Layer 1 copy

Challenge

  • Tens of thousands of orders per month and increasingly a target for fraudsters

  • Using a homegrown system that identified suspicious orders with a value above a certain amount or were from certain geographies, among other rules

  • Rules-based are ineffective once fraudsters start changing their tactics

  • Costs quickly mounted into rapidly growing fraud problem

Results

  • 92% of orders Sift Science identified as fraudulent were indeed fraudulent

  • 8x return on investment on their monthly fees

Summary

Delivering world-class accuracy

Small but nimble, the fraud team at JackThreads successfully tackled their initial trickle of chargebacks using a homegrown system that identified suspicious orders with a value above a certain amount or were from certain geographies, among other rules. The team's elation soon turned to concern as the trickle of chargebacks became a deluge.

JackThreads discovered that rules-based systems cast a wide net and are ineffective once fraudsters start changing their tactics, forcing the team to play cat-and-mouse to adjust rules while reviewing a growing list of suspicious orders. Costs quickly mounted.

JackThreads quickly settled on Sift Science after researching potential solutions. A quick sign-up process, a well-documented API their developers loved, a powerful but intuitive dashboard that made manual reviews easy and, most importantly, a machine-learning based approach that delivered world-class accuracy made their decision simple.

The results speak for themselves. Nearly 92% of orders Sift Science identified as fraudulent were indeed fraudulent - the industry average is 25%.

By using Sift Science, JackThreads saves thousands of dollars a month in chargeback costs and realizes an 8x return on investment on their monthly fees.