lightbulb moment that led to Locarise.
Why should robots be constrained to bodies? Why not take intelligence outside of casing, remove the ghost from the shell? What if senses weren’t embodied in a centralized position at all, but distributed, and decision-making was done by a combination of robotic analytics and human interpretation?
The first problem Béal imagined that could benefit from this solution was retail analytics. Pretty soon he’d recruited his two French friends from Orange Tokyo, and they were off.
Smartphones, carried by an estimated 60% of customers in Tokyo retail stores, emit public signals, which Locarise’s sensors use to track shoppers’ traffic patterns. The Locarise system anonymizes phone data, to protect owners’ privacy, but recognizes devices later. So it can identify repeat customers, spot first-time shoppers or mega-buyers like Gucci addict Buzz Bissinger, as outliers. Key variables are “dwell time,” the length of time a customer spends looking at a display, and “bounce rate,” the percentage of people walking outside a store who don’t come in, or don’t stay. What Locarise promises to determine is: What makes your would-be shoppers leave? What entices them to buy?
Locarise analyzes and visualizes this data for the retailer in an online dashboard, somewhat like the one Chartbeat provides for online media, or Circle Media is developing to gauge audience engagement in marketing events. Retailers can use the patterns they see to design stores and organize their staff based on data, not just instinct.
Stores selling anything from groceries to electronics, books, booze, clothes, coffee, or even haircuts, would benefit from deeper data on customer traffic patterns, argues Béal. But the new metrics and tools Locarise is offering face some resistance in Japan’s conservative business climate, where data-driven approaches to decision-making run counter to established protocols of top-down business decisions based on rank and tradition. And phone tracking itself has a shady reputation these days, in the wake of Edward Snowden’s leak of the NSA’s cyber surveillance program. So retail surveillance service providers have a tough sell to make with consumers.
Locarise’s online approach to the offline world hangs on tracking smartphones. Spooky? Doesn’t need to be, Locarise’s creators say. Shopper tracking may scare off customers, the way Nordstrom did recently by following shoppers’ phones, if it reminds them of government spying. But when the data is taken anonymously and in aggregate, not for targeted marketing but feedback to the retailer, they say, it’s a win-win for customers and sellers.
Locarise’s system anonymises all phone data, and no customer data is retained beyond 90 days. In addition, the system for coding phones is separate for each client store–so no single customer could be tracked shopping, say, at both H&M and Zara. This sets Locarise apart from competitors like ByteLight and Shopkick, whose services target specific phones with special coupons and deals, linking a specific shopper’s online and offline behavior.
“We could do something like that too,” Béal says, “but we don’t want to, because it would compromise on the anonymity of our analytics. We don’t want to be in a position of saying ‘on the one hand, we are linking your phone with your ID and on the other hand we are not,’ so we prefer to focus on the analytics.”
Japan made an appealing target market for Locarise, given its ubiquity of smartphone usage–Japan has the 3rd highest smartphone penetration in the world, at 39.9%, after Korea’s 67.6% and Norway’s 55%–and its dominance as the second biggest retail market on Earth, accounting for 55% of the whole Asian market.
“If you go in the countryside in the U.S.,” Béal explained from Tokyo, “you would probably have a lot less smartphone penetration, whereas here it’s very dense, so you always have a lot of phones, a lot of signals around you, wherever you go in Japan.”
If Locarise succeeds outside of Japan, it will be breaking new ground. Remarkably, although Asia has the largest number of Internet users–44.8%, as opposed to North America’s 11.4 and Europe’s 21.5%–you’d be hard-pressed to name a web service you use regularly, assuming you’re a Westerner, that was made in a country like Korea, Singapore, or India.
Japan has innovative web services of its own–the video-sharing service NicoNicoDouga（ニコニコ動画), in which users write comments onto streaming video, and the bulletinboard system 2-channel (２ちゃねる), for example–but these are popular within Japan, not worldwide like Facebook or Twitter. The culture of startups there, as in Korea, is weaker than in the U.S. LINE, the mobile messaging service owned by Korean firm NHN, has 230 million users–18 million from Thailand, 17 million from Taiwan, 15 million from Spain, and 14 million in Indonesia–but not much range beyond Asia.
ONLab hopes to free Japanese tech from this island problem, making Tokyo-based companies global. With any luck, they’re betting, Locarise may be one seed that grows.
[Image: Flickr user Mrhayata]