A Cloud-Native Decision Intelligence Architecture for Sustainable CPG Supply Chain Networks

Many retail and consumer packaged goods (CPG) companies use disconnected data pipelines, which can slow down decisions and increase costs. This paper introduces a cloud-native data architecture that brings together sell-in, sell-out, marketing, e-commerce, and financial data into one managed source of truth. This setup helps teams make timely and reliable decisions. Built on Snowflake, the pipeline uses contract-based ingestion, standardized dimensions, and automated testing. It also sets clear goals for data freshness (media within 6 hours, POS within 48 hours), reliability (at least 99% successful runs), and performance (95% of runs finish within 60 minutes). When tested in three markets and eight product categories, this approach cut the median decision cycle by 25% (from 8.0 to 6.0 hours) and lowered compute costs by 40%. Using standardized KPIs, incremental modeling, and smart retries, the system achieved 95% alignment between planned and actual campaign ROI across over 200 campaigns. FinOps features like auto-suspension, workload isolation, and detailed credit-per-row tracking reduced idle compute spending by at least 30% without slowing performance. The design also supports GreenOps goals by reducing scanned data through pruning and right-sizing, which led to measurable drops in CO₂ emissions without sacrificing analytical accuracy. Overall, these results show a proven, ESG-friendly model for fast and auditable decision-making. The design can be expanded to include streaming data, geo-based experiments, and carbon-aware scheduling, with expected efficiency gains of 10 to 20%. This approach also offers better data governance, stronger privacy controls, and easy scaling to new markets.A