How Banks Can Make AI-Based Insights Work

How Banks Can Make AI-Based Insights Work

Here are four challenges banks need to overcome to start generating AI-based insights that improve the customer experience.

Many banks recognize the potential for using AI-based insights to personalize offerings and offer more relevant services. However, most banks’ culture present challenges in terms of siloed systems emphasizing business segments rather than customers. This effectively blocks the bank from deploying AI across all customer touchpoints, from online and mobile sites to branches and ATMs.

Download Report: AI in Financial Services

By adopting a top-down focus on customer experience (CX) and committing to a 360-degree view of customers, banks are able to implement AI and analytics across all its points of interaction. One specific bank that Cognizant worked with, found that by implementing AI, its personalized offers reached 97% of customers, and the revenue from 200 targeted campaigns soared 160% from the year prior.

Four Hurdles to Overcome
Here are four changes banks need to make if they want to improve the customer journey with AI-based insights:

1. Consolidate data across channels. While financial services firms have traditionally been great at collecting information, today’s data comes in many formats and types. Not only has data volume exploded; most of it no longer fits neatly into rows and columns and database joins.

But to get the most out of AI, it’s absolutely necessary to consolidate this data. This is true whether you’re applying it to the customer experience or improving fraud detection, which is also an important factor in boosting customer confidence and loyalty.

Doing so, however, requires new thinking about who can access what data. Rather than keeping data closely guarded, with strict permissions policies, centralized data requires a more democratic, enterprise-wide approach, balanced by new security and governance controls such as tokenization, user authentication and role-based access.

We partnered with a credit-card and mortgage company that’s a case in point. It centralized its data and moved to an AI-based system not only to better understand customers but also to assist with compliance and accelerate decision making. The upshot was substantial: With the system’s ability to do real-time prospect analysis and make intelligent recommendations, conversion rates jumped from 8% to 21%.

2. Execute real-time campaigns. Once you know who your customers are, you need to be able to interact with them in real-time. This includes extending relevant offers at the right moment, quickly determining the success of these offers and adeptly course-correcting if needed.

Real-time response is also essential when it comes to credit-card activity and especially declined transactions, which are a nuisance for cardholders and a huge dollar loss for card providers. We partnered with a financial services provider that wanted to use AI-based insights to more quickly determine the root cause of credit card declines. Our team analyzed the key features affecting acceptance and then developed multiple machine-learning models to predict merchants with the highest probability of card declines. In addition to reducing volume loss, the card provider was able to identify merchants whose point of sale systems weren’t able to accept cards issued by its alliance partners and planned site visits for POS configuration.

3. Transition to a customer-centric culture. To reap the game-changing benefits of AI-powered customer journeys, businesses need to distribute the insights it generates throughout the enterprise, including employees who aren’t customer-facing, from software developers to the folks who work in the supply chain.

This requires banks to embed a customer-centric mindset into the banks’ everyday environment. In our recent report on the factors that define digital maturity and business success (registration required), a key difference between leaders and laggards was the ability to align new technologies with human requirements, such as improving consumer and employee experiences.

To nurture this culture, banks should make customer centricity a metric, and tie compensation to it through cash incentives or bonuses. For example, bonuses could be awarded to teams responsible for creating an app that improves customer interaction. Some businesses require all employees to spend a day in the call center to learn more about customer needs in a personal way.

4. Establish the right level of convenience. From Gen Z-ers to baby boomers, customers of every age want their experiences to be frictionless. Yet convenience is relative. Using AI, banks can assess how each customer defines ease of use and personalize interactions to those preferences. The experience of a large U.S. bank illustrates how AI-based insights can make such hyper-personalization possible. The bank was curious about behavioral patterns that predict customers at risk for closing their account. Our team created AI-driven models to sift through 10GB of data from three sources spanning 2,141 variables and more than 68,000 households.

After analyzing the data of customers who actually closed their accounts, the system turned up clear harbingers of churn. For example, in the three months prior to closure, online transactions slowed 9X, and ATM and gasoline transactions dropped by 4X. Seventeen percent of churned customers closed a savings account within the same window of time. By noting the patterns, the bank was able to create an intervention plan – and reduce churn by 15%.

To fulfill its potential, AI needs to be applied at every point in the banking customer journey – not just some of them. It’s not called a customer “journey” for nothing: Customer interactions are not just a sequence of steps, and the journey is never really over. There’s always more to do. Now is the time to begin tackling the hard stuff.

Related Publication

Survey: The Status of Automation in Financial Services

Process automation is vital to banking’s future; 90% of the 302 industry leaders in our study appear convinced that process automation is either important or critical to their business. Yet many are struggling to move beyond early proofs of concept. Why is that?