Split has redesigned its S3 Inbound Integration data pipeline to enhance scalability and reduce costs in response to increasing customer demands. The new pipeline leverages feature flagging for A/B testing, enabling controlled rollouts and seamless production testing without affecting the customer experience. Compared to the existing setup, the new architecture employs Spark streaming, facilitating parallel data processing and significantly lowering resource consumption by 80%. It also improves status report quality with enriched data, allowing customers to self-correct errors. Split's feature flag strategy not only helps in fine-tuning data ingestion but also provides a quick rollback to the previous pipeline if necessary. With successful deployment over a quarter, the new pipeline processes data faster and more efficiently, demonstrating enhanced performance and reliability in real-time environments.