6 Industry use cases for intelligent forecasting

• Automated the core enterprise revenue forecast and separate product-level roll-ups using a suite of integrated predictive models
• Results included: 30 percent increase in accuracy, significant reduction in staff time, and enhanced confidence in business planning on an 18-month view

Supply chain: Market demand planning

• Improved accuracy, time horizon and trust in demand planning for the overall market of a chemical company’s key product
• Tested the new data across dozens of model types to determine the optimal combination
• Delivered a 70 percent reduction in error rate on demand planning
• Identified powerful new business drivers

• Improved scenario-modeling of optimal locations for new retail locations
• Deployed machine learning models to estimate the two-year demand at any new location
• Improved sales accuracy by 2X
• Added increased visibility into areas like labor scheduling, intra-day forecasting and demand planning

Finance: Gross profit and net income

• Developed bespoke models for each aspect of the client’s P&L to automatically generate monthly gross profit and net income forecasts for the next year
• Implemented models into client’s dashboard software to allow for scenario adjustments
• Improved accuracy in net sales and cost of revenue of 75% and 26%, respectively

Finance: Regional and product category sales volume

• Enhanced the accuracy of an IoT-based forecast model using advanced analysis and statistical testing techniques
• Identified a dozen key weaknesses and new enhancements for the model framework
• Improved accuracy by 70–80 percent
• Established an ML-based process to monitor data anomalies going forward

Finance: Sales revenue

• Eliminated a time-draining data bottleneck between sales pipeline and order ledger models with an integrated model framework
• Increased sales forecast accuracy, speed and confidence a full year out
• Significant reduction in time spent on manual bottom-up analysis

Finance: Operating expenses

• Developed and deployed a fully ML-driven system for forecasting all PO expenses
• PO spend was forecasted daily using multiple predictive algorithms, with a unique algorithm for each unique forecast based on accuracy testing
• Built trust with users in the ML forecasts through training, and ultimately reduced time on manual validation

CRM: Subscriber churn

• Improved customer churn forecasting at a hyper-local geographic level
• Deployed an optimized model into a managed signals repository
• Delivered a 70 percent reduction in error rate on demand planning
• New approach allows company to continuously monitor subscriber health and forecast growth/churn by market on a weekly basis