DUBLIN -- Merchants at MRC Dublin 2018 received some tips on what they can do to prevent “friendly” fraud, as well as recoup losses from it, courtesy of two experts from Uber.

Friendly fraud occurs when a consumer makes a purchase online with their credit card, and then disputes the charge after receiving goods or services. Alice Cheung, product manager for Uber, began by noting that friendly fraud has become a major problem for the retail industry. Although retailers on the whole are far more concerned with “true” fraud, friendly fraud is not something to be ignored; it is estimated that friendly fraud losses will total $25 billion by 2020.

PREVENTING FRIENDLY FRAUD

Tal Yeshanov, principal, strategic payment risk initiatives for Uber, noted that while machine learning models are can be proficient at detecting true fraud, they not particularly good at detecting friendly fraud. This is because they are designed to identify major anomalies. She presented two sets of card charges on Uber Pool—one for several fares in San Francisco that occurred over a few days, and one for fares in much higher amounts that occurred in multiple countries, all in one day. While the former was actually friendly fraud, it doesn’t arouse suspicions and machine learning models can’t catch it. The latter, in contrast, stuck out immediately because it was obvious criminal fraud.

Yeshanov provided some tips to attendees on how they can prevent friendly fraud.

Implement a way for customers to self-resolve issues. Yeshanov noted that, on occasion, Uber charges $5 cancelation fees when customers opt not to take a ride in their vehicles. “We know that customers get upset when they’re charged for a ride they didn’t take, so we give them the option to go into an internal help center within the Uber app, to report that they had an issue,” she said. “We tell them why they’re charging them, and we give them access to our cancellation policy to let them know that this isn’t coming out of nowhere.”

If the rider still wants their money back after that, they can submit a complaint. And most of the time, if the user doesn’t abuse this service, Uber refunds the money. “This helps us deflect the chargeback from even happening,” Yeshanov said.

Use merchant descriptors to give cardholder context about what their transaction is. Sometimes, cardholders can get confused and don’t remember that a transaction occurred. Uber uses two different descriptors (one for Uber Trips and one for Uber Eats) that can show users what the product was that they purchased. “And you can update the merchant descriptor field. Work with your payment provider. You can give them the name of a sub-merchant. You can give them the exact time,” Yeshanov said.

Make it really easy for your customers to reach you. Your customers are most likely going to go to the bank. You want to encourage them to go directly to you, the merchant. So provide a help center, phone support, social media, etc. so that you can connect to your customers. “When you encourage your customers to reach out to you, not only do you get valuable information from the customer about their experience, but you can also deflect the chargeback from coming in,” Yeshanov said.

Clearly create a robust, helpful help center. Taking the previous tip one step further, Yeshanov recommends building out your help center so that customers can find solutions to their problems. “In our help center, we have a completely separate section for accounts and payment options,” she said. “Within that subsection, we have another subsection called, ‘I have an unknown charge.’ An unknown charge can happen for a number of different reasons; it could be true fraud, but it could also happen because you just don’t recognize the charge. So we let users go into the help center and give them this context as to what’s what.”

Clearly lay out your terms of use. Most users don’t actually read the terms of use, but for the ones that do, it could be helpful to them. And it can be helpful to you. “When you’re representing your chargeback, you can say to your issuing bank, ‘We provided a service, and these are the policies behind how our service operates,” Yeshanov said. “When you can tell your issuing and acquiring banks what your terms of use are, and you can say that the cardholder has agreed to those terms of use, you have a very strong case that will empower you to win your representments.

Clearly communicate your policies to your users. This is one step beyond laying out the terms of use; you should also display your policies within your app. Let your customers know when they will or won’t be charged.

RECOUPING YOUR LOSSES

Cheung presented attendees with some ways that they can recoup some losses from friendly fraud.

Apply a negative balance. When Uber incurs a chargeback, it will put a negative balance on customer’s account. The customer will then be unable to take a ride until they repay the money. “That really helps us to not let people abuse the system,” Cheung said.

Do representments—when it’s worth it. Once you start to identify your friendly fraud criteria, you can decide whether you want to represent or not. “Representments for a lot of issuing banks can cost a dollar or more per representment, so you have to think about whether it’s worth it for you to represent for a transaction,” Cheung said. “For us, we don’t represent anything under $5, because even if you would win that representment, you would still be ROI negative overall in the case.”

If representing, create strong compelling evidence documentation. If you decide to represent, you have to put together a set of compelling evidence to send to the bank and dispute the chargeback. Cheung provided an example of evidence that Uber shares with its issuing bank. “One is the proof of purchase, which is the receipt that we get from Uber. Within the receipt are a lot of details, the amount to be paid at the time of service, the type of product, the pickup location and the drop-off location. It has a lot of information about your product, which really helps us prove to the bank that we provided the service to the customers,” she said.

Run tests to optimize your representment strategy. Cheung noted that  representments are difficult to get right the first time. “Within Uber, we do a lot of A/B testing. It’s not only testing on how we detect friendly fraud; we try different algorithms and different signals,” Cheung said. “We also do a lot of testing in terms of what kind of compelling evidence would be useful. Play with your data, play around with your transaction, and really figure out what the best strategy is for you.”

Chargebacks are always a hotbutton issue for merchants, and are sure to be a topic of discussion at the upcoming AFP Retail Roundtable on June 21, 2018 in Denver.