Forecasting is backbone of entire supply chain planning. Forecasting is done to plan your raw material inventory, distribution planning and planning your FG. Organization strive had to get the forecast number right. Higher the forecast accuracy lower will be inventory requirement for a company and less number of stock outs which means better customer service level and lower Working capital requirement.
Best in class organization improve Forecast accuracy by
- Using better forecasting algorithm and continuously refining the model
- Collaborating with the customer to forecast
- Enhanced analytics to understand the component of demand (Promotion, Seasonality, trend etc) by decomposing the demand
Forecast accuracy can be measured in different ways like
- Root Mean Squared Error (RMSE)
- Mean Absolute Deviation (MAD)
- Mean Absolute Percentage Error (MAPE) /Forecast Accuracy
- Forecast Bias
The commonly used method is MAPE because its more intuitive for business users to understand but the challenge of using MAPE is it does not give a clear picture of errors as it a relative number and absolute number. So to better understand the direction in which the error is we use Bias as a measure.
There are two approaches of measuring MAPE
ABS (F – A)/F – This is used for product that are in GROWTH phase of life cycle. If this is the formula used one can clearly see to reduce the error the planner will be encouraged to slightly over forecast. But this being a growth SKU over forecasting will not cause the inventory levels to rise. The cost of losing sale for Growth SKU is higher than the cost of having high inventory
ABS (F – A)/A – This is used for Products that are in Mature (stable) or EOL phase. For mature and EOL type of SKU it’s always advisable to maintain just enough inventory to satisfy demand. Having high inventory for such type of SKU’s will lead to increase in obsolesce cost and markdown cost
Forecast Bias – is an indicator to understand which way the forecasting is moving- is it over-forecasted or under-forecasted SKU.
The negative Forecast Bias indicates the forecasting technique is continuously under-forecasting the demand. This gives the trigger for the planner to increase the forecast number and improve forecast accuracy.
The Positive Forecast Bias indicates the forecasting technique is continuously Over-forecasting the demand and the planner needs to take step to reduce the forecast number to improve forecast accuracy.