Artificial intelligence in finance: Predicting customer actions

Artificial intelligence can give you a valuable roadmap for your customers’ financial portfolio. Learn why predictive analytics is changing how bankers do business.

By David Berglund, senior vice president and artificial intelligence lead, U.S. Bank Innovation
Tags: Innovation, Artificial intelligence, AI
Published: July 11, 2018

What does artificial intelligence (AI) mean for the financial industry? We’re only starting to find out, but the potential grows every day. At the consumer level, chatbots and virtual assistants help guide customers without direct human involvement. At the corporate level, businesses continue to pursue automated processing tools that can eliminate excess busywork and identify security risks before they happen.

One of the most compelling arguments for artificial intelligence (and the variety of terms that often define it) lies with prediction. The ability to predict future actions can help corporations position themselves to meet expectations earlier in the customer relationship.

Predictive analytics is the catchall term for technology that leverages data to make predictions about future events. It’s a subset of the broader data science practice that is driven by AI technology. To be clear, it is not perfect – and it doesn’t replace relationships forged through interpersonal communication.

However, it can help corporate organizations keep up with the ever-changing needs of its customers. By analyzing past behaviors and patterns, AI-driven predictive tools can help businesses tailor services to specific customer needs.

In my work with the U.S. Bank innovation team, I’ve seen firsthand how these predictive engines can reshape how we all do business. But as U.S. Bank Chief Innovation Office Dominic Venturo mentioned in a recent interview, bankers shouldn’t fear bots.

Here are some areas where predictive engines are already reshaping the commercial and corporate sectors.


Customer relationships and the overall experience

Predictive analytics can affect every part of the customer experience, from the initial account opening to ongoing relationship management. A recent Infosys study noted that predictive analytics could help identify upcoming customer demands, allowing organizations to provide better-targeted offerings and more proactive recommendations.

Here’s a practical example from the banking sector, based on the use of conversational banking with virtual assistants:

Virtual assistants, like Apple’s Siri or Amazon’s Echo, use a combination of machine learning and natural language processing to provide content on a user’s request. Whenever you speak a command to one of those devices, it references any past interactions to respond appropriately.

From a financial services standpoint, this could mean anything from automated customer support to recommendations on new products or services. Even large corporates can join in on the virtual action – Samsung recently announced that they intend to pursue a “digital identity” platform with virtual assistants and blockchain ledgers for their customers.

The shift to a conversational interface is built on top of prior trends for push to mobile and social channels, then to messaging. With advancements in AI, machine learning and natural language processing, it’s simply possible to do much more through guided or unstructured dialog than before. As technology and user experience expectations evolve, we expect that we’ll be engaging more and more with bots and smart devices in this way.

Marketing and sales opportunities

From the business point of view, predictive analytics can offer a path to better marketing and sales goals. The more you know about customers and their needs, the better you can serve up relevant solutions. Data scientists can analyze a customer’s past behaviors with marketing offers, and the predictive engine can assess a score regarding how likely the customer would be to accept a new offer. This lead scoring process, which aims to simplify the identification of qualified leads, can help your sales teams focus on customers that are more accessible.

Here’s how that process might look for a corporate marketing and sales team:

You already receive recommendations for future content when browsing Google or Amazon. When it comes to marketing offers from a bank, the AI platform culls through your entire history of interactions and predicts whether you’ll accept a shiny new offer.

This serves to eliminate much of the uncertainty in marketing and sales prospecting. It’s less of a gut feeling, and it provides executives with better assurances about future performance. Rather than reacting to previous behaviors, the business can now make proactive changes.

Fraud prevention and regulatory compliance

The goal of any artificial intelligence interface is to mimic human reasoning and processing as best as possible. From there, the predictive engine can assess what might occur next, while a prescriptive engine would recommend steps to reach (or avoid) the current course.

Fraud prevention and regulatory compliance rules come into play with this process. Armed with data on past attacks, companies with predictive engines can determine the likelihood of another attack, prepare for it, and avoid or deter it.

Risk management is a common goal for predictive engines, because of the ability to scan for potential threats and mitigate future breaches. The use of data to determine future actions is nothing new, but with the onset of artificial platforms, data processing is faster and more accurate.

Here’s how a company could use a predictive-prescriptive analytics process for better risk management:

A predictive engine would enhance many existing fraud prevention tools – like stress testing and compliance tracking – by providing a real-time database of risks by geography, physical location (internal and external), service or product, and timeframe. The engine would use this data to make informed decisions on future threats, allowing the business to prepare and/or mitigate the danger.

What separates this process from earlier fraud prevention models is a higher level of machine cognition. Computers are getting smarter, and can adapt their rules as humans provide new insights. This can reduce the amount of false positives, and save the risk management team from wasting resources on bad leads.


Automation and prediction: The tangible goals of AI

When we talk about AI at U.S. Bank, we find that people aren’t necessarily sure what to do with it. All these concepts sound great, but how could they best contribute to a business infrastructure and process?

There’s no singular answer, of course. But the ability to automate data collection and predict future customer behaviors can dramatically affect the relationship, marketing, sales and security branches of your business.


David serves as senior vice president and artificial intelligence lead on the U.S. Bank Innovation team. He focuses on driving technology innovation and product development, accelerating business growth, building strong team cultures and finding exponential advantages with technology.