Forecasting Case Volume for 24/7 Critical Support
I combined historical support demand, customer growth, offering mix, and churn data from Salesforce and SQL Server to forecast monthly case volume and recommend staffing across three global shifts.
View methodology and decisions
Business problem
The Critical Situation Support team needed to anticipate demand from a growing high-priority customer base without understaffing critical hours or overloading available engineers.
Approach
- Reconciled case, account, offering, and churn data.
- Built an interpretable multiple linear regression model.
- Analyzed historical demand by geography and hour.
- Translated forecast demand into shift-level headcount recommendations.
Why this model
A three-day delivery window favored an explainable model that leadership could understand and use immediately. The model served as a planning baseline rather than a guarantee of future demand.
Outcome
The forecast informed workforce planning as the team expanded from 3 to 11 members and established continuous coverage. It remained useful for approximately two years without requiring a redesign.
The retained project record reports approximately 87% accuracy with a ±5% tolerance. Team growth was approved and executed by operational leadership; the forecast informed rather than solely caused that decision.