Bankers have seized upon “customer-centricity” as a core value, recognizing that building shareholder value requires banks to create value for their customers. But why doesn’t this core value translate to data mining?
Banks enjoy a forensic level of data about their customers. In the B2B space, banks can infer:
- How their customers manage and invest cash.
- The dynamics of the supplier and customer base, and even
- The degree to which customers have automated their accounting and financial supply chain activities.
Yet all too often, data mining efforts have produced modest, at best, results – and in some cases, outright failure. Mention the phrase “lead lists,” to a commercial banker – chances are, their eyes will roll. “Lead lists” are emblematic of what is wrong with data mining in banking.
Why do Lead Lists Suck?
Your typical lead list views the customer as a potential wallet of loans, deposits and fees. Inspiring, right? Just like a butcher sees a cow as cuts of meat, wallet models see the customer as chunks of money.
The model identifies “missing products” that the sales person has neglected to sell the customer and instructs the sales person to “go get” the missing products. This is a transactional model that views the customer as a finite, zero-sum set of services and the banker’s job as matching products to customer spend profiles. I’ve sometimes heard this type of sales role described as a “dork with an order book.”
Sometimes the model will be sufficiently sophisticated that it will compare the loan wallet to the deposit and fee wallet, identifying customers who are relying upon the bank’s hard-earned capital and failing to reciprocate with a commensurate level of deposits and fees. This more sophisticated model assumes credit is a scarce good and a loss leader that must be offset with valuable deposits and fees. The banker’s job is to “get their fair share.” This approach works very well for customers with weaker credit health – but how many banks want to build a whole portfolio of such customers?
Any good sales person will naturally rebel against these kinds of lead lists. At Deluxe we have observed banks in one breadth tell their sales people they are strategic advisors – and in the next breath, hand them a lead list with a set of products the customer should buy. Inevitably, when these lists fail, the reaction isn’t: maybe this is the wrong approach. The reaction is to view the sales people as stubborn and unwilling to use data in their sales approach. In fact, sales people should be applauded for throwing these lists into the trash.
Treasury 3.0 and Sales as a Value-Creating Activity
Treasury Strategies coined the term “Treasury 3.0” to denote a fundamental shift in how Corporate Treasurers function – and how successful banks will operate. In Treasury 3.0, Treasury groups are focused on higher-order functions – managing liquidity, risk, financial supply chain – and rely upon their bankers to understand their needs and craft solutions – solutions that incorporate multiple products integrated into the customer’s processes, data architecture and governance. In a Treasury 3.0 world two different banks can extract wildly divergent value from the same product set:
Advisory Bank “A”
- Diagnoses customer needs
- Brings knowledge of industry vertical and financial supply chain
- Sells broadly and deeply by meeting operational and strategic needs of customer
- Prices in a manner consistent with the value bank is creating for customer
Transactional Bank “B”
- Understands customer use of banking products
- Cross-sells products into bank’s credit base
- Prices based on credit position and strength of relationship
Which bank do you think has the more valuable franchise?
Customer-Centric Data Mining
What would data mining look like if it were customer-centric? First, we would not solely focus on making money for the bank – but would instead focus on creating value with the customer. Don’t worry, we are still going to make money!
The data mining Deluxe conducts with its bank clients begins with opportunities to create value for customers. We have inventoried value creation opportunities for businesses, ranging from the trivial to the sophisticated:
- Save time and money by remotely depositing checks vs. manually depositing checks at the branch (trivial, but this opportunity still exists!).
- Automate cash application by using artificial intelligence to capture information across all media types (including PDFs, emails) and match collections to the open Accounts Receivable ledger.
- Deploy statistical tests to determine the best methodology (channel, frequency, messaging) to collect outstanding receivables.
Each of these opportunities has a business case that you can show to a CFO, VP of Finance, Treasurer – and they will recognize it as correct. That’s the easy part – though it’s the step most banks skip. The trick is to infer the variables that go into these equations – and that is where data mining comes in.
The rich data banks enjoy can help infer the variables to drive these business cases. Consider the cash application example – how do you begin to estimate a company’s current and potential cash application rate? This seems like an impossible task, because it is an internal variable that is not reported to the bank. However, such variables can be estimated with a shocking level of accuracy by using available bank data. Consider the following for a collection bank:
- You know the customer’s industry – and thereby the range of current and best-in-class cash application rates
- You know the customer’s mix of incoming collections and through this can infer the likely mix of remittance types
- You know the customer’s relative sophistication in extracting and leveraging digital information – and through this can infer their likely level of success in developing a quality posting file for matching purposes.
Yes, there are going to be idiosyncratic behaviors that the model will not capture – however, these directional estimates are highly reliable for understanding the relative magnitude of an opportunity facing a customer. And what’s more, a good sales process will position the bank to update the business cases with better data, improving the accuracy and reliability of the business case.
Because banks are for-profit concerns, we also calculate the benefit the bank would receive were it to help the customer solve this opportunity (fees, deposits, retention, reduction in bank costs). As a result, the bank can meet its needs by focusing on prospects that offer value both to the customer and the bank.
The resulting lead list is no longer a collection of missing products – it is a set of ideas for how customers can make their businesses better through improved efficiency, increased revenue yield, reduced risk.
As a result, the sales people are now engaged – you’ve just given them tangible ideas with which they can help their customers improve. And as they use the output of the data mining, they will be an ongoing source of ideas as to how to improve the models and their application.
A Call to Action
The next time you engage in a data mining exercise, take a step back and ask yourself: what is the value we are going to create for the customer? If you can’t answer this question – and if the course of the data mining doesn’t give you a good answer, you are wasting your time. Start over and focus on the value you are going to create for your customers. Treasury 3.0.