Challenge

The fashion business can be fast-paced due to the dynamic nature of trends and sales cycles, often resulting in sales spikes. These sales spikes are frequent and often unpredictable, particularly for small sellers. For Yoycol, these spikes often make fraud detection challenging, as fraud spikes can be difficult to distinguish from flash sales.

Yoycol focuses on automation and efficiency as key strengths, but managing fraud is often far from efficient, especially with the risk of being placed in monitoring programmes by card brands that want to see detailed remediation plans. Customised clothing products further increase the complexity that results from fraud since shipped products are often unique and challenging to replace or restock. And because many of the third-party e-commerce platform partners that Yoycol integrates with work with small sellers, it's important that it maintains a reputation as a reliable partner.

Solution

Yoycol partnered with Stripe professional services for an eight-week Fraud Strategy Advisory engagement to identify high-impact opportunities and improvements to mitigate spikes of suspicious behaviour before turning into fraud more efficiently and effectively.

Establishing a continuous fraud monitoring process

Preventing fraud requires a vigilant approach. Stripe recommends a continuous, data-driven approach to fraud management using Stripe Radar for Fraud Teams and Stripe data products, such as Stripe Sigma or Stripe Data Pipeline. These tools provide Yoycol with reliable data essential for monitoring fraud trends, and accurately predicting metrics from monitoring programmes and acting accordingly.

Stripe’s approach to the engagement started with a threat modelling exercise focusing on Yoycol’s individual risk exposure and risk appetite. Yoycol’s combination of diverse user personas, from high-volume corporate organisations to small and medium-sized businesses, required mapping Radar rule sets to these personas individually. By analysing the patterns these different personas exhibited and building metrics to track them, Stripe could make specific recommendations to target different kinds of fraud while keeping users happy.

“The Stripe team was very professional when it came to fraud prevention, taking a clear and concise interpretation of our platform's policies and those of the card issuer to determine an optimal solution to remove the card from the credit card monitoring list,” said Wang Xiao Xi, CFO of Yoycol.

Results

At the onset of the engagement, Yoycol's disputes, in combination with the business's growth, put significant operational pressure on its internal risk teams due to the extra effort of tracking card brands monitoring metrics. Stripe professional services quickly employed tools such as threat modelling to reduce Yoycol's operational burden during the engagement, reducing the amount of disputes and early fraud warnings to the lowest levels in the history of the account.

With the use of Stripe Checkout and Link, Yoycol could enable advanced fraud detection with Radar for Fraud Teams, which had a strong impact on Radar's machine learning model, resulting in a false positive rate of less than 0.2% for high-risk Radar blocks. Furthermore, the chargeback rate dropped to less than 0.1% (including fraudulent disputes) while the conversion rate stayed stable, with legitimate transactions.

“In dealing with the fraud crisis, we worked well with the Stripe professional services team's working cadence and timely responses,” said Wang.