One-to-One at Scale: The Confluence of Behavioral Science and Technology and How It’s Changing Business

Consumer and business customers have increasing expectations that businesses provide products and services customized for their unique needs. Adaptive intelligence and machine learning technology, combined with insights into behavior, make this customization possible. The financial services industry is moving aggressively to take advantage of these new capabilities. In March 2018, Bank of America launched Erica, a virtual personal assistant—a chatbot—powered by AI. In just three months, Erica surpassed one million users.

But to achieve personalization at scale requires an IT infrastructure that can handle huge amounts of data and process it in real time. Engineered systems purpose-built for these cognitive workloads provide the foundation that helps make this one-to-one personalization possible.

Bradley Leimer, Managing Director and Head of Fintech Strategy at Explorer Advisory & Capital, provides consulting and investment advisory services to start-ups, accelerators, and established financial services companies. As the former Head of Innovation and Fintech Strategy at Santander U.S., his team connected the bank to the fintech ecosystem. Bradley spoke with us recently about how behavioral science is evolving in the financial services industry and how new technological capabilities, when tied to human behavior, are changing the way organizations respond to customer needs.

I know you’re fascinated by behavioral science. How does it frame what you do in the financial sector?

Behavioral science is fascinating because the study of human behavior itself is so intriguing. One of the many books I was influenced by early in my career was Paco Underhill’s 1999 book Why We Buy. The science around purchase behavior and how companies leverage our behavior to create buying decisions that fall in their favor—down to where products are placed and the colors that are used to attract the eye—these are techniques that have been used since before the Mad Men era of advertising.

I’m intrigued by the psychology behind the decisions we make. People are a massive puzzle to solve at scale. Humans are known to be irrational, but they are irrational in predictable ways. Leveraging behavioral science, along with things like design thinking and human-computer interaction, have been a part of building products and customer experiences in financial services for some time. To nudge customers to sign up for a service or take an additional product or to perform behaviors that are sometimes painful like budgeting, saving more, investing, consolidating, or optimizing the use of credit all involve deeply understanding human behavior.

Student debt reached $1.5 trillion in Q1 2018. Can behavioral analytics be used to help students better manage their personal finances?

What’s driving this intersection between behavioral science and fintech?

Companies have been using the ideas of behavioral science in strategic planning and marketing for some time, but it’s only been in the last decade that the technology to act upon the massive amount of new data we collect has been available. The type of data we used to struggle to plug into a mainframe through data reels now flies freely within a cloud of shared service layers. So beyond new analytic tools and AI, there are few other things that are important.

People interact with brands differently now. To become a customer now in financial services, it most often means that you’re interacting through an app, or a website, not in any physical form. It’s not necessarily how a branch is laid out anymore; it’s how the navigation works in your application, and what you can do in how few steps, how quickly you can onboard. This is what is really driving the future of revenue opportunity in the financial space.

At the same time, the competition for customers is increasing. Investments in the behavioral science area are a must-have now because the competition gets smarter every day and the applications to understand human behavior are simultaneously getting more accessible. We use behavioral science to understand and refine our precious opportunities to build empathy and relationships. 

You’ve mentioned the evolution of behavioral science in the financial services industry. How is it evolving and what’s the impact?

Behavioral science is nothing without the right type of pertinent, clean data. We have entered the era of engagement banking: a marketing, sales, and service model that deploys technology to achieve customer intimacy at scale. But humans are not just 1’s and 0’s. You need a variety of teams within banks and fintechs to leverage data in the right way, to make sure it addresses real human needs.

The real impact of these new tools has only started to be really felt. We have an opportunity to broaden the global use of financial services to reduce the number of the underbanked, to open new markets for payments and credit, to optimize every unit of currency for our customers more fully and lift up a generation by ending poverty and reducing wealth inequality.

40% of Americans could not come up with $400 for an emergency expense. Behavioral science can help move people out of poverty and reduce wealth inequality.

How does artificial intelligence facilitate this evolution?

Financial institutions are challenged with innovating a century-old service model, and the addition of advanced analytics, artificial intelligence tools and how they can be used within the enterprise is still a work in progress. Our metamorphosis has been slowed by the dual weight of digital transformation and the broader implications of ever-evolving customers.

Banks have vast amounts of unstructured and disparate data throughout their complicated, mostly legacy, systems. We used to see static data modeling efforts based on hundreds of inputs. That’s transitioned to an infinitely more complex set of thousands of variables. In response, we are developing and deploying applications that make use of machine learning, deep learning, pattern recognition, and natural language processing among other functionalities.

Using AI applications, we have seen efficiency gains in customer onboarding/know-your-customer (KYC), automation of credit decisioning and fraud detection, personalized and contextual messaging, supply-chain improvements, infinitely tailored product development, and more effective communication strategies based on real-time, multivariate data. AI is critical to improving the entire lifecycle of the customer experience.

What’s the role of behavioral analytics in this trend?

Behavioral analytics combines specific user data: transaction histories, where people shop, how they manage their spending and savings habits, the use of credit, historical trends in balances, how they use digital applications, how often they use different channels like ATMs and branches, along with technology usage data like navigation path, clicks, social media interactions, and responsiveness to marketing. It takes a more holistic and human view of data, connecting individual data points to tell us not only what is happening, but also how and why it is happening.

You’ve built out these customization and personalization capabilities in banks and fintechs. Tell us about the basic steps any enterprise can take to build these capabilities.

As an organization, you need to clearly define your business goals. What are the metrics you want to improve? Is it faster onboarding, lower cost of acquisition, quicker turn toward profitable products, etc.? And how can a more customer-centric, personalized experience assist those goals?

As you develop these, make sure you understand who needs to be in the room. Many banks don’t have a true data science team, or they are a sort of hybrid analytical marketing team that has many masters. That’s a mistake. You need deep understanding of advanced analytics to derive the most efficiencies out of these projects. Then you need a strong collaborative team that includes marketing, digital banking, customer experience, and representation from those teams that interacts with clients. Truly user-centric teams leverage data to create a complete understanding of their users’ challenges. They develop insight into what features their customers use and what they don’t and build knowledge of how customers get the most value out of their products. And then they continually iterate and adjust.

You also need to look at your partnerships, including those with fintechs. There are several lessons derived from fintech platforms such as attention to growth through business model flexibility, devotion to speed-to-market, and a focus on creating new forms of customer value through leveraging these tools to customize everything from onboarding to the new user experience as well as how they communicate and customize the relationship over time.

What would be the optimum technology stack to support real-time contextual messages, products, or services?

Choosing the right technology stack for behavioral analytics is not that different than for any other type of application. You have to find the solution that maps most economically and efficiently to your particular problem set. This means implementing a technology that can solve the core business problems, can be maintained and supported efficiently, and minimizes your total cost of ownership.

In banking, it has to reduce risk while maximizing your opportunities for success. The legacy systems that many banks still deploy were built on relational databases and not designed for real-time processing, providing access via Restful APIs and the cloud-based data lakes we see today. Nor did they have the ability to connect and analyze any form of data. The types of data we now have to consider is just breathtaking and growing daily. In choosing technology partners, you want to make sure what you’re buying is built for this new world from the beginning, that the platform is flexible. You have to be able to migrate between on-premises solutions to the cloud, along with a variety of virtual machines being used today.

If I can paraphrase what you’re saying, it’s that financial services companies need a big data solution to manage all these streams of structured and unstructured data coming in from AI/ML, and other advanced applications. Additionally, a big data solution that simplifies deployment by offering identical functionality on-premises, in the cloud, and in the Oracle public Cloud behind your firewall would also be a big plus.

Are there any other must-haves in terms of performance, analytics, etc., to build an effective AI-based solution?

Must-haves include flexibility to consume all types of data, especially that which is gathered from the web and from digital applications. It needs to be very good at data aggregation—that is, reducing large data sets down to more manageable proportions that are still representative. It must be good at transitioning from aggregation to the detail level and back to optimize different analytical tools. It should be strong in quickly identifying cardinality—how many types of variables can there be within a given field.

Some other things to look for in a supporting infrastructure are direct access through query tools (SQL), support for data transformation within the platform (ETL and ELT tools), flexible data model or unstructured access to all data, algorithmic data transformation, ability to add and access one-off data sets simply (like through ODBC), flexible ways to use APIs to load and extract information, that kind of thing. A good system needs to be real time to help customers in taking the most optimized journey within digital applications. 

To wrap up our discussion, what tips would you give the enterprise IT chief about how to incorporate these new AI capabilities to help the organization reach its goals around delivering a better customer experience?

First, realize that this isn’t just a technology problem—it will require engineers, data scientists, system architects, and data specialists sure, but you also need a collaborative team that involves many parts of the business and builds tools that are accessible.

Start with simple KPIs to improve. Reducing the cost of acquisition or improving onboarding workflows, improving release time for customer-facing applications, reducing particular types of unnecessary customer churn—these are good places to start. They improve efficiencies and impact the bottom line. They help build the case around necessary new technology spend and create momentum.

Understand that the future of the financial services model is all about the customer—understanding their needs and helping the business meet them. Our greatest source of innovation is, in the end, our empathy.

You’ve given us a lot to think about, Bradley. Based on our discussion, it seems that the world of financial services is changing and banks today will require an effective AI-based solution that leverages behavioral science and personalization capabilities.

In order for banks to sustain a competitive advantage and lead in the market, they must invest in an effective big data warehousing strategy. Therefore, business and IT leaders need a solution that can store, acquire, process large data workloads at scale, and has cognitive workload capabilities to give you the advanced insights needed to run your business most effectively. It is important that the technology is tailor-made for advancing businesses’ analytical capabilities that leverage familiar big data and analytics open source tools. And Oracle Big Data Appliance provides that high-performance, cloud-ready secure platform for running diverse workloads using Hadoop, Spark, and NoSQL systems.