Fighting Fraud: On-Demand Businesses
Protecting your online business from fraud has never been easy. With the rise of today’s on-demand economy, fighting fraud is even harder. While merchandise-focused e-commerce businesses have ample time to review transactions prior to shipping, orders in the on-demand world are placed and fulfilled in near real-time. The rise of mobile commerce and increasing popularity of app-based services have trained consumers to expect instant gratification. Without the luxury of manual-review time, companies like Uber, Instacart, EatStreet, Turo (formerly RelayRides), and EatNow rely on Sift Science’s large-scale and customizable solutions to fight fraud for their on-demand offerings.
The Problem
Most on-demand businesses want the same thing: to grow their customer bases as quickly as possible. To minimize friction for the customer, these businesses collect only basic information (e.g. name, email, billing address) to make the purchase experience as simple and efficient as possible. Orders are then approved and fulfilled with little time for review, making businesses vulnerable to issues like credit card fraud and fraudulent chargebacks. Bad users rely on this immediacy to take advantage of such organizations. For businesses that promise a good or service within minutes instead of days—like a ride pick-up or meal delivery—fraudsters can easily maximize their returns and ditch the stolen credit cards before being caught.
Additionally, some businesses use a referral system to incentivize existing customers to spread the word. By providing rewards like free rides or delivery credit, businesses aim to attract new users organically and build a loyal following.
Unfortunately for these companies, with referrals comes referral fraud. Customers can abuse referral systems in various ways. The most common referral fraud occurs when customers create multiple fake accounts in order to accumulate more referral credit for themselves. Referral abuse can be very harmful if unchecked by a fraud prevention platform—not only do companies lose the cost of the free “credits,” but they’re also not gaining the new customers they set out to acquire with these programs. Referral fraud makes it difficult to determine whether a business’s referral system is actually attracting as many legitimate customers as it seems.
The Solution
To help on-demand businesses fight fraud, Sift Science uses large-scale, customizable machine learning to make predictions about users in real time. Real-time fraud predictions allow companies to prevent fraud before it happens. More often than not, once a transaction is completed, the window of fraud-stopping opportunity has already passed. Many businesses rely on the Sift Score to automate the decision-making process. For example, you can set your order management system to automatically approve users with Sift Scores under 80, prohibit users with scores above 95 from acting, and manually review those users in between. On-demand businesses that collect limited customer information can also benefit from the intelligent, real-time machine learning that enables Sift to accurately predict fraud even with only a few data points for each user. By using our console and Labels API to identify fraudulent behavior, customers continue to improve their fraud prediction model’s ability to analyze signals and make accurate decisions.
Sift Science customers benefit immediately from a machine-learning model that has analyzed the many types of fraud we’ve seen across the globe. To provide even better fraud protection, we also build a unique model that’s tailored to fit specific needs, businesses, and verticals. Every customer is different, so it’s important to find a way to fight the fraud that is unique to every business. If your fraud-related decision-making relies on only black-and-white rules, you risk losing revenue to smart fraudsters and missing sales from good customers. On the other hand, reviewing every single order slows down the buying process, creating friction for legitimate customers. By using advanced machine learning to identify suspicious activity, Sift Science can help you strike the right balance between automation and manual review so you feel protected against online fraud without hurting your good customers.
Whether you’re an on-demand business or traditional e-commerce retailer, Sift Science can help you identify and eliminate online fraud.
To learn more about e-commerce fraud, visit our free Guides and Resources page. Have questions? Send us a message any time at support@siftscience.com.
