The cost of fraud for online retailers isn't just the amount of goods or sales lost. There are a number of other associated costs, not least the risk of being overly stringent in detecting fraud in online sales channels, which can put genuine customers off and cause some to abandon their online purchases before they reach checkout.

In fact, a study by LexisNexis found that every dollar of fraud actually costs merchants $2.40 when you take into account factors other than the fraud itself, such as penalties, wasted shipping costs and lost time dealing with credit card chargebacks.

An insight paper by ACI Universal Payments – Driving up conversion with effective fraud management – looks at how merchants can build an effective fraud filter in their sales channel without blocking genuine sales and without affecting checkout conversion rates. Some of the top strategies discussed in the paper include:

Mine your data for a picture of fraud

Your company's transaction data provides a valuable insight into what differentiates a normal or a fraudulent purchase. Use the payments data to help you understand what characterises a genuine purchase and what to look out for in a fraudulent transaction. This will enable you to use tools to detect fraud while allowing genuine customers to go ahead with their purchases.

Choose the right tools

Traditional fraud indicator tools can work very effectively as part of an overall fraud management solution but if applied incorrectly they could affect conversion rates. It's worth noting that using fraud management tools in markets where consumers aren't used to them, or if the settings are too constrictive, could cause an increase in checkout abandonment.

What to look out for

Fraud indicator tools are available to carry out checks including device fingerprinting and plausibility checks. These rarely cause any issues for conversion rates, according to ACI's report. Tools can also check the velocity – in other words, the number of purchases originating from one account. A higher number from one account/entity can be an indication of fraud, but not in all sectors (this is an useful tool for airlines but not for the gaming industry, for example). The tools can also check the buyer's location and IP address but manual checks are often needed to verify if a payments from a suspect address is genuine or not. Transaction size is also a factor.

Keep up with the fraud evolution

Merchants need to stay ahead, or at least keep up with, the evolving techniques used by cyber fraudsters. More sophisticated techniques can be combined with traditional fraud indicator tools as part of a more holistic and effective way to manage fraud. For example, machine learning models can be configured to block fraud behind the scenes, invisible to shoppers, with no harm to conversion rates.