TaiwanMedia and entertainment2022 – ongoingUDN Group
User preference prediction model for online ad targeting
UDN Group raised the clickthrough rate of some ad categories up to four times with a user preference model
Problem
The Taiwanese newspaper group sells online advertising on udn.com and wanted more accurate ad targeting.
Approach
In 2022 UDN Group started building a user preference prediction model with the marketing partner Merkle, using Vertex AI. The algorithms extract keywords from UDN's online content for tagging, match content tags to ad categories and generate target audiences. The team ran A/B tests to pick the best algorithms and finished development in eight months.
Results
Takeaway
The gain was reported for certain ad categories, not for all advertising. The company did not say which categories or how the baseline was set.
Categories
- Industry
- Media and entertainment
- Business function
- Marketing and sales
- Innovation type
- Process innovationProduct and service innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Partner co-development
- Evidence
- Vendor case study
Sources
Source summaries
Google Cloud's customer story describes how UDN Group and Merkle built an ML model on Vertex AI to predict user preferences for ad targeting. It reports an increase in unique clickthrough rate of up to four times for certain ad categories, and that 40% of ad orders are supported by data-driven advertising services. The page carries no date; the Internet Archive first saved it on July 25, 2025.
Full details are in the original sources.
