The stress of the holiday season is intense, and all the more so for eCommerce fraud teams. Soaring order volume means there is less time to thoroughly review orders, and teams that hire seasonal help to keep up with the load have to rely on analysts that are less experienced at the tedious task of uncovering Card Not Present (CNP) fraud. Among the influx of orders are many new customers who present a huge opportunity for merchants, but they can also appear threatening–especially if they’re from new and unfamiliar markets. It’s crucial for merchants to be prepared for the fraud patterns that are likely to emerge between Black Friday and New Years.
Riskified helps hundreds of eCommerce merchants manage their fraud during the busy holiday season. We’ve drawn on our experience to present some tips on analyzing and organizing your data, so that your fraud team is situated to optimally handle the surge in orders, drive revenue, minimize false-positive declines, and maintain a smooth shopping experience for your customers.
Analyze past performance so you know what to expect
Fraud detection, like most sciences, is all about learning and adapting. And as with any science, past data should be constantly reviewed in order to refine methods. Merchants who fail to learn from previous mistakes have little reason to believe that the rate of costly chargebacks should decrease from one year to the next. But, deciphering a common thread among last year’s incorrectly approved orders can mean a big uptick in efficiency this year.
According to Riskified data, retailers usually see a 100% increase in the number of purchases made with international cards in November. Merchants can benefit greatly from scouring past data to determine which overseas orders were surprisingly safe, and which should be treated with caution.
Most merchants are likely to discover that they’ve been overly risk averse during the holidays. Our analysts have determined that top holiday sales days are actually far safer than average shopping days, and that any given order placed during the holidays is 55% less likely to be fraudulent. Partially because merchants were unaware of this, 4 out of 5 orders rejected during last year’s holidays were, in fact, legitimate.
Bearing all that in mind, it is unwise to become too dependent on historical data. Any economist or statistician will happily remind you that the past is not necessarily a good predictor of the future. This is particularly true when conditions change. For starters, the ratio of eCommerce to in-person sales has been consistently increasing year over year. Even more profound is that fraudsters are constantly changing their behavior; they utilize more complex proxies than ever before (including mobile proxies), and the rate of account takeovers, a particularly devious form of fraud, has only been increasing. Analysis of past data is crucial, but rigid rules-based detection systems can backfire. Much more effective are machine learning algorithms, which refine their rules in real time.
Returning customers are paramount during the holidays
One of the most basic, but crucial, things to know about an order is whether the customer behind it has shopped with you before. The fraud rate among returning customers is about half that of new customers: 1.4% compared to 2.6%. This means that returning customers should be treated very differently than new ones. This distinction is especially critical during the holidays, when order volume is so much greater.
Unfortunately, detecting returning customers is surprisingly tricky. Most merchants link orders only by name and email, but this is often insufficient. If a customer with an unrecognized name and email places an order, but has a familiar IP address, it’s possible she just got married and changed her name! A sophisticated linking system which can recognize customers based on fields like IP address, device fingerprinting, email domain and product is likely to help identify returning customers and reduce false decline rates.
Riskified’s elastic linking data gives merchants an even more robust look at their customers. Even if a customer is new to one merchant, we are often able to find previous shopping data on this customer within our database. We already “know” 17% of customers that our merchants have never seen before, and are able to safely approve 96% of orders from these customers.
Start preparing for your January chargeback disputes now
Chargeback disputes are all about data. The more historic data a merchant can access, the more likely they are to corroborate the identity of returning customers. This link is one of the most compelling data points for chargeback disputes.
This means it’s extremely important for merchants to ensure they have a well maintained log of customer correspondences; data from September will help with post-holiday disputes in February. Particularly important to catalogue are shipping confirmations. We recommend merchants save these as screenshots or PDFs, because after 2 months they are no longer available in shippers’ systems! Additionally, merchants should make sure they are enriching their data with third party sources like White Pages, Emailage or Facebook. Proving address or email matches across these sources is very compelling evidence for the absence of true fraud.
Even if a merchant keeps immaculate records, a third party fraud prevention partner should be able to provide further evidence to help with chargeback disputes. Make sure to check if your partner can provide additional data related to credit card, phone, IP address and more.
We hope some of these tips help you to prepare for a successful, fraud-free holiday season. Riskified is always happy to help – whether by lightening the load on your team, recouping revenue from declines, or simply sharing best practices.
For more advice on preparing your fraud team for the holidays, download our free guide.
To learn more, feel free to contact us.