Real-Time Fraud Detection – In Your Next Pizza Order

PinIt

To combat online chargeback fraud, delivery.com turned to machine learning and behavioral analytics.

Name of Organization: Delivery.com

Industry: Retail, e-commerce

Location: New York, NY USA

Business Opportunity or Challenge Encountered:

The web has become the world’s busiest ordering venue for everything from food to car parts to travel reservations. For online retailers that offer delivery, the opportunities are immense, but so is the potential for mistakes or even outright fraud. Pymts.com estimates that online fraud accounted for 3.4 percent of all e-commerce sales in the third quarter of 2015, and about 2.1 percent of all transactions.

The challenge for any e-commerce outfit is to be able to quickly separate legitimate orders from fraudulent ones without disrupting customer service.

For delivery.com, the on-demand nature of its business requires that the company quickly confirm and fulfill orders. The company offers an online portal to facilitate delivery from a customer’s favorite local businesses, including restaurants, wine and spirits shops, grocery stores, and laundry and dry-cleaning providers. A smooth checkout process is essential to delivery.com’s business model — but fraud and the operational overhead in dealing with attempted fraud cut into profits.

There are two forms of chargeback fraud – one involves stolen credit card information. The other, sometimes referred to as “friendly fraud,” involves customers attempting to obtain a refund even though they received their requested goods or services.

To achieve the fastest possible service, delivery.com had to approve as many orders as possible with limited fraud protection in place. Manual review of fraud was an operational challenge since customers expected such fast delivery — within an hour for most orders. That timeframe that didn’t allow for lengthy fraud review.

Attempting to identify shady customers was a challenge, and a heavy-handed anti-fraud approach could impact legitimate customers, who might experience delays or be rejected.

The result was a level of fraud that had both a high operational cost and a high hard-dollar cost, as chargebacks began cut into revenue.

How This Business Opportunity or Challenge Was Met:

Delivery.com turned to an online fraud prevention solution intended to reduce overhead and chargeback costs within a high-velocity business. The solution, offered by Forter, is called “Decision as a Service,” providing real-time, fully automated, guaranteed fraud decisions for every single transaction.

As a customer takes steps to place their order, Forter’s machine-learning technology works behind the scenes. It combines behavioral and identity analytics, examining thousands of data points when analyzing transactions, and leverages cyber intelligence, elastic identity, and behavioral analysis to evaluate each strand of a customer’s profile individually and as part of the whole.

The moment a potential customer lands on the website, the system begins tracking their behavior and comparing even subtle elements of their web activity to known and tested behavioral trends. At purchase, the model reviews the numerous attributes, and balances the relative importance of each element to derive a total picture behind the transaction.

Once a customer has clicked to purchase, the system returns an approve/decline decision in milliseconds. An online dashboard displays some of the information that was key to making the decision, such as geo-location, behavioral data and payment details.

 Measurable/Quantifiable and “Soft” Benefits:

As a result of the Forter implementation, delivery.com has boosted its transaction approval rate to 99.5 percent and has decreased chargebacks by 69 percent.

Since every decision is real time, and customers are never asked for additional information, the system is effectively invisible to consumers.

As a consumer-facing e-commerce company, conversion and customer satisfaction are paramount for delivery.com, so maintaining the best possible user experience was essential. The system is designed to treat all customers as “innocent until proven guilty,” according to Forter.

Until recently, fraud prevention which aims to drive sales would not have been possible. But advances in machine learning, behavioral analytics and cyber intelligence have changed that.

(Source: Forter)

Related:

How to apply machine learning to real-time events: an online guide

Special report: Context-aware recommendations for e-commerce

Archives: compliance and anti-fraud analytics

Leave a Reply