|
Forecasting - Unique Data |
Question
How can we streamline our unique forecasting process?
|
Challenge
A client needed to have the forecasting approach for their flag-ship brand embedded into a desk-top application. Over the years they had developed an internal forecast methodology driven by data from very diverse sources (governmental statistics, industry publications, as well as internal data). Readily available forecasting software products were not equipped to handle the unique data characteristics, neither were they able to accommodate the proprietary forecasting approach the client had been comfortable with for years. The client was looking for an application that provided them three major features
- A faster turnaround (the government was moving from a quarterly to a monthly release of its statistics)
- An application that provided output that could feed directly into their internal systems for marketing, production as well as finance for bidding
- Enabled them to pick the "best fit" forecasts from among hundreds of parameterized simulations according to custom metrics.
|
Solution
First we took the client's forecasting methodology and incorporated various error and benefit gauges. Next we embedded the concept into a desk-top application automating the input procedures from the various data sources. The application was given tabular as well as graphical presentation screens. Last we added a batch module, which enabled the user to launch a search of the "best" fit by examining hundreds or thousands of different combinations of parameter settings.
|
Outcome
The application reduced the time to create the client's customary forecasts by about 80%. The batch process feature allowed the forecast analyst to launch the "forecast search" as an over-night job. The forecast staff meetings are now spent on talking about the implications suggested by the what-if scenarios and not on validating proper data aggregation or input methods.
Back
|
