One number is not accurate.
In a recent article, economist professors at the University of Pennsylvania Wharton School made an honest statement about their forecasts of the American economy “Truth be told, our best guesses just aren’t very good”. Should this comment also be said by those engaged in forecasting future sales for their business?
In making their forecasts more useful, the economists propose they recognise uncertainty when considering the future. They note that serious forecasters embrace uncertainty. For example, weather forecasters will advise of a 70% chance of rain next Tuesday (and therefore a 30% chance of no rain); not that it will or will not rain. From this you can understand the risk of leaving home without an umbrella.
This shows that more value can be gained by forecasting the range between optimistic and pessimistic outcomes, each with a weighting to indicate the level of certainty. The forecast range and associated risks can then be used by planners to allocate resources for the situation being forecast.
If you forecast sales in your business, or feel the effects of bad forecasts, then a similar approach will be of value, especially if the forecast is the aggregate, top down approach. This requires the sales group to develop high level forecasts by each product group or stock keeping unit (SKU); planners then use quantitative methods to break down the forecast to sales by each SKU by location. However, expecting the sales group to achieve the forecast (plus or minus 5%) by SKU is an unreasonable expectation.
To improve the approach:
- gain acceptance from colleagues that forecasting anything into the future (including their own lives) is unlikely to have a ‘correct’ outcome
- establish a pilot program for the new forecasting regime, using one or a few product groups
- identify the assumptions and risks associated with sales of the selected products
- identify optimistic and pessimistic sales levels and weight the likelihood of achieving the sales figures
The most important activity the planners can do with the forecast range is that expected demands for resources can be better matched with available supply. This means that items with long lead times can be acquired at the optimistic forecast level, while items with short lead times can be acquired at the pessimistic sales level, but with provisional orders to increase the quantities at short notice.
To achieve this outcome has challenges. I have been trying to influence clients and students for many years to adopt this approach, mostly without success. One of the reasons is that few computerised forecasting applications allow the use of ranges; another is that managers (who are people) do not want to change from the familiar – they are resistant to change, as are most others in the community.
