How we optimized operations for a global HVAC provider with an S&OP process, improving forecast accuracy by over 50% via increased transparency through BI, supply planning process with digital tools, and ML-based demand forecasting algorithms.
A global industrial company providing HVAC solutions faced a dual challenge: stockouts and excess inventory across categories. Operating under a sales-first model, the company pushed forecasts into the supply chain, resulting in low demand accuracy and frequent short-notice changes.
With lead times exceeding two months—and up to six for some products—a decision was made to introduce Sales & Operations Planning (S&OP) to restore balance.
Unique challenge: Initially, only symptoms were evident; we identified root causes and defined the problem during the engagement. Embedded operationally within the client's company, our goal was to design, implement, and ultimately transfer ownership of the process to an organization which was still to be established.
Our team tackled this multifaceted challenge through three integrated workstreams—supply planning, demand forecasting, and business intelligence—employing a rigorous diagnose, prototype, pilot, and integrate framework:
Diagnosed misalignments between sales, production, and procurement, revealing a lack of coordinated planning across global operations. Prototyped a streamlined supply planning process at a single high-volume site, designing tools to identify production bottlenecks from sales forecasts within minutes. Piloted successfully, then scaled the process to all six production plants worldwide. Incorporated raw material bottleneck analysis, upgraded tools and integrated both production and procurement into the S&OP cycle, ultimately transitioning ownership to a new client team.
Improved a manual forecasting system with basic SAP linear regressions for a 4,000-product portfolio. Uncovered a critical dependency: only ~800 primary products drove demand, with the rest as dependent accessories. Developed and coded a custom algorithm in R to forecast accessories based on primary trends, cutting complexity and lifting accuracy by over 50%. Created an Excel-based GUI to make predictions explainable, accelerating adoption and enabling real-time adjustments.
Established a robust reporting ecosystem, transitioning to structured Power BI and SAP dashboards that provided real-time visibility into inventory, forecasts, and supply risks. Developed conceptual and new Power BI reports for SCM reporting, conducted employee training sessions, and created a strategy for leveraging Artificial Intelligence in SCM reporting including execution of a pilot project.
Data-driven forecasting using Palantir Foundry to optimize spare parts inventory and availability.
Identified €600M in untapped sales potential and reduced RfQ processing time by 20%.
3-phase analysis enabling data-driven investment decision with 20% model accuracy improvement.
