Our client’s finance team manually creates quarterly budgets for three business functions across eight regions using data dumps from multiple sources. The process lacked flexibility and parameter control resulting in inaccurate forecasts.
In addition, the forecast development was effort intensive and time consuming – took approximately seven hours per region. With multiple regions and consolidation, this effort consumed hundreds of hours every month. The tedious manual process included collation of data, pre-processing, outlier handling, selecting the best model based on error, creation of validation metrics, and visualizing forecasting results.
Scaling the model across new divisions and regions consumed a lot of resources; hence the company sought to automate repetitive tasks and implement an automated forecasting solution.
Infocepts developed an automated forecasting solution that automated expense forecasts for the subsequent 24 months across businesses & regions. The solution had built-in validation to test the accuracy of forecasts against previous cycles/models.
Infocepts built this end-to-end solution using the Dataiku platform, seamlessly integrating it with existing applications to enable process automation. It uses eight predictive time-series models (using various machine learning algorithms) and a unique model selection and forecast combination approach. From multiple available models, it allows users to select the most accurate and best-fit model for unique time-series data, then finalize and share forecasts to the business for budget planning.
The solution’s scalability and ability to generalize allowed easily addition of variables and prediction parameters for forecasting, such as registration costs and other departmental expenses. The solution was later expanded to include newer divisions and regions.
Automated forecasting enables the finance team to make on-the-go budget planning decisions. Our solution has enabled our client to realize:
- Cost savings of over $1M through shifting FTE time away from manual tasks to value added analysis activities
- Enabled top-down budgeting, planning, and revenue management capabilities ultimately giving better control of cash flow
- Improved decision-making (e.g., easily defined monthly, quarterly, and yearly forecasts).
- 50% reduction in errors, with increased accuracy and reduced bias from predictive analytics models.
- 95+% reduction in manual effort—average time for regional forecasts now 20 minutes rather than ~7 hours; overall time down to ~2.5 hours from the original 8-10 business days
Please fill the below details
Want to know more? Talk to us today.