Payments Big Data: 3 Tips on Profit Maximization for Retailers


Big Data, you may have heard it used as a trendy buzzword, but what does it really mean? It is simply the practice of analyzing extremely large data sets to reveal trends and patterns. Big Data can be used to predict outcomes in many disciplines, but as a business owner and retailer, have you ever wondered how people are applying Big Data to payments?

Data analysis is nothing new. Marketers have used it to segment their target audience to craft the perfectly relatable marketing messages for a long time now. Data is currently being used in the same way in the payment processing industry to prevent fraud. Visa has used it to identify tens of thousands of dollars in fraud.

For internet retailers and eCommerce sellers, using payment processing data to prevent fraud is essential to keeping payment processing. In fact, chargebacks due to fraud are one of the leading causes of getting a merchant account frozen or shut down permanently.

Like Visa, leading retailers are utilizing Big Data to spot dangers and opportunities by marrying payments data with advertising data. Here are three areas which are ripe for continued mining:

  1. Reduce Chargebacks and Fraud.

Cybersource estimates online fraud takes approximately 0.9% of a retailer’s earnings. Analyzing chargebacks and fraud is a critical step in chargeback mitigation and keeping your payment processing solutions running. To maximize the results, the analysis must be run across dozens of metrics including lead source and campaign data. Some marketing campaigns result in higher chargebacks due to an increase in customer confusion. While in other cases, the marketing campaigns had no relation to the chargeback. Adding in payment information, such as what we know about cardholders—which is typically their name, address, phone, card type, issuing bank, device fingerprint, IP address, and Web browser info—allows for a more robust analysis.

  1. Payment attributes as a predictor of customer lifetime value.

By combining the campaign and payment data, marketers are able to gain clearer insight into how factors such as the type of card a consumer uses (e.g., debit vs. credit) influences the lifetime value of the customer. Some of our savvy retailers are using this data to more accurately forecast customer lifetime values and more effectively distribute their ad spend.

  1. Employing payment features and routing.

Many retailers are using payments attributes, or third-party tools that use them, for making real-time go/no-go purchase acceptance decisions. However, there is another world:

  • Reduce declines. Many larger enterprises are using payment attributes to route the payment to maximize approval throughput. A basic example of this principle is routing to the closest acquirer (if a multinational has an acquiring relationship in Spain and the United States, routing the U.S. customer to the U.S. bank will typically be approved more often). This principle is applied heavily for recurring payments, where declines can account for a significant percentage of churn. Analyzing when best to attempt that recurring charge can yield big returns on investment.
  • Reduce costs. Pricing can also vary based on card type, transaction amount, etc. Your best negotiated price may vary across multiple vendors. Retailers are routing to reduce costs based on the rules of their contracts (e.g., a deal for debit card processing vs. credit).
  • Target marketing offers. Some retailers will provide specific offers influenced by the payment attributes. These may be special or limited offers, intended to drive up customer lifetime value for that specific segment.

Analyzing payments is very similar to analyzing marketing data for trends. It can be more complex, because much of payments data is sensitive, needs to remain secure, and can be difficult to access. In some instances, it’s just impossible to do with certain payments vendors.

Where does analyzing marketing and payments data fit into your strategic priorities for the upcoming year?