Navigate to: Settings > Configuration > Forecast
Definition
Definition
Automatically reduce peaks in the sales history to lessen their impact on forecasting. Selecting High will reduce more peaks than Low and result in lower forecasts.
Use case
Use case
Setting this parameter to High ensures that an unusual spike in sales does not cause a matching spike in the forecast. This prevents over-ordering and possible excess stock.
Because this is a global setting, it applies to all items in the business. It is important to understand your business as a whole before adjusting this value:
What are the majority of demand types across your items?
What do the model stock and fill rate indicate?
Is your bigger problem excess stock or stockouts?
Simply put, would under-forecasting or over-forecasting be more detrimental?
Based on your answers, you can select your setting:
If under-forecasting is a greater risk, set the parameter to Low.
If over-forecasting is a greater risk, set it to High.
When in doubt, set it to Medium, which works well for regularly selling items or those with a linear trend.
A common misconception is that if most items have sporadic demand, this setting should be Low to avoid reducing forecasts.
Imagine an item that sells once a year. In the month it sells, you do not want the forecast to be lower than expected. If Peak Replacement is set to High, it may reduce that spike incorrectly.
The app avoids this by recognizing demand types that are young, slow-moving, sporadic, or seasonal, and it will not apply peak replacement in those cases.
Explanation
Explanation
Imagine an item that sells 250 units every month. In January, it sells 750 units.
The Peak Replacement parameter determines how much you want to smooth out that spike in sales, which would otherwise cause a matching spike in the forecast.
In simple terms:
Low: The system uses the full 750 units from January.
High: The system treats January as 250 units, as if the spike never occurred.
Medium: The system averages the impact, treating January as 500 units.
This is a simplified example that does not include demand type or other forecasting complexities.
For example, consider an item with steady sales of 250 units per month and no seasonality, growth, or decline. Its forecast is based on a weighted moving average of its sales history. We expect the forecast to be 250 units each month, based on sales history of 250 units each month.
Due to the spike in sales in January, the unadjusted forecast is pulled up to 290 units a month.
Here is how the Peak Replacement setting would affect that new forecast:
With a setting of Low, the forecast remains high at 290 units a month.
With a setting of Medium, the spike is partially reduced, and the forecast will be 270 units a month.
With a setting of High, the system treats the spike as if it never happened, and the forecast will be 250 units a month.
This parameter can be viewed as a less extreme, automated form of Event Correction.
Forget about these 👇 😞 😐 😃 Have your say here!