The gift concierge service that uses artificial intelligence is part of 1-800-Flowers’ technology and service strategy.
Since implementing a service from fraud prevention vendor Subuno, A4C is flagging 65% fewer orders as potentially fraudulent. The system gives the retailer more flexibility with fraud screening by allowing A4C to set different rules based on the amount of orders and other variables.
A4C.com, an online retailer of electronics and cell phone accessories, had a problem it wanted to solve before the all-important holiday season.
The retailer was hit with about 300 fraudulent orders during the 2012 holiday season, costing it around $60,000. To keep criminal activity at bay, it went to the fraud-prevention extreme by manually reviewing every order and also flagging too many orders as potentially fraudulent, says Dov Barber, director of marketing at A4C.com. For example, if the shipping and billing address on an order didn’t match, an order was flagged as risky and sent off for manual reviews that often took around five minutes apiece.
Making matters worse, customers were not informed if their order did not go through. This process left many customers hanging, and wondering what happened when they were not billed or never received their order.
“If the customer has left the site thinking the order is fine only to find out the order wasn’t processed, we most likely lost that customer, whereas if we could let them know right then and there, they could perhaps adjust their info and retry,” Barber says. “We were losing too many transactions, which was hampering our sales growth.”
To deal with the issue, the retailer in October implemented a fraud-prevention system from Subuno. The system, which A4C.com integrated into its Magento shopping cart, allows merchants to configure how they want orders checked against various fraud detection vendors without having to create separate connections and accounts with each vendor.
Subuno customizes the system to a retailer’s specifications, Barber says. For example, one of the vendors Subuno connects to is Maxmind, which runs orders through a points system based on various order details. Each order is scored on a scale of 1 to 100; the lower the number the better the chance of automated approval. Other services integrated with Subuno verify shoppers’ e-mail accounts and look at geographic locations and I.P. addresses, flagging orders that stem from high-risk regions.
Subuno gives A4C added flexibility and the ability to account for what Barber calls grey areas better than any single vendor can. For example, with the Maxmind points system, an order with a score between 35 and 65 may always be flagged for review. With Subuno, A4C is able to set different requirements for lower-priced carts, or set different rules for when shipping and billing addresses do not match and for international versus domestic orders. For example if an order is under $10 and being shipped to a U.S. address, screening is less stringent than for an order that is more than $50 and being shipped to an address outside the U.S.
Using these rules, the system checks instantly for potential fraud and informs the customer if the order goes through or not, Barber says.
“We can preset how deep we want to go in any scenario and when we want to be notified about a potentially risky order,” Barber says. “There is an intense amount of automation built in. There is no need to monitor orders live as they come through unless one is flagged. There is no setting to hold for review. This allows for a lot more efficiency to the process and maximizes productivity on our end.”
Subuno charges a base price according to order volume. The cost per order for review ranges from $.005 to $.08 per query, depending on if the check is automated or manual.
Barber says with Subuno, A4C is now able to automatically process many orders that it would have flagged before. Since implementing Subuno, A4C has reduced flagged orders by more than 65%, and only six of the approximately 75,000 orders placed with the retailer since it implemented the service have turned out to be fraudulent.