Fraudulent purchases cost retailers money. But the technology to prevent fraud also costs.
Online merchant Sportitude needed to keep its fraud at bay, but its costs continued to increase. After its fraud prevention costs doubled in 2020 compared with 2016, the Australian-based sports footwear and apparel merchant decided to switch its technology vendors, says Roumen Staykov, CEO of Sportitude.
Sportitude chose ClearSale, as the vendor’s pricing model differed from its previous provider, and its costs decreased 55% in 2022 compared with its 2020 pricing, Staykov says.
“It was not a small amount. It was well worth the move,” Staykov says.
Sportitude has ClearSale’s performance-based pricing model, in which the vendor charges a fee for every approved transaction that it processes. The percent per transaction is smaller than a credit card processing fee (which is typically 1%-3%), Staykov says, without adding more or disclosing its previous fraud vendor.
“We treat it as the cost of doing business,” Staykov says.
False positives decrease
In addition to the decrease in cost, the number of transactions that ClearSale flags as possible fraud has decreased as well. This saves the merchant time, Staykov says, as the retailer has fewer transactions to verify.
“It’s a win from many angles,” he says.
Both vendors appropriately flagged fraudulent orders, and Sportitude’s rate of fraud attempts and chargebacks has remained about the same, he says.
Of all of Sportitude.com’s credit card transactions, about 1%-2% are flagged for fraud. Of those, more than 90% are actually fraud, Staykov says. This is good, he says, as it means it’s not getting that many false positives.
This is beneficial for both shoppers — as it is not inconveniencing too many shoppers with authenticating their purchase — and the merchant. For example, once the software flags an item as potentially fraudulent, two senior members of Sportitude’s customer service team will manually review each transaction.
“It used to be a lot more time,” Staykov says. “It would be a larger percent of that person’s role. The more frequently it occurs, it takes a lot more time out of day, follow up and investigating.”
Verifying if flagged orders are fraud
One test the merchant conducts to confirm credit card ownership is processing a refund. Sportitude employees will process a return for a few cents on the credit card and have the purchaser confirm the refund amount. Half the time, the purchaser never responds, in which case, Sportitude knows the transaction was fraudulent and will decline the purchase.
Sportitude has played around with faster ways to confirm credit card ownership — as processing a refund takes a few days — but it has found this is the best balance of catching fraud while maintaining good customer service. Luckily, it’s a narrow group of legitimate customers that have to go through this process, and most are understanding when the retailer frames it as a “security step,” he says.
With Sportitude’s previous fraud-prevention provider, a lot more transitions were flagged for fraud that turned out to be legitimate. This inconvenienced more of its legitimate customers to authenticate their purchases, plus took more of Sportitude’s time. Now, its customer service representatives can handle more of those typical issues rather than having to spend large chunks of their day reviewing fraudulent orders, he says.
When a chargeback occurs
If Sportitude does confirm a transaction that winds up to be fraud, this will show up as a chargeback from the bank. What this means is that the criminal used stolen credit card details and that card owner saw a purchase that wasn’t hers on the statement, and had the credit company reverse the charges.
If a chargeback does go through, Sportitude can go into ClearSale’s portal and note which transaction wound up being approved even though it was fraud. This achieves two things: first, ClearSale will refund the purchase to Sportitude if it provides the shipping notification and the chargeback notification.
Second, ClearSale will update its artificial intelligence model that its technology uses to flag potentially fraudulent orders. When a retailer notes a nefarious transaction, its model then learns from the markings of this transaction to hopefully catch it in the future, says Salvador Tello, global senior, enterprise presales, at ClearSale.
ClearSale uses a machine learning model to assess if transactions are risky or good. Like most retailers, it also has a manual review team that will look at orders that its technology flags to then approve or decline. These fraud analysts use eight external sources that are “very powerful” in helping dig into different elements of the transaction’s fraud score before making a decision, Tello says.
For example, analysts can use proxy piercing technology to determine if the purchase is coming from a consumer who is disguising her IP address and using a proxy server. Another example is using device fingerprint technology to identify computers placing transactions. In addition, analysts can also look at ClearSale’s own tracking technology about customer behavior across all of its clients.
One example of an error that the artificial intelligence tool may flag as possible fraud that a human could verify is a typo in the address or email name, says Tello. Even though the information is incorrect, a human could easily see the error. Humans can also gain additional insights from analyzing social network activity or doing a reverse phone and address lookup. All this information is later fed into ClearSale’s artificial intelligence model.
Overall spikes in fraud
It’s important to catch a fraudulent order not only for the cost implications, but also because it could lead to more fraud, Staykov says.
Sportitude typically notices a spike in fraudulent activity if one or two chargebacks slips through, he says. Criminals may be using a new pattern of transactions that got through that the AI model hasn’t caught up to yet, and then criminals will exploit it as much as possible until the retailer reports the issue and the technology catches up, Staykov says.
“If you catch the transition quickly and nip them in the bud, it definitely causes the decline,” Staykov says.
The retailer also notices a spike in fraud during high-volume sales periods. For Sportitude, that is during November and May, which is when many retailers in Australia host sales, he says.
Stay on top of the latest developments in the ecommerce industry. Sign up for a complimentary subscription to Digital Commerce 360 Retail News.
Follow us on LinkedIn, Twitter and Facebook. Be the first to know when Digital Commerce 360 publishes news content.Favorite