What Best-in-Class Fraud Prevention Looks Like: Benchmarking Fraud Performance in Banking

For financial institutions, the focus of fraud prevention isn’t just to trim losses. It’s to maintain the bedrock of banking: customer trust. However, for too long, fraud teams have operated in a vacuum. Boards and executives frequently ask, “Are we doing as well as other banks our size?” Until now, the answers were buried in anecdotal comparisons or high-level industry stats that were often too broad to guide a specific strategy.

Fraud benchmarking breaks this pattern by moving the conversation from guesswork to hard data. By aligning your performance against industry-standard tiers, you can identify the specific technological and operational gaps holding your team back. In this article, we’ll break down the key metrics that define “Best-in-Class” performance and provide a clear roadmap for institutions ready to move into the top tier of fraud fighters.

Key Findings

• Fraud benchmarking provides an objective view of how a financial institution’s fraud prevention performance compares against its peers by using key metrics.
• The process anonymizes and aggregates data across different institutions, allowing financial institutions to understand their fraud prevention strengths and areas for improvement.
• Feedzai recommends concentrating on a program’s efficiency (value detection rate) and effectiveness (false positive rate).
• Feedzai insights find that positives can cost nearly 3x more than the fraud itself from lost revenue, customer churn, and addressing complaints.
• Benchmarking establishes performance tiers ranging from top “Best-in-Class” organizations that demonstrate a value detection rate (VDR) of more than 70% to “Developing” organizations with a VDR of less than 40%.

Key Metrics Used to Benchmark Fraud Performance

To capture the essence of a fraud program, Feedzai recommends concentrating on two primary metrics: efficiency and effectiveness.

Value Detection Rate (VDR)

VDR is the proportion of total attempted fraud value successfully identified and stopped. It is the ultimate measure of how much money is being saved. For example, a 60% VDR means the bank stopped 60% of would-be fraud value. 

False Positive Rate (FPR)

Effectiveness is measured by FPR, the ratio of legitimate transactions flagged for every one confirmed fraud case. This is a measure of customer disruption. A lower FPR indicates a precise system that inconveniences fewer genuine customers. This metric is also vital for financial health; Feedzai data indicates that the hidden costs of false positives, including customer attrition, lost sales, and service resources, can be nearly triple the value of the actual fraud losses.

What is Fraud Benchmarking?

At its core, fraud benchmarking provides an objective, industry-level view of how banks perform against their peers on key metrics. By anonymizing and aggregating data across institutions, banks can see where they stand in comparison to the market, identifying both areas of hidden strength and clear opportunities for improvement.

Benchmarking establishes performance tiers (Developing, Strong, Market-Leading, and Best-in-Class) that serve as targets for aspiration. It moves fraud management away from vague goals toward concrete, data-driven targets.

• Best-in-Class: These top-tier banks are the gold standard, stopping the vast majority of fraud while maintaining incredibly low friction for customers. They have invested heavily in integrated technology and real-time data, allowing them to catch over 70% of fraud value.
• Market-Leading: This group stays ahead of the industry pack with solid detection rates that sit clearly above the median. While they are highly effective, they may still face a slight lag when adapting to the newest, most sophisticated fraud patterns.
• Strong: These institutions often use a mix of traditional rules and machine learning to achieve respectable results. However, this performance often comes at a higher cost, such as more false alarms and a heavier workload for their review teams.
• Developing: Programs in this tier usually rely on older, rules-based systems that struggle to keep up with evolving threats. While they currently catch less than 40% of fraud, high-impact changes in data and technology can quickly help them climb to a higher tier.

Why Fraud Benchmarking Matters?

Fraud management results are highly visible. Yet many institutions lack a clear measuring tool. Benchmarking fills this gap by:

• Informing Internal KPIs: Banks can see exactly what top-tier performers achieve, helping to set realistic and competitive internal goals.
• Guiding Investment Decisions: When you know the specific delta between your performance and “Best-in-Class,” it becomes easier to justify budget for new technology or data sources.
• Leveling the Playing Field: By using a common scenario (e.g., a fixed 0.1% intervention rate) benchmarking ensures that differences in performance reflect true efficacy rather than just a bank’s “appetite” for flagging transactions.

What Defines Best-in-Class Fraud Prevention?

The “Best-in-Class” tier represents the gold standard in fraud fighting. These institutions achieve a VDR above 70% while keeping false positive ratios around 12:1 to 14:1.

So, what are “Best-in-Class” tier banks doing differently?

1. Unified Data: Best-in-Class organizations use a 360-view of multidimensional data. In other words, it’s not just about the ability to obtain signals, but combine them into one platform. 
2. Platform Integration: external intelligence orchestrated into different points of the journeys
3. Rapid Deployment: Maintain operational agility while still being able to quickly deploy rules and models into production.

“Banks want to understand how their fraud management results stack up against industry peers. Our State of Fraud Performance report is the first to provide both strategic insights on performance, as well as recommendations on how to make meaningful improvements with new tools, techniques, and technologies.” — Dan Holmes, Vice President, Global Product Planning & Strategy, Feedzai

The Biggest Gaps Between Strong and Best-in-Class Banks

Many banks fall into the “Strong” category, catching 40-60% of fraud. While respectable, the gap to “Best-in-Class” is often defined by two specific areas: feature coverage and agility.

Strong banks often rely on a hybrid of legacy rules and machine learning, but they may struggle with high false positive ratios, sometimes exceeding 30:1. This creates material customer friction and an immense operational review burden. The jump to the top tier requires decommissioning legacy infrastructure that constrains the ability to react quickly to shifting attack vectors.

How Leading Banks Improve Fraud Detection Rates

Improving detection isn’t about one single “silver bullet”. It is the outcome of coordinated progress across several dimensions.

Data Hygiene and Orchestration

High-performing institutions treat data as a strategic asset. They ensure that data exists and is available in real time. “Best-in-Class” banks have the capability to rapidly onboard new data sources and, just as importantly, retire them when they become “noise”. It’s also essential to label data accordingly. Improving the quality of fraud labels (e.g., distinguishing between authorized scams and unauthorized fraud) allows models to learn faster.

Integrated Environments

One of the most significant advantages for top-tier banks is an integrated environment. Instead of stitching together separate tools for engineering, training, and deployment, they use a unified platform. This compresses the time between identifying a new fraud pattern and having a calibrated response in production.

What Fraud Teams Should Prioritize Next

If a bank wants to move the needle on its performance tier, these three pillars should be the priority:

• Champion/Challenger Frameworks: Running new models against live traffic without disrupting production allows for rapid, evidence-based iteration. Make sure to invest in technology that is going to get you a step closer to helping you simplify operations, including by consolidating vendors and progressing to your end state.
• Multi-layered controls approach: Instead of relying on simple binary choices (approve/decline), a multi-layered approach leverages a spectrum of contextual controls. This allows for tailored interventions, such as dynamic friction, step-up authentication, or silent monitoring, ensuring customer experience is protected while risk is managed in real-time.
• External Data: To achieve a 360-degree view of risk, top-tier institutions enrich proprietary data with external signals. This comprehensive approach is essential for proactively detecting and adapting to sophisticated threats like synthetic identity fraud and authorized scams.

Fraud Performance Is No Longer Static 

Fraud performance results from intentional choices in data, technology, and culture. While performance gaps are vast, they are surmountable. “Best-in-Class” leaders currently stop over 70% of fraud at a 0.1% intervention rate. Institutions in Developing or Strong tiers can achieve rapid gains through targeted improvements like better data orchestration or refined customer segmentation.

Benchmarking is a critical asset to staying ahead of new fraud trends. Adopting a RiskOps philosophy that integrates millisecond signals and continuous feedback reduces losses while building trust and competitive advantage. By clarifying a benchmarked foundation, your financial institution can define what excellence looks like today and provide a clear path for tomorrow.

Additional Resources

• Blog: Application Fraud: How to Detect & Prevent It For Your Business
• Report: Fraud Detection Benchmarking Report
• Analyst Report: Merging Fraud Detection, AML, and Data Orchestration
• Solution: Stop Transaction Fraud and Protect Honest Customers