Improving forecast accuracy was the number one issue identified in my VP of sales survey. While there are many other aspects of forecasting that are equally as critical, we will focus simply on the accuracy aspect here. Done right, there is no reason that forecast accuracy can’t be improved by at least 30%. Let me show you how we attacked this problem.
In researching what is required to create an accurate forecast, everything pointed to the forecasting environment being the most critical aspect. When I considered the forecasting environment, I came up with three critical areas that required optimal conditions to achieve improved accuracy. These areas are:
- Data entry environment
- Availability and the ease-of-use of ancillary data sets such as shipments, backlog, uncommitted backlog, annual plans etc.
Let us look at these areas in a little more detail.
Too often forecasting is an end-of-the-month exercise that focuses more on completion than accuracy. Having a forecasting environment that is available 24/7 allows the forecaster to enter updated information on a regular basis, at their convenience and when it is freshest in their minds. Inherently this is a more accurate process.
Many of us grew up using Excel as our main forecasting tool and in many ways, it still offers one of the cleanest and easiest-to-use data entry environments. Part of that is due to familiarity, but it mainly due to the visibility of what you are forecasting and the simplicity in entering the data. Lock the few columns on the left that contain the key attribute data such as customer, part number etc. and scroll through the remaining columns to enter the data by month, quarter or whatever timeframe you use. Simple, effective, and a data entry environment I wish more CRM/forecasting tools would emulate. For me, this type of data entry environment is a requirement. Any system that requires me to click in and out of several pages to enter a forecast is ruled out immediately. Several systems on the market emulate this data entry and add significant improvements such as auto-fill algorithms.
The final piece of the puzzle is the availability and usage of ancillary data — the most critical being shipment history and backlog. This data is typically available in one form or another, but the real issue is how easy it is to use. I have had a number of sales managers provide me forecasts that were below their backlog and while I had fun at their expense, the truth is it is very frustrating and kills the confidence in the entire process when these type of incidents happen. The ideal situation is having real time access to shipment and backlog data viewable within (on the same screen) the data entry environment.
Through personal experience, forecasting inaccuracy was reduced from greater than 25% to less than 8%, a 60% reduction of inaccuracies. Based on that and discussions with customers, I believe that optimizing these three areas will lead to at least a 30% improvement in forecasting accuracy.