One of the world’s largest producers of food, beverages, home care, beauty, and personal care products—operating over 400 brands across 190+ countries—sought to strengthen its supply chain decision making through enhanced demand sensing and forecasting. With global scale and complex product portfolios, the client required both accurate long term forecasts for strategic planning and agile short term sensing to respond to sudden market shifts.
Traditional demand forecasting approaches, while effective for historical trend analysis, lacked the responsiveness needed to capture short term demand signals such as point of sale data, promotions, inventory movements, and event driven demand changes. The client partnered with Infosys to design a data driven demand sensing solution on Google Cloud that could unify data, improve forecast accuracy, and accelerate decision making.
Infosys designed and implemented machine learning based forecasting and sensing models using Dataproc and Vertex AI, with all data sourced, processed, and visualized in BigQuery. The solution delivered both short term (weeks) and long term (months) demand predictions, supported by intuitive charts and tables. A cognitive chatbot powered by Google LLMs enabled demand planners to query forecast insights conversationally—enhancing usability and adoption.
+6%
Forecast accuracy improvement
71%
Overall forecast accuracy
20%
Faster forecast generation
Upto104weeks
Forecast extention
Key Challenges
- Limited visibility into short term demand fluctuations
- Siloed demand signals across POS, inventory, and events
- Long forecast processing times impacting planning cycles
- Difficulty scaling forecasting models across product lines
- Need for real time analysis within existing data ecosystems
Ready to experience?
TALK TO EXPERTS
Infosys Approach
- Centralized all demand signals in BigQuery for unified analytics
- Performed exploratory data analysis (EDA) to refine predictors
- Built ML and time series forecasting models directly on GCP
- Generated short term and long term forecasts stored in BigQuery
- Fine tuned models to improve accuracy and reduce bias
- Optimized Dataproc clusters for faster and cost efficient execution
- Deployed models with production support and monitoring
The Solution
Cloud native demand sensing solution delivering faster insights and improved supply chain responsiveness
Infosys designed a scalable demand sensing and forecasting solution on Google Cloud to support both tactical and strategic supply chain planning. All data sources—including historical shipments, POS sales, inventory levels, and event calendars—were ingested and stored in BigQuery as a single source of truth.
Machine learning and time series models were built and executed using Dataproc and Vertex AI, accessing data directly from BigQuery and persisting forecast outputs back for downstream analysis. The architecture enabled generation of short term sensing forecasts as well as long term demand projections.
Demand planners and analysts could query forecast results using SQL and visualize trends directly within BigQuery, eliminating data movement and simplifying insights. A cognitive chatbot powered by Google LLMs further enhanced usability by enabling conversational access to forecasts, anomalies, and comparisons—accelerating planning decisions and improving supply chain agility.
Business Outcomes
Improved 13 week forecast accuracy by 6%, reaching 71% overall accuracy
Reduced forecast generation time by 20% through Dataproc optimization
Enabled faster decision making through in platform analytics and visualization
Improved anomaly detection and demand responsiveness
Established reusable forecasting architecture for future extensions
Benefits
Cloud native demand sensing improves accuracy, speed, scalability, and decision making across the global supply chain.
- Faster demand sensing through optimized data pipelines
- Improved forecast accuracy for short term planning
- Scalable architecture supporting extended forecasting horizons
- SQL based analytics enabling rapid insights for planners
- Reduced manual intervention in forecast generation
- Seamless integration within the Google Cloud ecosystem
- Reusable model framework for new product lines
