TaiwanRetail and e-commerce2019 – ongoingCarrefour Taiwan
Machine learning models for precision online advertising
Carrefour Taiwan cut its cost per online ad action by 40% with machine learning models for customer targeting
Problem
Carrefour Taiwan set a goal in 2018 to grow its ecommerce revenue sharply. Online advertising drove much of that revenue, and the retailer needed a more cost-effective way to target it.
Approach
In 2019 the retailer uploaded customer data to Google Cloud's AutoML and built four models for customer lifetime value, segmentation, conversion rate and churn. Google Cloud's professional services team mapped out the development steps, and the Google Ads team advised on the performance indicators. The lifetime value model identified high-value customers, who received most of the ads, and the segmentation model supported more personalized ads.
Results
Takeaway
The retailer had no staff experienced in complex models and relied on the vendor's services team to set them up. It tied the models directly to ad targeting and measured them on advertising results.
Categories
- Size
- Large enterprise
- Industry
- Retail and e-commerce
- Business function
- Marketing and sales
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Google Cloud's customer story describes how Carrefour Taiwan built four AutoML models in 2019 to target online ads, alongside a later move of its systems to Google Cloud. It reports a 40% lower cost per action and a return on ad spend 2.64 times higher than before. The page carries no date; the Internet Archive first saved it on November 22, 2022.
Full details are in the original sources.
