Revenue Forecasting Automation to Improve Accuracy and Enable Weekly, Location-Based Forecasts

Our Solution

BizAcuity initiated the engagement by collecting historical forecast and actual revenue data from the existing systems to establish a reliable baseline.

Key activities included:

• Identifying all attributes and parameters that influence revenue.
• Conducting detailed discussions with each business unit to capture department-specific variables.
• Incorporating both internal and external variables into the forecasting framework.
• Preparing and consolidating data from multiple disparate sources.
• Performing correlation analysis between identified variables and actual revenue.

Only data points with strong statistical correlation—such as sales pipeline and group rental nights—were selected for modeling.

• An Auto ARIMA model was tested using selected predictor variables.
• A Holt-Winters model was used to forecast annual revenue based on historical data, trends, and seasonality without relying on predictors.
• Holiday impact on revenue was analyzed using a Generalized Additive Model (GAM).
• Pre- and post-holiday effects were evaluated and validated with business stakeholders.
• Holiday impact was incorporated into the final revenue forecasting automation model to further improve accuracy.