SAP APO Demand Planning - usability versus functionality

I have heard from Consultants and business managers across many companies: Very rarely there is user adoption of standard APO forecasting techniques.  Users typically use an off-line forecasting process and enter the forecast into the planning book or even upload it with IT help.  Why is it?

Are the standard models and algorithms not sufficient for the business models?  Or is it becuase the users and IT staff do not understand the statistics and functionality behind APO?

Let me address some possible reasons for this unfortunate situation.

The complexity and scale of the data and functionality actually drowns out the requirement for business process mapping and leveraging the functionality to meet the business needs.  So the key ingredient of building the bridge between what the users want versus what the tool can do is perhaps left out in the cold – either due to the hectic project management schedules or due to the lack of understanding by IT staff/Integrators etc.

I think this is more than just “selling to management” – I am sure management will love to buy the concept of increasing the usability of any gadget/tool. Nothing worse than creating shelfware that is trashed after a few years and few million dollars down.

In many implementations, the business process blue printing is either poorly done or not done at all. May be it is in the project plan, but this task is marked complete sheerly by a group of tech consultants mapping the old database to the new system or just the same inefficient process into a system.  Although we all acknowledge that the business process need to be mapped, re-designed and documented before the system implementation begins, very rarely this is funded by corporate managers as a worthy effort.  They like the tool and they want the tool……..

There is also this other possibility that the statistical knowledge is lacking from both the user representation as well as from the consulting resources. For example, I have found many implementation consultants still struggle with questions in implementations:

1. How does one manipulate the trend profiles to adjust trend?

2. What is the difference between using the ex-post forecast vs. Median?

3. Why does a moving average forecast model produces a forecast that rarely moves?

4. How much history should one need to create usable models?

5. When does one use Croston’s models?

and a host of other simple questions that can be resolved with an expert user/consultant that understands the business of demandplanning and the math behind the models.

Finally, there are some short comings and challenges with the tool itself, just like any other tool. Consultants and Integrators either do NOT know what these are or just afraid to acknowledge these. It is important to recognize this, so they can help the users navigate around these pitfalls to improve usability of the tool.

Please read http://demandplanning.net/solutions.htm to understand what can be done to improve the tool.  Some of the stuff is simple and perhaps will mean consulting costs in the order of 5% of your total budget to implement the tool in the first place.