With sensors, Big Data and machine learning all about to hit our shops, are we taking the next step towards a scary future of retail?
There’s a scene in the 2002 film Minority Report when Tom Cruise’s character walks through a shopping centre and a wall of holographic advertisements address him by name.
In that film, optical recognition technology allows a series of sensors dotted throughout the centre to pin down the specific identity of customers, call on a stored database of information and tailor adverts in real time to appeal to an individual person’s tastes. This is framed in Minority Report as a terrifying intrusion of consumerism into our private lives. It is also surprisingly close to what will happen across many shops over the next few years.
The reality of this Minority Report-style consumerist future was toted at a recent Intel and WPP event I attended. Entitled “Retailing to the Customer of One in the Internet of Things Age”, the Customer of One is advertising material tailored to the individual. If advertisements are traditionally aimed at segments, the Internet of Things promises to narrow this down considerably. Instead of marketing to a vague demographic, adverts will be selected based on the individual characteristics of customers.
Internet of Things and the Customer of OneCreepy associations are something that both advertisers such as WPP and technology companies such as Intel are keenly aware of. In his opening keynote, the CEO of The Store WPP, David Roth, spoke about the way facial recognition, data analytics and machine learning will revolutionise the retail industry over the next decade, but that using these tools without scaring customers is a tricky task.
During the presentation Roth demonstrated a small camera attached to a cute-looking robot that – like a Debenhams Terminator – was able to identify both his mood and the brands of clothing he was wearing. This information could then be used to display appropriate advertisements on a nearby monitor.
There is undoubtedly a degree of creepiness that goes along with being analysed in this way by a machine but, crucially, the robot eye was not able to identify Roth’s name or personal details. I say “not able” – it most likely could if the programmers linked it to a suitable database. Nevertheless, it was prevented from being able to attach a specific identity to the person being sized up by the software and this, I learned, is key to how WPP and Intel draw the line between what is and isn’t an acceptable use of the technology.
The line between intrusion and assistance
What if the reason Minority Report is unsettling isn’t because of the targeting ads, but because those adverts refer to the main character by name? If a sensor were able to judge certain characteristics of a person, and recommend items accordingly while maintaining a sense of anonymity, would that be an invasion of privacy? Or would that be a digital extension of what shop assistants have been doing for decades?
To get a better idea of how techniques like these would work, I spoke to Joe Jensen, general manager for Intel’s Retail Solutions Division. Jensen told me in more detail about Intel’s vision for how the IoT would function in a retail environment.
“Say I have a store in a neighbourhood and with sensors I’ve observed that when it’s early afternoon, and there’s a female shopper, and she’s moving quickly, she tends to do this,” he explained. “You don’t have to know anything about that shopper, except that she’s female and moving quickly and, historically, in this store, in this time window, someone in that situation tends to, say, pick up milk.”
(Above: Samsung Display digital viewing platform combines Intel Real Sense technology with OLED technology)
As Jensen described it, an array of interconnected devices in any given shop, using everything from facial recognition to motion sensors, could form a profile of a consumer – not based on private information, but on details picked up about that customer during their visit to the shop: their sex, size, speed of movement, and so on. “It’s not so much that you need to know ‘Elaine’, it’s that you need to know that this shopper has these characteristics and, in the past, that when those characteristics happen, this is what a person tends to do,” said Jenson. “It might not be right – maybe you guess by accident that I’m a girl and I want to see a prom dress – but I’m not going to be insulted. It’s about making the hit rate of content better, not necessarily perfect.”
Monitoring customers and building predictive models is a different kettle of fish to delving into specific customer accounts, but it is still surveillance. The big question is whether or not this is an intrusion. Is monitoring the physical characteristics of shoppers and using this as a basis to select advertisements an unwarranted invasion of personal space, or is it a useful way of streamlining the shopping experience? Jensen argues that this method retains anonymity but whichever way you look at it, convincing customers not to smash profiling sensors with hammers hinges on an enormous amount of trust.
MAC address tracking
Jensen told me that trust is indeed a very important part of the relationship between customer, consumer, and technology. “Whenever we work with brands and retailers we always emphasise that, at some point, trust will be broken,” he said. “And with social media, a company will die. Let me give you an example: Nordstrom [a US retailer] was doing an experiment in one store, where they were looking at the ISN number – MAC address – on your phone. Some blogger said Nordstrom implemented tracking and there was a huge controversy.”
Nordstrom did indeed use the Wi-Fi signals on its customer’s smartphones to track movement in the shop. Nordstrom’s defence was that they were only doing what online outlets like Amazon already do, but people reacted negatively all the same. It’s easy enough to see why. There’s a vast difference between a customer’s relation to an online shop and a physical shop, not to mention having your physical presence invisibly tracked with a device many associate with private communication.
(Above: Smart bins installed in the City of London were removed in 2013 after it was revealed that they had been collecting MAC addresses)
Nordstrom may have bore the brunt, but it isn’t the only retailer delving into these tactics. Jensen told me that any time you log onto free Wi-Fi you are likely to have your MAC address tracked. “A huge number of retailers are using MAC address tracking. In Asia, we had a customer tell us that in their digital science network in Beijing, they can track every single cell phone in the whole of Beijing. It’s a hugely prevalent thing. I’m personally not a fan of it. I think it crosses a privacy line. […] We don’t want to use the word tracking, because we’re not actually trying to track people. We’re trying to observe behaviours.”
The Internet of Things and the shop of the future
WPP and Intel talk about bringing smart devices into the retail environment as less of a far-off sounding possibility and more of an impending inevitability. The argument isn’t whether or not customer monitoring will come to shops, but rather where you draw the line of conduct when it gets here.
We adapt to things very quickly, and privacy is by no means a solid concept. How long will it take for us to get used to being watched by sensors as we shop? From the outset, monitoring and targeted advertisements sound like a creepy infringement of our personal space, but with the promise of a tailored shopping environment, will consumers fight against this new technology or embrace it?
For all the rhetoric about introducing sensors as a means to improve our shopping experience, there is ultimately one reason for these technologies to be introduced: to make advertisers and shops money. That, in itself, is nothing new. They are businesses after all. But when the line between a clever use of data and intrusive tracking is less to do with the capabilities of the technology, and more to do with the way it’s used, that future is something we as citizens need to pay attention to.
No comments:
Post a Comment