RZR’s Approach

Solution:
Deep Neural Network (DNN)

DHgate shared 20 user segments with the RZR team who used them to train DNN. This model training in concert with other proprietary algorithms distilled into four of the best performing segments (dynamic user cohorts) based on ROAS generated:

• Purchase behavior: Users who have made a purchase on the app at some point.
• Checkout users: Users who reached the payment page on the app, but did not make a purchase.
• Added to cart: Users who abandoned the app.
• Active users: Users who are active purchasers on the app

Once narrowed, the DNN model was further optimized by using geography and device model features intoaccount to enhance ML prediction accuracy.

‍

‍

‍

‍

Solution:
Unified Creative Strategy

The creative team’s strategy resulted in 33% increase in conversion rate that drove sales through:

• Subtle Promotions: The ads avoided overt ad language and offered a more organic presentation.
• Price Reductions: Highlighted low prices products to encourage more purchases.
• Targeted Appeal: Product placements were tailored to resonate with specific target audiences, making the ad more relevant.

‍

Conclusion

While DHgate’s spend with RZR saw a substantial 52% increase over the course of the campaign, RZR delivered by consistently exceeding the 450% D30 ROAS goals by 1.7x and reaching 784%.