Biometric retail technology is ready for its close-up; but are shoppers ready for it?
The global market for retail biometrics is by all accounts in its early stages, and is significantly fragmented between regions. Retailers in Asia are testing biometric systems for retail payments and customer interaction, while most trials and deployments in North America so far seem to be focused on security. The range of applications is broad enough for biometrics to be a potentially disruptive force for the retail industry worldwide in the foreseeable future.
Amazon opened its first Go stores, which are partially automated to provide payments without a cashier or self-checkout, to the public at the beginning of 2018. The company plans to roll out up to 3,000 Go stores by 2021, in a unique divergence from the regional split. Pay with a smile was launched with Alipay technology at KPRO by KFC stores last year in China, and facial recognition has more recently been applied to controlled retail settings for trials in Finland and Korea.
The majority of retail biometrics projects in the U.S. so far, however, are focused on loss prevention and violence prevention, FaceFirst CEO Peter Trepp tells Biometric Update. After more production roll-outs of that application, different ones will eventually follow, he expects, but not right away.
“There is another step though that exists which has more to do with consumer loyalty, and consumer experience, that is not quite as expensive an endeavor, and I think there are lots of folks looking at ways of doing that in a friendly opt-in environment, where privacy is not the cornerstone issue, and consumers are opting into systems that allow them to engage with them,” Trepp says.
SensibleVision CEO George Brostoff also sees a range of possible uses for facial recognition in stores. “When it comes to retail the applications kind of expand out from traditional authentication,” he said in an interview. SensibleVision is in discussions with businesses in Asia about several different retail applications of its facial recognition technology. Like Trepp, Brostoff sees customer loyalty programs, which inherently involve a customer opt-in, as a logical next step in Western markets like the U.S. and Europe.
Goode Intelligence Founder and Managing Director Alan Goode also sees potential for customer loyalty programs in those markets to leverage facial recognition in the near future, but suggests a couple of other applications are also likely to emerge in the near future. Age verification, which in some cases is handled on-device with apps produced by providers like Yoti, is a logical next-step for automated self-check out systems selling age-restricted items like tobacco or alcohol. Likewise, Clear is providing biometric age verification and payments for concessions in Seattle – where the original Amazon Go stores are located – in what Goode identifies as a trend towards hybridization.
“What we’re seeing is immense hybridization going on in terms of payment, ticketing, identity all getting molded into one,” Goode explains to Biometric Update.
That hybridization will be one of the drivers of a rapid increase in the adoption of biometric payments, according to a recent Goode Intelligence report, which forecasts 2.6 billion people globally will use biometrics for payments by 2023, when 579 million biometric payment cards will be in use.
Arturo Falck, CEO of startup Whoo.ai, also sees customer loyalty as the next logical step, but he sees privacy concerns coming up even then.
“Once companies are using this type of technology for crime prevention purposes, there’s no reason why they should not be using it for upselling their customers,” Falck told Biometric Update. “In a way you can see the natural progression. If you remember way back when Gmail first started, we didn’t really think through the fact that pretty soon the advertisements that we were seeing when we were browsing the internet were targeted to us based on the emails that we were sending to each other. And you can imagine how that extends to the real world and how people are nervous about it.”
A recent survey from the Brookings Institution indicates that half of American consumers have an unfavorable opinion of the use of facial recognition for retail stores to prevent theft, compared to only 27 percent who are favorable of the application. People are more comfortable with the use of the technology in airports, stadiums, and even schools, according to the report.
There are many different ways that lack of comfort could play out, from legal or regulatory barriers all the way to a reversal of attitude among those who currently have negative views of the technology’s use in retail.
“Retailers need to be very careful to have things be both opt-in and not creepy,” Brostoff cautions.
Goode, Trepp, Brostoff, and Falck all believe that as the technology proves its value, and consumers become comfortable with it, different applications will be more easily accepted. Each of the technology company executives finds a comparison with earlier stages of the internet, when certain behaviors that are normal today were considered risky. Trepp recalls when most consumers were uncomfortable entering their credit card information online.
“Those days are gone,” he says. “We trust these companies to do the right thing with this information, and that’s how these companies, like Amazon and others, anybody you do business with online, become big growing companies. They treat their customers’ information correctly, and respectfully, and they respect their privacy.”
The same process of growing trust must be navigated by retailers if they are going to successfully leverage the power of facial biometrics, but Trepp is confidant they will do so. “Their incentives are very well aligned with their customers,” he notes.
Falck estimates, based on Whoo.ai’s early projects, that roughly 3 percent of the population is vocally opposed to participating in facial recognition systems, but even more are concerned about the potential loss of privacy from its increasing use. In response, Whoo.ai has developed a consumer application to provide people with information about retail facial recognition use, but control over it. The company plans to attract a network of retailers with not just positive public relations, but another way to connect with customers.
“It is a way for the businesses to turn what could potentially be bad public relations into engagement opportunities,” he explains.
Companies participating in the scheme can offer app users incentives to opt-in, as well as loyalty rewards or discounts on particular items they may be interested in. The back-end uses an open REST API for companies to integrate it with their facial recognition systems. When the system identifies a customer, it makes a call to submit the image, and WhoKnowsMe returns the user’s permission or denial. Falck believes it may help retailers demonstrate compliance with regulations in the future.
There is interest from a number of retailers, he says, but none want to be first, so he expects a coordinated announcement of participating companies sometime after the app launch.
Brostoff is unsure if a customer engagement app is necessary to ease public concern, though he agrees broadly with Falck’s assessment public comfort levels.
“There is a small group of people who are highly sceptical and concerned, and those are the ones we hear about the most,” Brostoff concurs. “There’s a wide swath of people who don’t have strong opinions, and have to see it, and you have a small group of early adopters.”
The change that will unlock a wider array of applications of facial recognition for the retail industry in North America and Europe, according to Brostoff, is the evolution of public understanding of and comfort with the technology based on experience.
As customers become more comfortable, new applications of biometrics will be widely deployed, and customer’s comfort will be further reinforced. As always, the devil is in the details.
Those details include not just what applications will leverage biometrics, and in what order, but also break down into different modalities, and delivery models. Facial recognition can be 2D or 3D, and can include a range of liveness detection or anti-spoofing mechanisms.
Goode points out that retailers and their technology partners in Europe are almost exclusively working on device-based biometric applications, because although GDPR does not directly ban or discourage the use of cloud-based systems, they are considered to make compliance significantly more challenging.
“It’s just not worth it in terms of risk, reputational damage et cetera if something goes on with that, so they’re looking at decentralized as much as possible,” Goode says. “China, India, Russia, parts of Latin America — Japan has being using centralized biometric databases for a long time, so they’re going to be fine with it. Ultimately, we have to talk about specific deployments.”
Trepp notes that FaceFirst is receiving a lot of interest from retailers in North America about transitioning towards cashier-less environments, but that doing so today would be major undertaking, which could cost tens of millions of dollars. Beyond that, in a hypothetical scenario in which two stores side by side offered the same goods at the same cost, but one was completely automated and the other was not, Trepp asks; which would people choose?
“We don’t know the answer to that question yet,” Trepp admits, though he has his own opinion. That makes gradual expansion of uses of facial recognition much more appealing in the short term.
Media scrutiny is increasing along with the technology’s use. Retailers and technology providers should both be concerned about the headline of a recent article in New York Magazine’s Intelligencer: “Smile! The Secretive Business of Facial-Recognition Software in Retail Stores.” The article calls the technology “almost completely unregulated” and says that no one can tell how widely it is used except its vendors.
The uncertainty prompted a warning in the National Law Review for retailers and technology companies to consider how privacy laws and regulations may evolve when planning new deployments.
Trepp and Brostoff both say industry stakeholders have a responsibility to their retail customers, and their consumer customers, to encourage effective communication about the use of biometrics.
“We feel it is incredibly important that they are as transparent as they can possibly be with their customers,” Trepp asserts. “We’re always advocates for signage. We’re advocates for FAQs that help people understand how the system’s being used. There’s a ton of misinformation out there about how this is being used, and how mismatched people go to jail and things like that. There is not a scenario where FaceFirst is involved, where someone could possibly be put in jail because they are matched on our system.”
Regional differences go even beyond customer attitudes and regulation. Some Asian markets are weighing the use of facial recognition for payments against a very different set of alternative payment options from those available in North America and Europe.
“I think what we’re likely to see is that the production roll-outs will happen on a broader basis in Asia before we see them in the United States, Canada, or Western Europe,” Brostoff predicts. “That doesn’t mean that the technology isn’t being trialed here, or that the capabilities are limited. We have an infrastructure in place that people are comfortable with now. In China, they don’t have that.”
In North America and Europe, familiarity and education could lead to greater adoption and acceptance, but could also lead to new demands from consumers.
“We believe that when the general population becomes more educated about exactly what’s going on with facial recognition, that there’s going to be a much larger percentage of the population which wants to control where there faces can and cannot be recognized,” Falck says. “They’re going to want more granular control.”
Ultimately, retailers are continuing to expand their considerations of facial recognition technology, out of necessity. Goode points out that back-of-house deployments to reduce “shrinkage,” or employee theft, are close enough to biometric time-and-attendance systems to include them among the low-hanging fruit for retail facial recognition, even though customers may be unaware of the application. Each application, however, builds familiarity between retailers and the possibilities the technology has.
“Facial recognition and the engines that power it are really uniquely positioned to be able to address some of these things where there’s not really any other technology that can do what people are hoping it can provide,” Trepp argues.
Goode agrees that the technology will continue to address a wider range of applications in more and more locations, but also says that retailers are slow to adopt new technology.
“Retailers are going to take a considerable amount of time for all of this to get integrated and embedded in stores,” he cautions.
That provides time for the infrastructure conditions to change to make advanced applications like facial biometric payments more appealing, Brostoff says, sounding a note both optimistic and cautionary.
“The infrastructure in this case is people’s perspective. That can change quickly, look at what Facebook was able to do to get us to open up about ourselves.”