Working with the Forecast
Understanding the Demand in AGR
How to Affect Forecasts
There are two ways to affect the statistical forecast, but you could also overwrite it manually using the planner.
While AGR’s forecasting software can forecast a statistically accurate demand, it relies entirely on data. AGR is more efficient in forecasting many items at once but will not stand a chance to the human insights that an experienced demand planner has when looking at items line-by-line. Sometimes, AGR’s statistical forecasts can be significantly improved by manual or automatic intervention.
Adjust the Sales History
By adjusting the sale history, the user can change the forecast any item.
Anyone can adjust the sale history and it will have an immediate effect on the forecast, but note that existing order proposals will need to be recalculated if they should take the new forecast into account.
- Adjust sales by using the item card
- Adjust by keying on a top level in the planner
- Automatic rules (talk to your consultant)
See Inventory: 2.2.1 Editing Product Detail.
Change Forecast Parameters
By default, AGR will use the entire sales history it has for the product as data for the forecast. For items with years’ worth of sales history, not all of it might still be relevant. Skip Forecast to Date in the item details will set a cut-off date for the forecasting module so anything before that date will be ignored.
Good practice is to set a dynamically moving date, i.e. always use 3 years of history if it exists. Reach out to your AGR consultant who can help define this value and set it in the system.
AGR needs at least 24 months of sales history to capture seasonality in the forecast.
Copy the Forecast into the Baseline Plan and Change the Future Demand
Using Plans within the Inventory module, any item’s forecast can be copied into the baseline plan using the “Copy values…” action, where it will effectively override the forecast. The benefit of copying the forecast into the baseline plan is that the baseline plan will be a day-by-day replica of the forecast so any daily, weekly or monthly trends will be copied over as well. This allows the users to key in a single number on an aggregated level that is then distributed to items according to the forecasted distribution. This is a standard functionality in AGR’s planner.
New Item that will Behave Similarly to Another Item
Consider a retailer that sells smartphones. They’ve been selling iPhone 3 for two years and Apple is just released iPhone 4 which is expected to sell similarly to iPhone 3. How can the retailer plan for the new iPhone release in AGR?
There are many ways to create demand for a new item in AGR, here are some of the proposed ways of working in AGR.
We know that it will sell similarly to an item group so we can copy the aggregated forecasted values of the item group for a period to the baseline of our new item.
1. Manually plan in AGR Planner
The simplest approach but often more than enough. Open the relevant time period in the planner, filter on the item number / item group and key in the values.
- Helpful tip: Make use of other data series that might be insightful, such as “Last Year’s Sale” for the previous version of iPhone
- Relevant: how values are distributed when keyed in on an aggregated level in planner
Insert picture of how to key it in, inspect element overwrite names.
2. Connected Items
When a new product is released and will certainly behave similarly to an existing product, i.e. when the iPhone 4 is released it will sell similarly to the iPhone 3, AGR can copy the existing sales history from iPhone 3 to iPhone 4. Once the iPhone 4 has received the sales history, AGR will generate a statistical forecast for iPhone 4 based on the sale history it just received.
a. Don’t Lose Track of the Original Sale
The connected sales history is kept separated from actual sales. The connected sale history is created as “Adjusted Sale” and organic sales of the new item come in as “Original Sale”. Once the sale history has been connected, it can be adjusted. See Adjust the sales history for further information on how sales history can be adjusted. The connected history can also be copied over multiplied by a scale-factor so for a new item that is expected to have a similar demand pattern but 50% more than the old item, it could be copied over with a scale of 1.5.
b. Setting Up Connected Items
This can entirely be managed within AGR using the Connected Items workspace as pictured below, reach out to your AGR consultant to get started with the workspace. They will also help to find ways of automating this further if the information already exists in the ERP system.
Demand from Stores
Demand from stores is a tool that the AGR system uses specifcally for warehouses ordering from suppliers. When it is turned on, the system calculates every night the demand from stores, i.e. how much stores will need to order from the warehouses for the next period (6, 9, 12 months - the period can be adjusted within the system settings). You can see the Demand from Stores as displayed within the Item Card as Purchase Plan Calculated Inventory: 2.2 Item Card - Purchase Plan. The demand from stores is different from the forecast because it takes into account current stock and undelivered, whereas the forecast just shows a store what they are predicted to sell in the next period. The demand from stores is then summed up for every item and then shows the warehouse how much the stores plan to order, which in turn shows the warehouse how much they will need to order from thweir suppliers. If the warehouse would order from their suppliers according to the forecast, then they are not taking into account what the stores have in their current stock and undelivered, mostly likely causing them to over order.
When the AGR system fulfills store replenishment, it is drvien by the forecast, but when is fulfills warehouse replenishment, it uses demand from stores.
When looking at the item card from a warehouse level, the purple line will be defined as “Demand from Stores,” but when looking at the item card from a store level, it will be defined at “Purchase Plan Calculated”.