Case Study: Knorex achieves sub-100ms real-time bidding and 30% cost savings with Google Cloud Platform

Knorex Turns to Google Cloud Machine Learning Engine to enhance real-time bidding in programmatic advertising

Knorex is a Singapore-based adtech company that runs XPO, a self-serve programmatic advertising platform for advertisers and agencies across APAC. Facing billions of daily bid requests, the company needed to decide and return bids in under 100 milliseconds while scaling globally, reducing latency across regions, and supporting hundreds to thousands of machine-learning models — requirements its previous infrastructure could not meet.

Knorex migrated to Google Cloud Platform, using Compute Engine, Cloud SQL, Cloud Storage, Google Kubernetes Engine and Cloud Machine Learning Engine to automate model training and deployment. The move cut model training and deployment from days to about one hour, reduced instance setup from two days to three hours, achieved sub-100 ms bid processing at high volume, and lowered costs by up to 30%.

Knorex

Phu Le

Head of R&D