Artificial intelligence and machine learning promise nothing short of a customer experience transformation in retail banking — but the best ideas for their application could fail if projects are not built on firm foundations of clean, accurate and complete data around customers and their behaviors and financial needs.
The focus on customer experience among today’s retail banks and credit unions should hardly be surprising. With financial services becoming even more commoditized, institutions increasingly must battle for consumers’ attention and wallet share against disruptive new market entrants as well as their traditional competitors. More than ever, financial institutions need to find some way to differentiate themselves.
In customer experience terms, AI and machine learning can help marketers in retail banking to predict client needs and deepen relationships. They can do this through more personalizing their approaches, fine-tuning campaigns for maximum effectiveness, targeting consumer segments that represent the best acquisition prospects, and identifying attrition risks and causes.
Building on an AI Base that’s Already Coming Together
Retail banks already rank among the most progressive organizations in deploying machine learning, according to a survey of 1,419 companies, including more than 150 from the financial services sector, conducted by MIT Technology Review Insights in association with Google.
“AI and machine learning can help marketers in retail banking to predict client needs and deepen relationships.”
The survey found that more than four out of ten (41%) of financial services marketers currently use machine learning, and another 30% plan to deploy the technology this year. Two-thirds of respondents (66%), meanwhile, agree that machine learning is fueling their strategic marketing efforts. The technology enables them to sift through reams of data to identify which strategies work best with specific geographic and demographic customer segments, as well as to predict future industry trends and customer buying habits.
But if these new technologies are to deliver meaningful insights, financial institutions cannot sidestep the effort that must first go into laying the groundwork. That requires putting their existing data in order, as well as identifying and incorporating new sources of third-party information that could help, such as geographic and socioeconomic data.
AI and machine learning, after all, are notoriously data-hungry processes. Regardless of the sophistication of the algorithms underpinning them, the answers they return can only ever be as smart as the information that feeds them. That makes data management a vital, prerequisite step on the journey to delivering better customer experiences.
With that in mind, a six-step, upfront effort is the best way for retail banks and credit unions to ensure that their explorations of AI and machine learning deliver a good return on the time and money invested in them.
1. Identify and Agree on Business Objectives and How to Measure Them
All AI projects should begin with a clear goal. No investment should be made until this is settled. The goal at a retail financial institution might be to increase responsiveness to digital marketing campaigns by X%, to increase mortgage originations by X%, or to increase wallet share per household. Only once representatives from the business and technology leaders have agreed on reachable overall target can any necessary investments be couched in the context of the expected returns from achieving that goal.
2. Build Your Business Case and Plan the Data Element
With a clearer idea of expected returns and expected costs, retail banks can then build a clear business case.
Here, Net Income Per Employee can be a useful metric. In other words, what is the revenue contribution if we achieve our objective, minus the cost of necessary investments, plus the productivity gain from making that investment?
Every business function — from commercial loan origination to new customer onboarding — requires timely access to the right data. Poor data quality, if left unfixed, can significantly impact a retail bank’s gross revenue, thanks to losses in productivity and missed revenue opportunities.
Read More:
- How Banks and Credit Unions Can Grow Deposits with AI
- AI Advantages in Banking Grow, Adding Pressure for Broad Adoption
- How Two Banks Are Chasing The AI Dream
3. Define Your Machine Learning Application’s Data Requirements
Bring together all relevant parties from across the bank or credit union determine what data will be critical to achieve success. Here, the input of employees involved in the day-to-day work of orchestrating marketing campaigns and onboarding new customers, for example, will be essential in outlining the information they need to do this work effectively. While the technology team will likely have a good understanding of the tools that are needed to manage that data, they can’t be expected to second-guess what that data looks like.
4. Conduct a ‘Current State’ Data Assessment
The task here is to identify what data the financial institution currently holds, where it comes from, its quality and availability, how well it’s managed, and what’s currently missing.
This should be a formal exercise, in which employees are asked to rank data on a scale of one to five, where one denotes “poor” and five denotes “good.” It’s often wise to bring in a third party to conduct this assessment because outside experts can bring an objective eye to the proceedings, operating above the subjective assessments that arise internally.
5. Define a Plan to Address Data Gaps
Based on the outcomes of a data assessment, it’s time to assess what investments are required to address the gaps. What will it take for employees to be able to access and share accurate, consistent and holistic views of their customers — and how might data management technology assist in achieving that goal?
At this point, retail banking leaders should be thinking beyond the confines of the project immediately to hand, to consider how these tools might be leveraged elsewhere in the organization and on other projects. Increasingly, retail financial institutions are looking to establish shared services for data that can scale to meet a wide range of business requirements. In this way, they’re not building a new data management environment for every requirement that comes up.
Read More:
- How BankMobile Uses Artificial Intelligence to Streamline the Experience
- Data and AI Power the Future of Customer Engagement in Financial Services
- Banking Brands Need AI to Keep Pace With Digital Giants
- Artificial Intelligence: The Financial Marketer’s Secret Weapon
6. Execute, Monitor, Refine and Report
The deployment of AI and machine learning calls for an agile approach, based on ongoing monitoring of business results and regular tweaks to ensure that business value is achieved. AI is an emerging technology and, with that in mind, it makes no sense to attempt to “boil the ocean” on the first attempt. Instead, it’s about incrementally leveraging what’s working, fixing problems, and adjusting course where necessary in order to move forwards in pursuit of overall goals.
Like any emerging technology, AI comes with some risks. However, the potential rewards are too great for retail banks and credit unions to ignore. For their applications of AI to work, they need to consider the strategic value of their data. By tying improvements in data management to business outcomes measured in hard dollars, they can simultaneously mitigate the risks associated with poor data quality, while reaping the rewards of new approaches that deliver a step change in customer engagement and experience.