January 28, 2021
Consumer finance, fraud detection and business intelligence are among the places in financial services where AI spending pays dividends, according to members of Protocol's Braintrust.
Good afternoon! In this week's Braintrust, we asked the experts to tell us how they evaluate AI investments in financial services by highlighting the factors they see as being most determinant of a good return. Questions or comments? Send us a note at email@example.com
President, Operations & Technology at Mastercard
When deciding to invest in AI and build out new capabilities, I see two main categories of innovation. The first is eliminating toil — enabling machines to do more routine tasks and freeing up people to do different and often more fulfilling work. This is an important engine to always have revving in any business; it typically delivers value that's consistent and low-risk, albeit incremental. But the bigger return on investment comes from the second category: creating new value and applying AI to achieve things that could never have been done before. This is when we get truly transformative benefits.
For us, that breakthrough AI application centers on combatting fraud. Billions of dollars are at stake every day, when it comes to cybercrime and fraud in the financial sector. As a payments leader — powering digital commerce globally — we experience 200 fraud attempts per minute.
Using existing data, such as transactions and authorizations, we've been able to develop and train our predictive AI models through supervised learning techniques. This enables us to predict and detect fraud with incredible precision and speed, while also reducing false positives — in other words, we want to avoid stopping legitimate transactions from occurring, as it hurts the merchant and consumer experience. Another reason this AI innovation has been so successful for us is the way it's built: our multi-layered intelligence approach leverages 13 different AI technologies, which are combined in a truly novel way. With $55 billion dollars worth of confirmed fraud blocked since it launched in 2017 and a 60% decrease in false declines, our AI-powered fraud detection platform continues to be one of the best technology investments we've made.
Chief Data Officer at Moelis & Company
At Moelis & Company, our AI/ML/Data Science & Analytics strategy is client-centric with a focus on differentiated and customized client content and services. We create business value beyond efficiency and use AI/ML and data science to help differentiate the firm across multiple business lines. As an investment bank offering advisory services, we evaluate ROI across several key areas including improved client experience and retention, new products and services, profitability, enhanced employee engagement and higher productivity.
As the adoption of AI/ML in financial services becomes more pervasive, it is important to look across the organization to identify opportunities to generate concrete business value from AI/ML beyond cost reduction. Enhancing client experiences by improving their business intelligence and decision-making can increase revenue, particularly when it helps a company create differentiated content, target prospective clients, design portfolio strategies or evaluate potential trades. By evaluating such opportunities through a commercial business lens, rather than purely as a technology issue, ROI comes starkly into view, and measures such as client satisfaction, revenue growth, cost reduction, and productivity gains are becoming prerequisites to scalability.
Complementary to the business metrics used to measure ROI, our client-centered approach to AI/ML reinforces our culture of strong, bespoke client service. Our initiatives are highly cross-functional and collaborative. We measure the potential impact and business value of every project, prioritizing based on client engagements and deals, which shapes our near- to long-term strategic focus across the organization. This enables us to be agile and pivot as needed to focus on business use cases that are commercially oriented and drive differentiable results for our clients and the firm. AI/ML skills are not just in the domain of data science, data engineering and IT teams though. The opportunities, outcomes and responsibilities should be cross-functionally sponsored and accountability maintained across all levels of the organization to continue to enhance our client service.
General Partner at Homebrew
The secret of AI/Ml is that the values really depends upon the underlying data. In financial services that's always been true but the types of data that have been utilized for underwriting risk have been limited and haven't changed.
The highest ROI of AI/ML in financial services comes from using new types of data in combination with AI/ML to better underwrite and understand risk. This is possible in lending, insurance, fraud and compliance, trading, collections and product personalization to both increase revenue and drive down costs.
Executive Vice President, Chief Technology / Information Officer at Nasdaq
We think of return in terms of the profound impact and fundamental change that AI and ML will bring. NLP and computer vision will be powerful as we are continually inundated with data across varying degrees of structured and unstructured formats. The ability of these technologies to crunch through data and intelligently report in an automated format will bring scale and productivity that will enable professionals in our industry to cut through noise and focus on the needle in the haystack to, for example, improve protection for your company, improve returns or create new opportunities. The cloud service providers have made headway here as have various startups.
We use many forms of AI and ML at Nasdaq. For instance, our sector-based Advisory Services team leveraging these technologies to deliver intelligence to our listed companies. They can cut through an enormous amount of data to enable a stronger foundation for their research. As they start from a higher quality baseline, our analysts can be more strategic and give our listed companies a greater level of service and value. This is the tip of the iceberg of what can be achieved in financial services if these same principles are applied elsewhere.
AI and ML will also have a profound impact on anti-financial crime. The United Nations estimates over $2 trillion is laundered annually. Banks spend over $1 trillion to fight financial crime, but on average 1.5% of alerts result in a suspicious activity report and 1% of criminal funds are seized. Due to the sheer volume and complexity of the data flows being managed, many banks are expending huge amounts of human capital that are using siloed, ineffective technology to comply with monitoring obligations. AI and ML capabilities can bring effectiveness and efficiency to investigations. We are hugely interested in this space and our recent launch into automated investigations technology and acquisition of Verafin underscores this. As we work with others to make the detection of financial crime more effective, we believe there is an opportunity to raise efficiencies that will eventually help financial institutions eradicate flows of illegal money — having a positive impact on the global economy.
Executive Vice President & General Manager, Identity, Fraud & DataLabs at Experian
At Experian, we see AI showing fantastic returns and for us it's been more of a foot race to identity those areas where it can make an impact, develop an AI driven solution and get it into production.
In financial services, we're leveraging AI to optimize credit underwriting, mitigate fraudulent behavior risks and enhance customer experiences in a fully digital environment. We're also applying it within our own operations to improve data hygiene, entity resolution, attribute development and optimizing our internal decision systems.
Where do we find the best ROI? It really depends on a few things — like how optimized is the existing process through data and analytics prior to the application of machine learning and the rate of change taking place in a given application.
It's most common for us to see the highest rate of return in the detection of new fraud signals. Mostly because:
- The dynamics of fraud are constantly changing.
- New data that provides insights into emerging signals frequently becomes available.
- It's transactional in nature, which means the "size of the prize" can be significant.
- And the non-linear optimization approaches that underpin many machine learning techniques match well to the signals that can be detected.
As a result it's not uncommon for us to see 10x to 20x multiples on ROI in terms of investment versus return.
Partner at CapitalG
The financial sector has experienced exciting innovation this past decade. The dual factors of increased mobile penetration and the weakening of big banks following the 2008 financial crisis presented startups with opportunities to reimagine consumer banking for a mobile-first world, especially in categories such as payments, checking, investing and insurance. Over the past few years, these upstarts have continued to grow, and several of them are poised to blossom into enduring franchises.
However, despite incredible early progress, we are still in the early stages of the transformation that's yet to come. While many fintechs are successfully offering products online, many of these products are largely based on offline analogs. The next exciting progression of finance will use AI and ML to build automation and personalization into consumer finance. The potential implications for the ecosystem — and, more importantly, for consumers — are absolutely enormous.
Today, many financial products have largely become commoditized. Brand equity matters, but consumers often shop around seeking improved rates and experiences. However, as the next generation of fintechs increasingly leverage AI and ML, they will be able to help consumer finance progress from commoditized "products" to higher value personalized "experiences." A consumer will be able to say "Plan for a vacation," and their providers will be able to automate decisions around investing, saving and borrowing based on their knowledge of the consumer's preferences.
This transformation represents a huge paradigm shift in how finance has operated for decades and an opportunity to transform the sector from one frequently dreaded by consumers to one that's actually beloved. The future is fast approaching — and I, for one, can't wait.
See who's who in Protocol's Braintrust (updated Jan. 27, 2020).
Questions, comments or suggestions? Email firstname.lastname@example.org.
Kevin McAllister ( @k__mcallister) is a Research Editor at Protocol, leading the development of Braintrust. Prior to joining the team, he was a rankings data reporter at The Wall Street Journal, where he oversaw structured data projects for the Journal's strategy team.
More from Braintrust