Genome projects map the relationship between human DNA and musical preferences, so why not that of people and their cards?
Many companies are mapping the so-called payments genome to make more accurate predictive marketing decisions and improve fraud detection. Some vendors say their systems can be used to determine the products and services consumers will buy next.
"If you think of every transaction as one of your genes," then those transactions can be woven together and analyzed as part of the "DNA [that] determines who or what you end up being," says Schwark Satyavolu, chief executive and co-founder of Truaxis in Redwood City, Calif.
The use of transactional data in the cards loyalty space grew out of a desire to get more granular data about customer purchasing behavior, says Alan Mattei, a partner for Novantas LLC.
The next step for financial institutions is to find less-profitable customers whose DNA resembles that of their best ones — and sell them products and services they'd probably like to have. "I can look at suboptimal [purchasing] behavior" and attempt to change it "so it becomes advantageous to the institution," Mattei says.
Until recently, most banks were staring into a data void with reams of unstructured data they could do little with, says Jim Craig, vice president of marketing for 1st Advantage Federal Credit Union of Yorktown, Va.
"We did not have an easy way to slice and dice any of our data other than to dump it into a pool and say this member spent X on the card for Y level of transactions," Craig says.
Transaction data has often been described as "a mile wide and an inch deep," since banks have a great view of the breadth of consumer spending but little detail about each individual transaction. By contrast, merchants have a vertical view of their customers' purchases, but not one of where else they are buying. The breakthrough for marketing purposes would be to connect both of these views, vendors say.
Bank data is shallow, but Satyavolu says plenty can be inferred about consumer behavior from what's there, and that is how Truaxis has built its algorithms. "There are a number of different data points that are leveraged to create this richness," he says.
To deepen typical bank transaction data, some vendors have turned to crowd sourcing, meaning they request input from consumers over an open platform like social media.
One vendor, T8 Webware of Cedar Falls, Iowa, claims to dissect consumer transactions using a modified version of an algorithm gene scientists use to create new species of crops, called a Hidden Markov Model. "The algorithm yields subsequences that can be used to compare new transactions against a known result," says Wade Arnold, T8's chief executive.
Its mobile financial management tool maps transactions to specific locations. Users can add data about that merchant for their own future reference, and in doing so, they help T8 figure out who is shopping where. As more users enter information about the same merchant, the data becomes more valuable to bankers. Such data would be useful not only to make recommendations and cross-sell products, but to sign the merchant up as a customer on the strength of the market intelligence they have, Craig says.
Similarly, the financial management and fraud detection provider BillGuard, of New York, uses crowd sourcing in its transaction database to fight fraud. To flag anomalies, its engine scours the Web to compare consumer complaints against transactions its own customers are logging on their statements. "It's a learning-based algorithm that gets smarter with each user and each click," says Yaron Samid, BillGuard's CEO and founder
It's inevitable that banks' efforts to fight fraud would intersect with their marketing activities, and transaction analysis may be the nexus, experts say. At banks, "the fraud experts are getting more information than the marketing people are," says Avivah Litan, vice president and distinguished analyst for Gartner Inc., because they are looking at customer activity across bank channels. "Now the marketing people are knocking on the fraud people's doors," Litan says.
Banks are touchy about anything that seems to imply they are invading customers' privacy. And they need to be careful here with transaction fingerprinting. "Anytime you talk about using this kind of information about the customer they are concerned, and rightfully so. So we address this by making sure the data used is anonymized," Craig says.
1st Advantage uses a product from Micronotes Inc. of Cambridge, Mass., that performs regression analysis on each customer's transaction data to cross-sell other products. "We think it is very powerful to use data to the benefit of the consumer, to whom the offer will feel on the mark and relevant, and to the merchant who has never been able before to describe with hundreds of predictive variables what their most profitable customers look like," Micronotes CEO Devon Kinkead says.