8 Strategies to Fight False Positives
Card-not-present (CNP) fraud was up 40% in 2016.
If this surge has you feeling defensive, a word of caution: clamping down in the wrong ways can increase false positives and actually harm profits more than the fraud losses you are trying to prevent.
The best way to avoid over-reacting is to use a multi-layered, enterprise-class fraud prevention solution. You get the accuracy and precision you need to confidently distinguish between fraudulent and unusual—but legitimate—orders.
Further, you’ll get robust data that will let you track Key Performance Indicators (KPIs) so you can easily employ the following 8 strategies to dial in approvals and declines so the revenue and profits generated by borderline approvals consistently exceeds the losses on those borderline approvals.
- Dynamic Risk Scoring
- Real-Time Data
- Artificial Intelligence (AI) and Machine Learning
- Order Linking
- Big Data: The Network Effect
- Automated Rules Engine
- Persona Technology
- Business Intelligence Reporting
- Dynamic Risk Scoring. Static point scoring systems and historical databases are simply not viable in today’s world of rapidly-metastasizing fraud. Dynamic Risk Scoring monitors a transaction post-approval, continuously assessing the transaction every 30 minutes for days after initial approval. Each time, all associated data, linkages and connections are re-evaluated to detect evolving behavior that might change the risk score. This continuous scoring acts like a time machine so you can return to a transaction and revise your approval decision, armed with richer data. You can reconsider the order and decide to:
- Halt shipments and cancel charges to avoid losses of the goods, while also preventing the expense of chargebacks.
- Cancel subscriptions or accounts to avoid or minimize losses on digital goods or services, while also preventing the expense of chargebacks.
- Real-Time Data. Analyzing fraud risk using data that is days-old or even hours-old causes blind spots. But the use of real-time data prevents fraudsters from escaping detection by rapidly swapping devices, credit cards, proxy servers, Internet Service Providers (ISPs), etc.
- Order Linking. Merchants using only internal customer data have limited information available to evaluate borderline orders. In the face of ambiguity, the natural inclination is to decline. But enterprise-class fraud prevention solutions use Order Linking to compare millions and millions of links across merchants, channels, countries, and time zones to find the connections that reveal fraud...or legitimacy.
- Artificial Intelligence (AI) and Machine Learning. Capitalize on the massive computing and memory capabilities of computers to analyze billions of discrete data points, detect associations, weigh probabilities, and quantify risk far more effectively and efficiently. AI sharpens the distinctions between the transactions with higher fraud risk and those with lower fraud risk in a dynamic and rapidly-changing environment. Machine Learning then makes it possible to act upon those distinctions in an automated way to increase order approvals without increasing fraud losses.
- Big Data: The Network Effect. Big Data can reveal patterns and behaviors that may appear harmless in isolation but reveal themselves as fraud when seen across millions of transactions in real-time. The other side of this coin is that understanding what true fraud looks like and knowing its most obscure markers enables clearer, more confident distinction between transactions that are truly fraudulent and those that merely look suspicious. The result is higher approval rates, fewer false positives, and increased revenue.
- Automated Rules Engine. Quantifying the risk for each of the hundreds of discrete data points associated with a transaction enables merchants to avoid single-factor decisions and to instead precisely weight multiple, dynamic attributes so decisions can be made in an incredibly granular fashion. This makes it impossible for even sophisticated fraudsters to identify the factor(s) that are leading to their detection and prevents them from gaming the system.
- The term “Persona” describes a series of proprietary statistical mathematical plotting algorithms. Personas make it easier for fraud prevention systems to identify normal behavior, so those legitimate orders can be automatically approved. This results in fewer reviews, faster throughput and higher sales.
- Business Intelligence Reporting. Criminal rings and fraudsters grow increasingly sophisticated every day. The ability of a merchant to gain insights from their transaction data stream is essential. This requires a business intelligence solution that provides an integrated dashboard incorporating data mining, reporting, and workflow tools to re-structure raw data into a usable format. With greater insight, you can more precisely define fraud risk, leading to better distinction between fraudulent orders and those that only look suspicious.
Download the eBook “The Silent Sales Killer: False Positives.” You’ll discover even more about to employ these 8 strategies and reduce false positives—increasing increase revenue without increasing fraud losses.