Using robotic process automation for financial forecasting

Using robotic process automation for financial forecasting



  • Transform Corporate Finance functions beginning with Revenue Forecasting and continuing into Compliance and Auditing functions
  • Currently, the processes are mostly manual with Excel sheets being exchanged, a long lead-time (up to two weeks) for regular cadences (e.g. quarterly forecast updates)
  • The cost to operate is also high given the labor footprint (up to 800 FTEs involved world over-involved in Forecasting alone)
  • Given the high volume, it’s impossible to validate all transactions, so sampling needs to be done (e.g. expense report validation) and risks missing out on some out-of-compliance “transactions”
  • The score, flag, and predict high-risk deals through measurable preventive means



  • Machine Learning-based ensemble models to extract the “seasonal” patterns and apply that to future forecasts
  • There is an ability to apply external input (exogenous variables) to adapt the forecast for new information (e.g. search engine meta-data)
  • Pattern recognition for fraud or abnormalities that can be applied to the entire data lake constructed from finance transactions
  • Optimized decision making in the presence of uncertainty and error, forecasting risks (e.g. minimizing high-risk deals – enterprise agreements)
  • Bot-based interface for interacting with the findings and helping with queries that managers/execs have with the forecasts



For Revenue Forecasting:

  • Faster: Lead-time reduced to two days before the cut-off
  • Better: Accuracy is 2-3 times manual method
  • Cheaper: FTE coverage drops to 25% of the original
  • Clearer: Forecasts standardized and documented for accountability