Is Predictive Analytics For Everyone? Or Just Megabanks?

Many believe predictive analytics will give some banks a big edge by identifying patterns well before their competitors. But is this power only going to be available to a privileged few? Or will the potential of artificial intelligence and advanced statistical analysis be something that any financial institution can leverage — even community banks and credit unions?

As banking evolves to meet consumer demands, financial marketers face mounting pressure to understand what works and where they need to improve. Consumers expect convenience and simple solutions. They want an immediate, personalized experience. These are today’s table stakes, not “value-added options.” But it’s virtually impossible to tailor any kind of experience — much less in real time — if you don’t know who you’re talking to.

Many in the banking industry believe the answer to these problems can be found in advanced statistical analytics — what is often called “artificial intelligence.”

One of those is Steven Ramirez, CEO of Beyond the Arc.

“For financial marketers who are often asked to do more with less, predictive analytics can be the key to unlocking greater ROI,” says Ramirez. “The trick is how to interpret all the data that’s available so you can answer strategic critical questions. Who are the most likely prospects? How do you identify which customers are at the greatest risk of leaving? What offers work best?”

With predictive analytic technologies, you can identify patterns in data that were previously hidden (or extremely difficult to spot using more traditional methods). Based on these statistical correlations, you can predict the likelihood of future outcomes. In the past, financial marketers may have estimated forecasts for broad customer segments like “Millennials” or “Single Moms.” Now, Ramirez says bankers can predict the behavior of each and every individual, and it can be done for a database with millions of people.

What makes this possible is leveraging a broad range of customer data:

  • Interactions – Email and chat transcripts, call center notes, web click-streams, and in-person dialogues.
  • Attitudes – Opinions, preferences, needs, and desires gathered through survey results and social media.
  • Descriptions – Attributes, characteristics, self-declared information, and demographics.
  • Behaviors — Orders, transactions, payment history, and usage history and more.

If, for example, the goal is retention, analysis usually includes consumer behavior like calls to the contact center or visits to a branch. Analysis to support marketing offers might also include credit scores, social media, transactions, census data, and more.

According to Ramirez, the asset size of the institution doesn’t matter. While advanced analytics once required complex IT systems, expensive software and PhD-level statisticians, today there are plenty of low-cost and open source tools that can empower a bank or credit union with predictive engines.

“You don’t need an enterprise data warehouse, nor millions of rows of data,” Ramirez explains. “A propensity model to predict who is most likely to accept a home equity offer can use data in a spreadsheet, and the model can be constructed and run on a laptop.”

However, Ramirez notes that the more sophisticated the modeling desired, the greater the requirement for a complete view of the customer — and that will need to extend across all of the institution’s silos.

Ramirez says that too many financial institutions are still doing very little analytics, instead resorting back to the good old “spray and pray” approach. The result is that consumers are dissatisfied, because too many people receive too many offers that simply aren’t relevant.

“When marketers have done analytics in the past, they were usually looking at historical information. They were looking in the rear view mirror instead of trying to predict consumer behavior.”

Ramirez says the initial focus for any financial institution building predictive models should be to answer some basic, fundamental questions:

  • Over time, who will be our best customers? (Predicting the lifetime value of customer relationships.)
  • Which customer experiences will have the greatest impact on customer loyalty? (Understanding the factors behind likeliness to recommend.)
  • If we offered a new product or service, which of our current customers would be most likely to buy it? (Predicting cross-sell, upsell, and next best product.)

“Consumers wonder, ‘How does my bank help me make money, save money, or save time?'” Ramirez says. “Predictive analytics helps to fulfill the promise of one-to-one marketing — the idea that marketing can become personalized to meet the needs of each and every customer.”

Taking the Guesswork Out of Acquisition and Retention Efforts

Where predictive analytics particularly shines is when it helps provide insights into how likely behavior is to change as incentives are packaged and presented in various ways.

“The effectiveness of offers can be very complex,” Ramirez explains. “That is why we need advanced algorithms and computing power — in order to identify the right mix of factors for each customer. An incentive alone isn’t typically enough to make a customer stick around, so there has to be an overarching, long-term value proposition.”

Ramirez adds that community banks and credit unions can use predictive analytics to help understand which factors are likely to drive a particular person away. For example, with retention modeling, a ‘retention score’ can be calculated for each customer. The bank can deploy this score to the front line so that tellers and other service reps are flagged with a “red alert” whenever someone is at risk of closing their accounts, in which case staff may choose to proactively extend a special retention offer.

Ramirez explained that predicting customer behavior depends on a multitude of factors: tenure with the institution, product holdings, size and trends of balances, and much, much more. On top of that are experiences the customer may have recently had, both positive and negative.

“Thankfully we can now make sense of this mind-bending array of variables, and find the patterns,” Ramirez says. “Predictive modeling may tell a bank marketer that certain high-value customers who have had a problem with a wire transfer over $20,000 might have the greatest risk of leaving in the next 30 days. That’s pretty exciting.”

He cautioned that finding the pattern is only the first step. It is up to marketing and customer service representatives to act on those insights.

You can learn more about how data analytics and artificial intelligence technologies can give you and edge on your competition during Steven Ramirez’s session, “How Predictive Analytics Can Deepen Customer Relationships,” at The Financial Brand Forum 2017. You’ll learn how to build algorithmic models that strengthen relationships, maximize revenues and reduce customer attrition by translating your raw data — customer behaviors, transaction patterns, buying habits, late payments, unusual activity, social media interactions, etc. — into actionable marketing strategies.

This article was originally published on . All content © 2024 by The Financial Brand and may not be reproduced by any means without permission.