It’s no secret that retailers are declining transactions from honest customers in their attempts to block online fraud.

Figures vary widely for the cost of “false positives” relative to fraud, with some reports suggesting the losses from false declines are several times the fraud losses, while others give a lower ratio.

Fraud costs can be underestimated. Besides loss of goods and chargeback fees, there are penalties, wasted shipping fees and lost time dealing with chargebacks.

ACI Worldwide, the Universal Payments company, cites research suggesting that every dollar of fraud actually costs merchants $2.40 with these additional factors. And if fraud persists, merchants may have to pay higher transaction fees, or risk losing their merchant account altogether.

On the other hand some overzealous fraud prevention systems place too many genuine shoppers into the same bucket as fraudsters. At least one-third of shoppers declined due to a false positive fraud flag abandon their purchase.

The goal, says ACI, is to find the balance that allows merchants to minimize fraud and maximize the checkout conversion rate. ACI’s new White Paper – “Driving up conversion with effective fraud management” – available at  focuses on how merchants can build an effective fraud filter for that sales funnel — one that is not too fine.

Here are some key points from the White Paper:

Know the fraud

Before building an effective fraud filter, merchants need to understand what their fraud looks like. For that they need plenty of data. A really effective understanding comes from updating fraud management systems with data from both internal and external sources — including hot card files, chargeback data and information traded on the dark web. Fraud exchange services can play a valuable role.

Use tools appropriate to the context

Merchants need the right fraud detection tool, configured to suit their business and customer base. ACI suggests that in some markets (Brazil and China) using a well regarded fraud prevention tool like 3D Secure can cause genuine customers, who are unused to it, to abandon the cart. But in other markets it can reassure customers about the security of their payment card and drive up conversion..

Some fraud indicator tools – for example device fingerprinting and plausibility checks — rarely cause any issues for conversion rates. However, others need more careful evaluation and configuration. ACI cites some examples:

Velocity checks

These check how many purchases are coming from a specific origin. For some transactions, eg airline ticket purchases, they can be a powerful tool, but in telecommunications and gaming they will decline many genuine shoppers.

Machine Learning

ML builds algorithms that
can predict the probability of a transaction being fraudulent. These can identify patterns too complex for humans or automated techniques to flag.

Because machine learning models learn from experience, however, they can struggle to spot monolithic events, and can underperform when customer buying patterns suddenly move away from the norm.

Tailored rules and alerts

Combining pattern recognition capabilities with robust, flexible fraud rules can provide a powerful, optimized fraud prevention system.

The tailoring of fraud rules can start at a sector level to help with broad trends. For instance, in many sectors, fraudsters are most exposed at the time of fulfillment, so rules flags can be used to introduce extra checks at that stage where needed.

Introducing fraud rules in silent mode can also be part of a tailored process. Profiles that are higher risk can be reviewed instantly using silent rule dashboards.

Identifying good customers

Many merchants will have in place a black list to help automate fraud decisions. A white list is also prudent, to ensure genuine customers are recognized and treated accordingly. And black lists can be temporary (say for 24 hours), allowing shoppers to try again later.

Analytics and behavioral profiling capabilities can also be effectively used to understand customers better, and to make quick changes to fraud rules. While high-risk shoppers are frequently automatically declined, with advanced analytics capabilities in place, merchants can instead accept these transactions and monitor shoppers’ subsequent activities.

In addition, payment methods can be tailored to shoppers’ risk profiles. Safe shoppers can be offered the entire range of options, while riskier shoppers are only offered more secure payment methods that often include additional risk checks.

Continuing role for fraud analysts

Outside of the manual review process, there is still a vital role for fraud experts to play in the day-to-day management of fraud strategies. Almost every fraud management tool and process needs to be configured and constantly monitored by experts to get the best value from it and to make informed decisions.

As a fraud platform itself takes on most of the analysis, fraud experts can use their knowledge plus the intelligence and tools at their fingertips to support broader areas of the business. They can help support promotions and business growth, and provide a valuable window to customer behavior and purchasing trends.

The White Paper includes case studies of how ACI has reduced fraud, chargebacks and manual review rates for leading companies.

For example, a leading football club was experiencing high rates of fraud on sales of its promotional kit. Using three months of the transaction data ACI built a customized fraud solution, that helped to reduce chargebacks to 0.22%, and manual reviews by 71%.

And a European airline was finding that fraudsters using US cards were jeopardising its US sales channel. In the first year of a collaboration with ACI, the airline found that an average 97% of all fraud was denied by the new rules, saving it $3 million. Manual review rates were reduced from 12% to less than 5% and chargeback rates fell to less than 0.1% on average.