How Musinsa scaled its audience engine with ClickHouse Cloud and reduced TCO by 71.4%

The Audience Engine stores an exploded mapping of users to audiences, producing roughly 1.1 billion rows for 16 million members and 2 500 audiences. On self-hosted ClickHouse, growth forced the team to add external compute resources because storage and compute could not be scaled separately on a single node. This coupling meant that a heavy query in one workload could degrade performance for business logic and batch jobs running on the same cluster.

Operational overhead became a bottleneck as the business evolved quickly; maintaining patches, upgrades, and capacity planning consumed engineering time that could have been spent on feature development. The team noted that managing ClickHouse themselves was more burdensome than anticipated, especially given the rapid pace of new audience definitions and marketing experiments.

Moving to ClickHouse Cloud addressed these constraints by decoupling storage from compute, allowing each to scale independently based on workload demands. The managed service also provided automated backups, patching, and access to solutions architects, which reduced the need for dedicated database administration staff.

After migration, Musinsa reported an 86.5% reduction in storage costs and up to a 71.4% drop in total cost of ownership. Real-time ingestion, previously reliant on tools such as Databricks Auto Loader, was simplified through ClickPipes, which directly streams data into the cloud service without intermediate processing steps.

Because only one feed covered this announcement, there is no cross-source corroboration to validate the claimed percentages or the specific architectural details. Engineers should treat the reported figures as self-reported outcomes and consider conducting their own proof-of-concept to verify cost and performance impacts in their environment.