Managing forecast consumption settings can be a tricky, error-prone activity. We asked SCM 2015 speaker Sean Mawhorter of SCM Connections to host an online Q&A to address users' questions and share tips on optimizing forecasts for more accurate demand and supply planning results. Check out an excerpt from this Q&A session and see whether your questions were answered.
Sean will be one of the featured speakers at SAPinsider’s SAP SCM 2015 conference in Las Vegas, March 30 – April 1. For more info on the event visit our website at Logistics & SCM, PLM, Manufacturing, and Procurement 2015 and follow us on Twitter @InsiderSCM
Comment from Rohit
We changed the packaging specifications for a couple of SKUs and gave them a new material number. How do we forecast for these new materials using the forecast of the old ones? We have tried life cycle planning, but the demand planners think it is too tedious to maintain all the entires. Is there any other option other than life cycle planning and realignment/copy?
Sean Mawhorter: Unfortunately, there are not too many more options to address these requirements (although they are fairly common). One option is to copy the sales history from the old item to the new item, but that can throw off your aggregate sales history. Plus, it’s a manual process.
Comment From Pavani
Can we assign a custom category to TLB order STO in APO system, so that we can influence consumption of forecast by this custom category?
Sean Mawhorter: It is possible to set this up in the consumption configuration. It is a combination of the requirements planning strategy and ATP categories and assignments.
Comment From Anna
Do you have any recommendations on when to use backward vs forward consumption methods and the time frame for the orders to be considered?
Sean Mawhorter: This is a great question, but would require a long answer. Short answer is that a combination of the sales order volume/quantities, order drift, and forecast bias should lead to segregation of your materials and/or material-location combinations and backward/forward settings should be established and managed using these groups.
Comment From Rangarajan
Can you explain order drift?
Sean Mawhorter: Order drift is the propensity for an order to be in a different bucket that its intended forecast. An indicator of this can be where the forecast accuracy for a single bucket (month) is usually much lower than the forecast accuracy of a multi-bucket window (e.g., 3 months). So timing is more the issue than the quantity itself.
Comment From Rangarajan
Can we maintain forecast consumption settings in SNP different for each material?
Sean Mawhorter: This is standard functionality. The consumption parameters are set in the product master at the product-location level.
Comment From Suat
Consumption of different forecasts: If we have more than one forecast line (50 pcs and 100 pcs for example) in SNP relevant for consumption by a sales order of 120 pcs, how we can find a way to consume fully the first one and partly the second one? The reason why we have more than one forecast figure is because we want to add specific characteristics to represent priorities, for example.
Sean Mawhorter: Have you investigated the use of extra forecast characteristics for this one (i.e. adding priority as an additional characteristic to the customer to be used in the consumption)?
Comment From André
How can we use forecast and safety stocks together to increase the SNP plan quality?
Sean Mawhorter: This is an area where many are confused as to what dial to turn when...
Safety stock is really the dial to turn to adjust inaccuracies in the forecast quantity itself, versus consumption settings, which are used to address inaccuracies in the timing of that forecast.
There are many variables that can affect these; the key is knowing which dial to turn when. 😉
Comment From Rangarajan
Can forecast consumption work other than the Active 000 Planning version?
Sean Mawhorter: It can, but you need to use the save_multi bapi to create the sales orders in the other version. Sales orders are normally sourced from ECC and only populate version 000 by default.
To view the rest of the transcript, click here.
