For decades lenders, retailers, marketers, and other professionals that rely on data for decisioning have staked their success on the assumption that the data sources they are using have enough information about their customer to make a robust decision. The assumption being there is enough information reported by various data users about a consumer or small business to a single data aggregator to make an informed decision. The classic example is a consumer lender furnishing credit history to a credit bureau which aggregates that consumer’s information across lenders into a single credit report (a give-to-get model). However in practice data users are constantly incorporating new data sources to try to squeeze out more informed decisions without considering their data sources have an implicit problem:
Many companies do not report to data aggregators resulting in incomplete customer profiles
The result is constant game of test, evaluate and on-board new data sources that eats up time and money trying to create a more complete picture of the customer.
Piece of the Pie
While the give-to-get model of data acquisition is workable in theory, in practice the result often creates an incomplete picture of the applicant, or of the customer. Many who use credit data try to complete the picture by acquiring data from multiple data aggregator sources (bureau A + bureau B + fraud vendor + public records vendor +…). However, this data integration exercise can require a large, expensive team just to integrate the various data aggregators into one cohesive decision. Additionally reporting customer behavior to different data aggregators requires a team that may not add to the bottom line. Even when the flow of data is well-managed, there is a lag in the data reported to the data aggregators that often makes even well-maintained data environments out-of-date. The task of acquiring and deploying a complete, up-to-date view of the customer remains relatively elusive for even sophisticated operators.
In the last several years, so-called “alternative data” has gained prominence as a solution to fill in the gaps of customer behavior. Data from non-prime bureaus, social media data, mobile phone data, web browsing data, and bank transaction data have each been developed and are now commonly available. While these new data sources can exacerbate the problem of data integration, clever users can nevertheless gain greater insights using alternative data. The emergence of alternative data sources has accelerated the deployment of big data solutions, and many users of data are making use of advanced machine learning and artificial intelligence to derive learnings from integrated data. In turn, much attention has focused on legal and compliance issues related to the fairness and interpretability of models and processes that come from these new analytics capabilities.
Enter Transaction Science
Amongst the alternative data sources, we believe that the most complete and relevant data for lenders are bank transactions. What other source knows as much about income, spending behavior, credit accounts that might not be reported to bureaus, and mobile payment spending? Bank transaction data can be the most complete, timely and relevant way to assess previously cloudy concepts like affordability and its relationship to creditworthiness. In addition, bank transaction data must be provided by the applicant or customer through their explicit consent. While this limits the applicability of the data to well-known processes such as credit pre-screens and pre-qualifications, bank transaction data engages the applicant in an intuitive and easily understood way.
Bank transaction data provides a customer picture of Behavior Made Clear
We at Transaction Science believe that bank transaction data is absolutely transformative in any industry, but it is inherently difficult to utilize in a way that offers true insights. These difficulties include:
- Difficulty in real-time processing Most, if not all users rely on a tiny fraction of bank transaction data (snapshot statistics) located in a header file to make important decisions because data engineering in just a couple of seconds is a daunting task.
- Difficulty in developing specialization with BT data Legacy credit bureau experience is common in the financial services job applicant pool, but very few have deep experience with the difficulties of bank transaction data. Development of a team to manage real-time bank transactions data can take years, while value goes unrealized.
- Difficulty with regulator concerns Whether bank transaction data should be used as the basis for consumer lending decisions. Some of these issues have recently been addressed by CFPB commentary and positions, but material issues remain to be addressed.
Transaction Science solves these problems and goes many steps further. In short, we make it all much easier:
- Solving production concerns The same attributes regardless of bank transaction provider – our patent pending process generates the same 5,800+ attributes for fraud, risk, and marketing purposes from any BT provider in a matter of seconds.
- Years of experience with BT Over 50 years of financial experience and real-world use of BT data means we are experts in BT and we provide that experience with data, models, and consulting capabilities.
- Address regulatory and compliance concerns directly Specific BT attributes focused on affordability measures that provide actual willingness and ability to pay attributes and indices – other sources have to estimate it.
- Products designed for your customer Attributes that provide the before and after of a purchase or credit product to determine how positively you have impacted your customer’s life by analyzing life events, impact of a credit product, changes in cash flows, etc.
We will be rolling out even more exciting capabilities with our partners in the consumer and small business lending space soon, but if you are anxious to know more right now – feel free to connect with us! Look us up on LinkedIn, send us an email, or call 858-935-4477.