Five years ago Bank of America only sold 10 different financial products. How hard could it be to intelligently target customers, especially before the proliferation of mobile and social channels?
Deceptively hard, it turns out. It was during this period when Bank of America first engaged with SAS to build out an intelligent process to send the right message to the right customer at the right time in the right channel.
“We expected we’d be running 500 billion decision variables,” Andres De Armas, the financial institution’s SVP of customer offers and targeting, said during his presentation at the SAS Premiere Business Leadership Series in Orlando.
This is because the financial behemoth had 10 products multiplied by 10 channels multiplied over the numerous types of interactions it might potentially have with different customers. The recent addition of digital and online elements to the system increased those decision variables four-fold.
Numerous elements factored into the complexity of these variables. Notably, Bank of America’s substantial data load: credit bureau data, statistical models, customer relationship data, segmentation and behavioral data, channel usage data, contact and offer data, customer response data, and third-party data—all of which influences the way Bank of America approaches each customer.
“We’re trying to figure out ways to maximize our relationship impact and, of course, the efficiency of our relationships when it comes to the depth and the profitability we’re getting from them,” De Armas said.
And the scoring system that Bank of America uses to determine the value of each relationship is also complex. While profitability from the purchase of a certain financial product is straightforward, Bank of America also scores service messages. Enrolling someone in overdraft protection, for instance, improves customer retention.
“So it’s not just the actual dollars of selling something and collecting money for that something,” De Armas explained. “It’s the longevity and the value of the relationship that goes into the SVA, Sales Value Added.”
All things considered, the technology build-out was the easy part of the SAS analytics installation. The hard part was creating business plans around all that new technology, managing the people who’d be using it, and updating the processes that would be affected by it.
When Bank of America began this project it executed channel optimization quarterly. Soon it became monthly. Today it’s weekly. This means Bank of America, which once got 90-day-old customer data, now has data that’s much fresher, which creates some new considerations in how to use it and how it will affect business processes.
“There were a few things we didn’t put price tags on that ended up being material,” De Armas said. “Either in terms of money or in terms of getting to milestones we thought we’d get to sooner.”
Here are De Armas’ lessons:
“Imagine what it’s like to lose subject matter expertise when you’re in the middle of a journey like this,” De Armas said.
Without commitment, the best people will leave the project—the very people who are the bridge between the slide deck that communicates the project to the financial committee and the actual execution of the work. These are the people who will keep everything on course.
“In this day and age of limited attention span, you sometimes have to reeducate the very same people,” De Armas explained.
The people who matter, De Armas said, weren’t just those who wrote the checks. They bought in rather easily. Others did not: marketers who didn’t want to change how they’d been performing their jobs, or channel leadership teams that dictated what went into their respective communication channels.
“Downstream of that centralized optimization, there are a number of places where the folks who have been doing things a certain way for a long time act as inhibitors,” De Armas explained. He didn’t mean this as a pejorative. “They feel they’re the best at optimizing the offers and conversations going through their channels. And that’s a true thing.”
De Armas and his team coaxed some of the more reluctant stakeholders to give ground. This meant he couldn’t curtly tell them that they were doing things incorrectly, or that they needed to change their job description.
Instead, he asked channel leadership if he could keep 25% of their population. They readily agreed. “And almost to an activity, when you look at the portfolio and the performance at the customer level for that channel, we do really, really well when we optimize,” he said.
He’d produced a tangible proof point rather than a confrontation.
“This is the most exasperating for us,” said De Armas, “not just because it’s not solved but it’s always been that push-pull in the battle of what defines success.”
The issue revolves around units versus sales. Typically, businesses assess sales staff measured by the number of products they sell. Management, however, looks at profit retention and sales value. This creates success metrics that don’t always align.
A credit card has a useful lifespan south of seven years, De Armas pointed out. “If I tell a business partnering card that this particular sale isn’t worth the trouble, that we won’t make the value we’re seeking and we’re better off engaging the customer with something else, it’s not uncommon to hear that the profit doesn’t come in for seven years.”
But how can Bank of America be sure that the expected value is what it will actually be seven years down the line? One issue is that Bank of American doesn’t have hard unit costs associated with every product, which creates that debate around the value Bank of America might derive from each sale.
Moreover, there are costs to the marketing messages that go beyond, say, the fraction-of-a-penny for each email message.
“One of the costs we’ve discovered in a number of our activities is if you show customers a certain thing in a certain way for a certain amount of time, they start to tune out,” De Armas said. He equated it to staring at the same meaningless ads on the same website every day. There comes a point in time when it’s best for Bank of America to simply walk away from an interaction because if it doesn’t, all the intelligent selection and sequencing won’t stop messages from becoming spam.
“And there comes a point when tuning out costs me so much more than losing that opportunity,” De Armas said. It costs the ability to continue effectively building a relationship with a customer.
People become enamored with technology so much that they keep building and never actually turn it on. But using the technology—even to alleviate smaller issues—provides a lot of value. First, it solves a problem. Second, it creates a proof point that can woo fence-sitters.
Earlier this year Bank of America had a service issue: a failure in the authentication process. The problem was that a tremendous amount of customer records were wrong—either incomplete or incorrect.
“We had 100 people working on this thing, kicking through how to do it, changing the script on the phone,” De Armas recalled, “And a member of our team quietly said, ‘Why not use a targeted offer.’”
Bank of America put one up at the online sign-in: a personalized message for customers with incomplete records, prompting them to fill out their phone number or their date of birth—whatever information was missing.
The bank fixed 500,000 customer records in a week.
“The biggest credibility builders are [fixing] the little things that have been a sore in everybody’s side; nobody fixes it but everybody knows it’s there,” De Armas said. “In an environment where the decision makers are changing and you have to show value, focus on showing what you can do at least as much as focus on what you want to do.” De Armas said, and then pointed toward the audience. “I would venture to guess that you have more technology than you can use today and I guarantee you have more people trying to get the next piece of technology in place than those trying to figure out what you can do with the stuff you have now.”
Many of Bank of America’s struggles came from its background in direct marketing, during which it successfully ran campaign-based programs by product. The process was straightforward: find the people eligible for a certain financial product, figure out who can be acquired at an acceptable rate, and have the sales team fire away.
But old habits die hard. “We still catch ourselves targeting the customer at this level before we give optimization a chance to make that decision,” De Armas said. Doing this undermines the optimization effort. It fails to take into account that a prospect who might initially seem like a good fit for a financial product is actually in a segment that, based on past behaviors or historical behavior, is highly unlikely to take action.
This is why De Armas wants to know who within Bank of America is accessing the system. “If you don’t know what’s going on with your inputs, somebody could be driving you in a different direction, or at least not the most optimal direction,” he said. “We’ve spent a very good amount of time in 2013 really understanding who has access [into the data environment]. And we need to document them.”
