SAP ECC Forecasting functionality usage

[former_member221091]

Hello Folks,

Have a query regarding usage of SAP ECC Forecasting functionality. Kindly note, this is not a how-to query. Also, it refers to ECC alone.

I have seen few clients use a third party tool for forecast generation even though the functionality is available with SAP. The interfaces do pose its typical challenges and despite that; I don't see the advantage accrued by the business when you do a cost-benefit analysis.

Would request your thoughts and guidance on usage of Std. SAP ECC Forecasting functionality within SAP and its merits/de-merits.

Thanks.

[DominikTylczyn]

Hi Sameer, you've brought up a very interesting topic. Frankly I don't believe in over-sophisticated forecasting tools or models. They use heavy statistics few people understand and are able to validate. Also the more complex / sophisticated model, the more historical data it needs. Therefore more and more history is used to forecast the future. Nowadays business changes fast, very fast and decision based on distance past seldom play you nicely in the future.

I attended very interesting session on forecasting during SAP SCM 2017. The guy made very strong argument that the only usefully model is first order exponential smoothing: https://help.sap.com/doc/saphelp_scm70/7.0/en-US/3a/2dc95360267614e10000000a174cb4/frameset.htm with alpha parameters = 0.2 - 0.4. That's it. Of course SAP ERP ECC provides the functionality out of the box.

My view on way companies implement sophisticated i.e. expensive third party solutions is as follows. First, forecasting is typically used to plan sales. Sales planning is intrinsically difficult. What is even more difficult to sale according to plan and to establish efficient motivation system for the sales people to stick to the plan. The "solution" here is to use heavy statistics to the rescue. Then whatever is the forecast / sales plan and the plan is not realized we can always say "we are not responsible for the plan; it's the system that calculated the plant".

Hope my post is going to start a nice discussion here.

Best regards

Dominik Tylczyński

[former_member221091]

Hi Dominik Tylczyński,

Thanks for providing a detailed input. I do share the same sentiments echoed out in your post and hope others take a cue from you on the discussion ahead.

We really don't need a sophisticated tool.

The current scenario is where we have a MTS scenario and the forecasts are used for just generating the process orders. Hence my thought process was to use the standard functionality. Yes; the First-order exponential smoothing model may more than suffice.

Besides, investing in a costly tool; the challenges associated with the interfaces; I don't think so we accrue any proportionate benefits at all.

Thanks again for sharing a good link too.

[DominikTylczyn]

Couldn't agree more 🙂

[former_member221091]

Folks,

Any thoughts on the above?

Regards,