Background and objective
MyDress, a Hong Kong e-commerce brand that launched in 2013, offers clothing and accessories from more than 200 fashion brands. While popular with shoppers, the brand found that customer-acquisition costs were rising in its crowded marketplace. MyDress had historically created its target audience segments based solely on past experience, but this limited data view was leading to spending inefficiencies.
MyDress decided to implement AI technology to increase sales conversions and drive revenue. The brand also wanted to improve its overall return on advertising spending (ROAS). To do this it would have to find its most valuable customers by using precise targeting and by delivering dynamic product recommendations.
Working with Appier, the brand's first step was to identify the most valuable user segments—those most likely to purchase. By examining a massive amount of available user data, including information such as frequency of site visits, products viewed and additions to shopping carts, Appier’s AI identified more than 1 million user segments. From there, Appier further isolated a set of the most valuable segments. These were:
- People with a cart value of more than HKD$500 (US$64.50) across three days
- People who made additions to the cart four times in one day
- People whose purchase value was higher than HKD$750 (US$97) in 180 days.
MyDress then employed Appier’s dynamic product recommendation tool with product feed implementation, placing dynamic display advertisements on both the open web and Facebook. This provided more opportunities for reaching potential customers and gave MyDress flexibility to customize product feeds and ad formats for specific events, such as promotions and deals.
Appier’s AI and machine-learning technologies optimized the campaign performance over time.
Over the course of the five-month campaign, customer-acquisition costs dropped by 26%, while conversions increased 121%. The brand's overall revenue during the campaign increased by 127%, and its ROAS by 30%. The dynamic creative approach triggered clicks, driving cost-per-click down by 56%.