But the right balance – and being able to build an effective fraud filter – requires merchants to first see fraud to understand what their fraud looks like, and to recognize how fraud differs from genuine transactions.
Start with data
Through using a wealth of data—based on thousands upon thousands of transactions—merchants can build a picture of both fraudster and customer behavior, which helps to establish the context for fraud management.
The best data solutions dig deep into every transaction, gathering intelligence from every conceivable data point. Payment providers and merchants need systems capable of capturing and collating these massive amounts of data, so that it can be analyzed for trends, even as those trends are still emerging or evolving.
The data that one merchant gathers can be used to sketch out patterns within their customer base, but is also critical to building rich intelligence and a good understanding of emerging fraud trends within – and across – market segments and geographies. This can only be achieved through constantly updating fraud management systems with information from both internal and external sources, including hot card files, chargeback data and information traded on the dark web.
Fraud exchange services can also play a valuable role here — connecting merchants and issuers in a multi-directional information exchange, which boosts the ability of all parties to make accurate and informed decisions.
The richer and more intelligent data becomes, the more accurate, effective and efficient fraud detection strategies become. This in turn reduces the impact on genuine customers and checkout conversion rates.
Leveraging data with machine learning
Machine learning applies pattern recognition techniques to transaction data, from both fraudulent and genuine transactions, to build algorithms that can predict the probability of a transaction being fraudulent. These predictive models, with their ability to extract meaning from complicated data, can identify patterns too complex for humans or automated techniques to flag.
Machine learning, and its ability to crunch huge data sets, is complementing more traditional fraud indicator tools. And when machine learning models are correctly ‘trained’ (using mass amounts of relevant transaction data) and configured correctly by experts, these techniques can be used to block fraud behind the scenes, invisible to shoppers, with no harm to conversion rates.
By more accurately pinpointing fraud, machine learning models also help to support better conversion rates, by reducing false positives and ensuring genuine customers do not get unnecessarily declined or delayed by manual review processes.
Because machine learning models learn from experience, they can struggle to spot monolithic events, and can underperform when customer buying patterns suddenly move away from the norm. It is therefore important that machine learning forms just one part of an overall fraud management solution, just as expert human analysts remain critical to interpreting and acting upon the wealth of transaction data now available.
For more advice on managing fraud and sales performance, download our new insight paper, ‘Driving Up Conversion with Effective Fraud Management’.