What to do when there are no trends or patterns detected in your historical data
A common misconception about demand forecasting is that straight-line forecasts are poor forecasts because your historical data is not a straight line.
Your data usually has inherent variability, making it jump around. Sometimes that variability exhibits a trend or pattern and sometimes it is just “noise.” In the absence of real patterns, the best forecast might just be a straight line.
For example, your historical data shows varying levels of demand across the past 12 months. Advanced statistical analysis can help you understand whether those variations have patterns and seasonality, or are actually random. If there are true patterns and seasonality, they will show in the forecast.
However, if the variation is actually random, the best forecast will assume the median position between the two extremes and gives a wide band of uncertainty around it. The uncertainty band will cover 95% of the variability seen in the historical data, but the median of the forecast will be a straight line. When dealing with random demand history, the best forecast will always be at the median point because it reduces the potential size of the error while accepting that error is inevitable.
How to leverage straight-line forecasts
Leveraging straight-line forecasts to your advantage requires you to look beyond the forecast itself and consider how it will be consumed in a supply planning environment. In other words, forget about forecast error. The error on a straight-line forecast may be higher than the error you see on other forecasts; not because the forecast was poor, but because the historical data is truly random.
With straight-line forecasts, the emphasis on the confidence interval is even greater. Carrying the uncertainty band (or 95% confidence interval) through to the supply plan allows buyers to understand the variability of past demand and weigh that against their desired service levels. So, while the forecast might indicate the median will produce the least amount of error for that item over the next planning period, supply planners may need to stock more product than the forecast suggests to ensure they do not stock out. In that scenario, planners choose to accept a higher cost of over-producing goods.
How Vanguard auto-optimizes all of it
Vanguard Software has helped companies understand their historical sales data, generate accurate demand forecasts, and optimize their supply plans. The cloud IBP platform, Predictive Planning, was built to auto-optimize inventory levels based on all of the information needed to make the right buy.