We hear a lot of talk about automated transaction reviews without human intervention being better. But is it really? Fast approval of clearly valid orders is essential – but what happens to questionable orders? If the automated system automatically declines potential fraud, you may be turning away good customers and losing their lifetime value.
Today’s AI-enabled algorithms depend on insights to learn about the most current fraud schemes. A truly balanced fraud prevention approach should automatically approve good orders and decline clear fraud — all while improving speed and effectiveness for both. That’s why contextual reviews of questionable orders are so important.
Let’s look at what’s involved in that review process.
As soon as a customer starts the checkout process, an AI-enabled algorithm begins analyzing data points and comparing them to known customer behavior, global and current fraud trends, and data sources such as:
Once the scan is complete, the algorithm assigns a fraud score to each order that delivers a decision:
Unlike most fraud solutions, in a contextual review, suspicious orders are not automatically declined. Why? Because, on average, 90% of those orders are legitimate.
Specially trained fraud analysts combine industry insight with extensive experience to make the transactional decisions that keep customers happy and revenue growing.
If analysts are concerned that a suspicious order is fraudulent, they will examine the order further. Analysts look at information that may go unnoticed by a machine but will provide valuable clues:
In rare cases, the analyst may contact the customer directly, using a preapproved call script. This tends to come across to customers as “white glove” customer service.
This approach—combining fraud filters with contextual review—secures the lowest false decline rates. For example, at ClearSale, across 6,000+ client accounts, 91.3% of orders are automatically approved – 8.3% are reviewed by an analyst but do not require a phone call, and only 0.4% are subject to a review involving a phone call.
Let’s look at what a fraud analyst considers during a contextual review.
Any good fraud analyst understands that approving as many orders as possible and never losing revenue by declining a good order is in the best interest of the business.
“Suspicious” behavior doesn’t always mean the order is fraudulent. Novice online shoppers and general user error can make a legitimate order look like fraud. Analysts use their experience and extensive expertise to determine if the order is legitimate and can be approved. And they feed what they learn about identifying fraud back into the algorithm, which allows it to recognize more fraud faster.
Analysts will often start here. If the order comes from a longstanding customer with no history of fraudulent behavior, analysts generally won’t add friction by requiring additional verification steps, and they’ll often approve the order immediately. But for new customers, or for existing customers making an unusual purchase, analysts will want to look at more information before approving the order.
A fraud score on an order is an all-in-one number that’s derived from a comprehensive analysis of a credit card transaction to evaluate the transaction’s risk. This score provides a quantitative way for analysts to assess the likelihood a transaction is unauthorized or fraudulent.
Frequently, a combination of suspicious indicators leads a fraud analyst to take a closer look when evaluating a transaction. These might include:
These red flags for fraud allow an experienced analyst to make a quick determination that the transaction is high risk and should be declined. For many businesses, only 2%-3% of orders are in this category.
Transaction reviews generally happen almost instantly. Most orders are legitimate, so there’s no reason for fraud analysts to hold them up for further review.
On questionable transactions, analysts will need to conduct extra research to confidently determine whether the order is legitimate. For a well-trained analyst, this can take between two and 24 hours to gather additional information before making a decision. For example, the analyst may want to verify an address, or the analyst may reach out and contact a customer directly to confirm the customer’s identity.
It’s this extra time that enables analysts to uncover fraud strategies that a simple fraud filter or automated fraud solution never could. And that little extra time pays dividends. With every order processed through an AI-enabled system, it “learns” more about your business. As a result, you’re able to approve more orders and deliver a superior customer experience.
Because there are so many possible indicators of fraud, many businesses rely on outsourced solutions to spot and prevent fraud.
These fraud management programs use customized software to conduct fast, expert analysis on every transaction to identify fraud before it has the chance to damage a retailer’s reputation and revenue. These programs also deal with fraud on a large-scale basis, ensuring they have the perspective and expertise to spot fraud trends and patterns that only become apparent when a range of orders are analyzed.
Consider purchases being made with a gift card. Most fraud protection solutions automatically approve such transactions. And yet, these often are in fact fraudulent purchases. In one recent scam, a fraudster was placing legitimate-looking orders with an online business. After the transaction was approved, the fraudster would call to cancel the transaction and request the refund be placed on a gift card (which are often untraceable). Upon receiving the gift card, the fraudster would place a new order and pay with the fraudulently obtained gift card.
In another case, a fraudster would place an order and request the order be delivered to the actual cardholder’s address – thereby ensuring the order would pass through the fraud filters undetected. Then, once the order was in transit, the fraudster would call FedEx pretending to be the business and request the order be delivered to a different address and right into the hands of the fraudster.
Patterns like these can be extremely difficult to spot without the trained eye of a professional fraud analyst.
When it comes to successfully defending your business against scheming fraudsters, you need a partner who focuses exclusively on fraud protection, has the expertise to prevent fraud from damaging your business, and is constantly monitoring the changing fraud landscape.
Here’s what you should look for in a partner:
The hybrid model pairs advanced machine learning with highly trained human analysts to address the friendly fraud threat in real time. Not only does that help protect your business over the long term, but by applying a global lens and a large database of orders across industries, a hybrid solution enables both AI and analysts to quickly recognize fraud trends and help clients eliminate fraud threats and prevent chargebacks — all while approving more orders, faster.
Chargebacks can siphon away a small business’s profits. Every dollar in chargebacks costs businesses $2.50 in time, fees, goods, and shipping, not to mention the costs associated with penalties and punitive actions if your chargeback rate crosses the 1% threshold.
That’s why small ecommerce businesses need a fraud prevention strategy that includes comprehensive chargeback services such as:
As a pioneer and innovator in the fraud prevention industry, ClearSale recognizes your business needs a customized solution for its complex needs.