Jim Powell

← Projects

IRR and Payback Automation

Predicting which new builds will actually earn their return

Automated IRR and payback analysis for residential and commercial new build, with machine learning to project 10-year growth.

Machine LearningPythonFinancial ModelingIRRData Integration

The problem

Return-on-investment for new-build projects was evaluated in Excel cost models fed by many source systems, so it was slow to refresh and hard to compare estimates against what actually happened.

The approach

  • Pulled data from multiple source systems and reverse-engineered the Excel cost model, rebuilding it so it could process actuals.
  • Extracted the original assumptions from hundreds of initial estimates.
  • Applied machine learning to forecast 10-year growth for completed projects that were three years into revenue generation, then identified which projects would meet their intended IRR.
  • Presented the findings to senior leadership.

Outcome

The analysis led to multiple initiatives to improve business performance, and it seeded the later CB lookback and payback dashboards that found small budget increases were falling through the cracks.