Demand planning in the intelligent enterprise requires the “right tool for the right job.” Arguably the most over used tool is the spreadsheet for many good reasons – availability, ease of use and user experience to name a few. Of course the downside is well documented – loss of control, multiple versions of the truth, and the eventual overwhelming nature of large data sets that characterize detailed demand planning processes.
On the other side of the argument is the important requirements of demand planning tools such as an infrastructure that provides a persistent system of record for pre- and post-forecast analysis, a robust time series forecast engine that’s automated and easy to use, and a collaboration platform for refining the forecast results. Let’s look at each characteristic in more detail.
Persistent system of record – implies the system is always available like an enterprise dial tone and contains the latest information along with all the necessary historical data used for the forecasting process. The data repository provides two main purposes. First, to store the data feeds for generating future forecasts that are accurately and consistently tied back to the true enterprise numbers. And second, to serve as the system of record against which ongoing comparative analysis for continuous improvement is performed. The value of the process can be greatly enhanced if the data repository not only contains the required demand input data for the process, but also includes complimentary variables such as inventory, cost structures, open orders and supply data.
Time Series Forecasting tool – is valuable if it provides a substantial increase in sophistication over spreadsheets without an increase in user complexity. Sophistication means the ability for the tool to automatically (without user interaction) extract the important components of a Make-to-Stock type forecast including trend, seasonality and external event (promotional) impact on the final periodic forecast results. The goal is to move to a 20% art and 80% science environment that — when automated — produces a more accurate baseline forecast for supporting the Sales and Operation planning process. The baseline forecast establishes the starting point for the final step in the forecasting process, which is collaboration.
Collaboration – is for refining (not creating) the baseline forecast. The primary importance of collaborative input is to capitalize on the collaborators’ unique local real-time knowledge about conditions not found in the past historical demand data. The pitfall of many collaborative environments is the attitude of the collaborators, such as field sales and marketing people. The value increasing part of the process comes when the field inputs are combined or used to add a different perspective rather than simply replacing the results of the automated time series forecast during the collaboration step. The fault when this occurs lies both with the field people and the operations people.
Field people are motivated to increase the forecast to ensure product availability, as well as their belief that they are better at predicting the future than a more scientific approach. Operations people, while concerned about inventory levels increasing due to over forecasting by field people, enjoy the opportunity to place the blame on someone else for incorrect forecasts – especially when it results in increased inventory value. Operations people also have the threat of poor customer service hanging over their heads due to inaccurate or incorrect forecast mix. All of these issues are sometimes referred to as corporate politics and usually serve to help reduce the effectiveness of the overall supply chain.
Using and believing in the right tools for the demand planning process can go a long way to improve the supply chain effectiveness in the intelligent enterprise but not without solid top-down and bottoms-up management that meets in the middle to add value without significant friction.
Categorised in: Intelligent Technology
This post was written by John Hughes