Automated anomaly detection is improving customer experience, lowering costs and aiding service consolidation
Many banks are undertaking digital transformation to improve customer service, reduce their operating costs and uncover new business opportunities. A leading new technology that promises to be a real differentiator in the banking industry is artificial intelligence (AI) built on machine learning (ML) algorithms.
A McKinsey & Company report says that to compete successfully and thrive, banks must adopt AI technologies as the foundation for new value propositions and distinctive customer experiences. Doing so could potentially unlock $1 trillion of incremental value for banks annually.
Where business intelligence tools fail
Monitoring metrics with dashboards and static thresholds based on human assumptions holds banks back from proactively meeting customer needs and operating at a large scale. When using these legacy systems, anomaly detection is a game of chance, consequently prolonging many of the incidents that hurt customer experience and brand reputation.
Traditional monitoring using static thresholds depends on a human setting maximum and minimum values for a time series metric, and getting alerts when data exceeds those limits (see Figure 1). The area in red falls within the shaded blue area representing normal data patterns but still tripped the static threshold, sending a false positive alert. However, the area in orange fell outside the normal baseline but failed to trigger an alert as it didn’t exceed the thresholds – commonly known as a false negative. This represents a lost opportunity for course corrections that can preserve revenues.
ML overcomes the challenges
ML algorithms scale monitoring across the entire business – and billions of metrics – and independently adapt baselines to each metric’s signature behavioral patterns. This is particularly important for business data, which is governed by the dynamic nature of human behavior.
Figure 2 illustrates an incident related to customer experience where models detected an anomaly in transaction success rate for VISA payments for a particular merchant. The drop in transactions (depicted in orange) would have gone unnoticed if static thresholds had been used, but AI quickly detected it. The alert prompted the bank to quickly reinstate transactions for that merchant, preventing additional losses in revenues.
The highest rewards from using advanced AI come from monitoring business data. According to Business Insider Intelligence research, the majority of potential cost savings for banks from AI applications are in front-office and middle-office use cases such as customer interactions and payments. This information can greatly improve customer experience, reduce costs and more easily consolidate services. Here are just a few examples:
Customer experience
Customers’ needs and expectations for personal banking are changing quickly, and their options for service providers are expanding as fintech companies take their offerings into areas served by traditional banks. AI can help banks do a deep and fast analysis of every individual customer to get better insight into their behaviours and respond with programmes or services to enhance their customer experience and reduce churn.
Operational costs
AI is a critical technology for keeping tabs on KPIs and receiving alerts when something is off. By monitoring factors such as the cost of acquiring new customers, the success of marketing campaigns, infrastructure operations and more, banks can lower their overall operating costs. Personal finance company Credit Karma uses Anodot’s Autonomous Business Monitoring to identify revenue anomalies in real time so they can be resolved before negative impacts occur.
Consolidation of services
The scale of business and the complexity of workflows are changing because applications now require extensive API communications to third-party services – for example, for affiliate marketing programmes. Banks need to monitor the performance of the APIs and the vast amounts of data flowing through them to understand whether a service is cost-effective or not. Only AI and ML can analyse this data and produce alerts in real time.
There are many different areas where AI can produce the critical insight to give a competitive edge in banking services. By automating revenue monitoring, banks can provide better customer experience, reduce costs and consolidate services to give them an edge in what has become an increasingly competitive market.
Visit Anodot to learn about the Autonomous Business Monitoring platform that Payoneer, Credit Karma, eToro, and other leading financial services companies use to proactively monitor their business.
David Drai knows the challenges of data monitoring well. Before launching Anodot, he served as CTO of app-based transportation service Gett; content delivery network and site acceleration services provider Cotendo, which he also founded; and Akamai Technologies.
In these roles, David realised that dashboards and business intelligence were keeping organisations from scaling their business monitoring, and contributing to prolonged incidents and revenue loss. Together with former Chief Data Scientist at HP Ira Cohen, and Shay Lang, Drai created a machine learning-based business monitoring solution that is business-first, data agnostic, and significantly more accurate than other monitoring tools.
Located across several continents, today Anodot enables Fortune 500 companies – from fintech to telecom and e-commerce to adtech – to scale their anomaly detection and forecasting, cutting time to detection by as much as 80 per cent and incident costs by as much as 70 per cent.