The company has linked with startup Yofi to protect itself from bad actors—those who perpetuate abuse, fraud or scams in violation of companies’ policies—in stores and online. And in the process, the sportswear retailer determined that a few of its bad actors had come from the inside.
Using AI that identifies and aggregates information about consumers—from how many email addresses they might have, to their typical shoe size—siblings Jordan and Alex Shamir’s startup has become adept at catching bad actors and fraudsters of all kinds.
Chris Bone, director of operations at JD Canada, said prior to implementing Yofi, it was a bit more difficult to track policy abuse infractions—but Yofi’s AI-powered systems aim to link together information on bad actors that might get glossed over by retailers’ standard back-end systems. Its prowess at linking together separate instances of abuse was why Bone and his team became interested in the product in the first place—originally on the customer side.
“The sneaker industry was banging and bot mitigation was a hot topic. Yofi was on that forefront, as far as bot mitigation goes, so that’s originally how we got onto them, is to help us do that,” Bone said, noting that he and his team were also attracted to Yofi’s user spoofing capabilities—that is to say, cleaning email lists to remove fraudulent addresses from marketing lists, in turn, decreasing marketing spend.
JD Canada has a keen interest in ensuring real, human customers—not bots—get first dibs at its items. By suppressing bots and disallowing their purchases, the company can better serve customers, who are more likely to find the item they are looking for during a limited-release period if bots lose the ability to snatch up stock.
“The reason why we don’t want bad actors to get the product is that we can get the good actor the product—the person that we know that we’re going to have for a longer retention,” he explained.
But bot facilitators are far from the only type of bad actors JD Canada deals with on a monthly basis. Yofi has also been able to help the retailer pinpoint instances of bad actors who had dozens of accounts, which they were likely using to purchase goods for resale.
“We were able to identify people that maybe we didn’t think were bad actors, but they actually were because they had 40, 50, 60 accounts,” he said.
In these cases, if the goods didn’t sell ahead of the deadline to return them, the buyers may send them back—or bring them back to the store. But that kind of behavior poses a profit loss issue for JD; the company has to handle the expenses and labor associated with handling returns, but it also risks having low stock on items loyal consumers want to purchase if the items are tied up in the hands of resellers.
Not all cases of abuse were external. In some cases, they have been hiding in plain sight.
With the help of Yofi’s AI systems, the company determined that a small percentage of its employees had been violating company policy by abusing employee discounts, then subsequently reselling products purchased.
“Our policy is for personal use, and you might look at some profiles and be like, ‘That [purchase] is not for personal use,’” Bone told Sourcing Journal. “There’s no reason why you should have every size from size four to size 12 women’s shoe, and they were bought on individual purchases to not make it look like you’re buying a whole size run.”
Depending on the severity of the issue the company has with an employee, they may be terminated, given warnings or face disciplinary actions.
“Not everything led to a termination; if you were buying a size nine and a size four, I might just ask you, ‘Why?’ And if you say, ‘Sorry, I bought my friend some,’ we’re just going to have a conversation around it. But there’s other cases where [violations] are more severe, so some of them have led to terminations.”
Bone said JD Canada’s work with Yofi on identifying internal bad actors is still in its early days.
For Yofi, pinpointing all kinds of bad actors—internally or externally—requires a holistic understanding of how they operate.
The information that helps make determinations about whether someone is a bad actor comes from thousands of data points, including information scraped from popular delivery companies and behavioral data—like the frequency with which they return items or how much the size they choose to purchase varies. While data like telephone number, email address and physical location can help crack that code, patterns also show up in the physical actions bad actors take when purchasing, Jordan Shamir, Yofi’s CEO, said.
“One of the big things we start with is what we call telemetry. We look at the device signal; we look at how they’re typing on their phone, whether the phone is moving—are they driving in a car?” he explained. “We start, before the transaction ever happens, forming what we call a digital body language.”
If those indicators—paired with others—raise flags for Yofi’s system, it may block a transaction, depending on what its individual clients are trying to prevent; in some cases, discount abuse; in others, bulk purchases by resellers that eventually get returned.
Yofi allows all its customers to join a consortium, where it anonymizes and aggregates data on consumers, then shares it with other customers. Shamir said every customer Yofi has onboarded has opted into that program—though that’s not necessarily what he expected when he first introduced it.
“Data sharing is a big leap of faith; most people don’t want to share data. However, this is one of the unique use cases where most retailers want to share data, because they’re only as good as their own system and fraud moves in a million different directions,” Shamir said. “It’s not necessarily Nike vs. Adidas trying to sell a pair of shoes; it’s, how do we work together to fight a common enemy, which is a bad actor, a fraudster?”
For Bone, the decision to share was easy—with more information about customers, his team could more easily identify bad actors and legitimate, loyal shoppers.
“Anytime we have more information to make decisions, it’s helpful, so the more information that Yofi can aggregate for us, the better,” he said.
Bone said that in addition to reducing the impact of bad actors, which helps loyal customers have a better shopping experience, JD Canada has started to work with Yofi to benefit what he calls “good actors.” While that process is a work in progress, the company has started to identify its best customers, which could later lead to extra discounts, expedited returns, free shipping or otherwise as a reward for their continued, legitimate business.
Given the trend toward higher levels of personalization and an onus on customer service and loyalty programs, additional information about good actors can help with segmentation and can make marketing dollars go further for brands and retailers—particularly smaller companies.
JD Canada’s interest in segmenting its customers has extended into its in-store operations, as well. Yofi and the sportswear retailer have begun partnering to link customers’ in-person actions to their digital actions; that piece of the program is still in its early days—and can be difficult to track with the increased anonymity in-person transactions can provide—but Bone said he expects it will allow the company to make better correlations and inferences about consumers.
Today, the fact that in-store and online systems often operate differently can be a boon for retail crimesters, return abusers and other types of bad actors. Simultaneously, it can cause headaches for good actors. As the Yofi team continues to find touchpoints to piece together customer profiles based on omnichannel interactions with brands and retailers, Shamir said using the technology in stores—which one in five Yofi customers already do—has potential to benefit brands and retailers from a loss prevention and abuse mitigation standpoint.
“A lot of what we think about is, how do we holistically prevent and protect a business?” Shamir said. “Abuse is abuse. It happens online, it happens in stores and it happens both from internal employees and external customers, and what’s really important for us is helping to tackle and understand this issue, [which means] you have to look at all sides.”
