Case study
Younique
Streaming commission calculations and field analytics for a global direct-sales cosmetics business.

The commission calculation bottleneck
Younique relies on an international network of Independent Presenters who sell through digital channels. During major promotions, its legacy systems often took eight hours or more to calculate presenter ranks, commissions, and team bonuses. Those delays limited visibility for presenters and slowed marketing decisions.
Calculating commissions as sales arrive
The Emperitas team rebuilt the batch calculation system around event-driven processing with Apache Kafka and Apache Spark. They developed services on Kubernetes to apply commission rules to incoming point-of-sale data and built executive and field dashboards powered by SingleStore.
They also developed a machine-learning model to identify indicators of presenter churn and the risk of rank demotion.
Reported results
Commission and rank calculations fell from eight hours to under five minutes. During peak events, the system handled more than 50,000 queries per second while dashboards rendered in less than a second.
The project reported an 18% improvement in 90-day retention for newly onboarded presenters through automated milestone alerts.