How Machine Learning for Credit Unions Can Power Growth While Minimizing Risk
For credit unions, 2018 has been a big year for expansion and innovation, which also makes it a big year for facing the new threats that are sure to follow. The financial space is evolving, and credit unions must evolve too. So, how does a credit union like yours scale smart? It starts by looking at the macro trends that are currently impacting customers, and discovering ways to leverage these trends for long-term growth.
Credit unions are growing
Credit unions are growing in many different ways this year, and the upward trends can provide a smart starting point for understanding your potential. Gains were strong across existing and new product lines and portfolios, as highlighted in the most recent performance data from the National Credit Union Administration.
Here are some of the highlights from the report:
- Total assets rose 5.9% year-over-year to reach $1.42 trillion.
- Credit union systems’ net worth ratio was 10.88%, compared to 10.69% the year before
- Insured shares and deposits are up 5.5%.
- Delinquency rate of federally insured credit unions declined slightly in the quarter to 66 basis points.
- Return on average assets reached 90 basis points year-over-year, compared to 71 basis points the year before.
This snapshot shows that credit unions are performing well, which means a greater chance of expanding into new markets, and also greater risks of becoming a target for fraud.
Understanding new members, and new risks
Credit union members, and those in the financial sector at large, are vocal about their desire for new digital services. This is easy to see in the popularity of fintech firms such as the budget tool Mint, mobile money sending service Venmo, or the bevy of apps that allow us to pay for a cup of coffee with our smartphones.
Mobile is the dominant theme. Juniper Research projects that 1.75 billion people will use their mobile phones for banking by 2019, up from 800 million in 2014. That’s a lot of potential members, but also a significant amount of data that could be at risk as every account and transaction heads online.
Business members face the same desires and accompanying threats. At the 2018 NACHA conference, 74% of survey respondents said that mobile apps for business banking would improve a company’s experience with financial institutions.
The good news for you is that members will trust you to get it right when it comes to protecting their data. Credit union service providers are the most trusted brands when it comes to data security, though you must work to maintain that trust with your members.
After you have the products and services in place to meet existing demand, it will be time to start looking for new businesses opportunities to scale.
There’s a lot of data available, though it can be difficult to understand. It’s especially hard not to chase the shiny object. For example, fintech tools grab a lot of attention and seem like a major demand. But, this might not be true for your particular market, given that 59% of U.S. consumers do not use any fintech tools.
This could mean that members want such services but need a trusted partner to adopt them. Analysis of your audience is important to understand what any particular trend means for you. For example, Frank Rinaudo, SVP of GrooveCar Inc., says that 30% of new car sales nationwide are leases, but that this increases to 70% in urban centers. Adopting a more aggressive leasing product portfolio could ensure that you’re competing for every car sale in your region.
Finding these patterns and understanding what’s relevant to you will require scaling the data you use and the tools to process it.
The NACHA survey mentioned above is important for another big reason: it shows the risks that arise with these developments. Professionals believe that smaller financial institutions need to update their legacy platforms, and 84% say that payment fraud will grow in the next two years.
Fraud risks and concerns will grow and adapt alongside your new service offerings. When new accounts are available, fraudsters will try to infiltrate your member ranks. When the U.S. adopted EMV, for instance, account opening fraud increased by 113%, according to Javelin.
At Feedzai we’ve monitored growing risks in the areas of:
- True name fraud where criminals use personal information, which is becoming more widely available, to apply for accounts and loans in a victim’s name.
- Synthetic fraud that relies on a fabricated identity including methods that tie fake names to real Social Security numbers.
Mobile service adoption is often outpacing the rate at which companies can protect themselves from fraud and theft. It will require a platform to tackle this fraud.
Turning to machine learning
Machine learning for credit unions can free you to become more adaptable to new circumstances and their associated risks, by identifying behavioral signals with hypergranular accuracy, and detecting fraud while limiting the number of legitimate transactions that are blocked.
At its core, machine learning for credit unions is designed to look for patterns and see how they relate to your data, allowing it to generate business intelligence. When you apply ML to existing business and historical information, you can start to see what systems members use more often and what they might be asking for already.
For instance, roughly 83% of people want their financial institution to assist them with their financial goals and to anticipate financial needs. Millennials like the idea of this as a digital service, with 79% expressing interest in a virtual financial wellness coach.
Not only can you use machine learning to determine the desire for such programs, but the same systems may also help you create the rules and advice that such a digital program would provide. AI-enabled technology could look at the goals that users have and, based on the success of similar-looking members, make recommendations that were more successful.
New lines of business, and the fraudsters slipping through the cracks
Reaching new customers is about discovering the data that’s available in the new markets you want to pursue, whether that’s different geographic locations or new types of customers. If you have low penetration in your existing markets, you could apply machine learning to the local chamber of commerce’s data to determine what types of businesses are growing in your area and which of their desired services you’re best able to offer.
You might even be able to determine the best cross-selling and marketing efforts for these new customers as you process more business, HR, purchasing, and other data.
Machine learning is a key tool to use when your credit union is facing new threats or seeks to streamline its anti-fraud efforts. The platform that Feedzai specializes in can reach across product families and services, from ATMs to mobile apps to underwriting departments, to connect data and start looking for patterns and red flags.
The goal of the modern fraudster is to find a gap in data and slip in quietly. They’ll move through different channels and probe you for weaknesses. These criminals are using AI themselves to take advantage of you, and AI doesn’t get tired or need a coffee break; it can keep looking and searching for these gaps at speeds that people cannot. In this game of cat and mouse, advanced machine learning techniques identify potential threats and send them to your team, then learn and refine themselves when your experts say a flagged item was a hit or a miss.
Smart machine learning can even overcome the absence of historical data to provide you with these protections. And there’s no need to be a large credit union to achieve these results. Small but growing brands, with only a small group of developers, can still afford and leverage AI and ML to meet the demands they have.
Adopting an ML tool might be the right move for your credit union to understand and safely achieve its scalability potential.
Latest posts by Ken Bui (see all)
- EU Fifth AML Directive: How Banks Can Prepare for Five Key Changes - November 18, 2019
- AI Best Practices to Improve Enterprise Risk Outcomes - October 22, 2019
- Why It’s So Hard for Challenger Banks to Fight Financial Crime - September 24, 2019
Subscribe to stay infomed