what are the tasks needs to be keep in mind during imple

[prasnop]

Hi Experts,

Could some one please help me on what are the thing needs to be considered during the DP implementation?

What are the tasks needs to be keep in mind to fetch the master and historical data from ECC to APO  or ECC -> BI -> APO ?

As a DP consultant on what specific tasks need to be focused in ECC to bring the master and transactional data to APO?

Thanks,

[Former Member]

There are infinite possible ways of implementing a best demand planning solution.

For a start you may need ECC for generating your own data sources (t-code SBIW). There are several possibilities of getting the data you want to replicate from ECC to APO. The choice depends on a "VISION" of the end solution. So start with a pen and paper and write down the story yourself.

The best best practice is simplicity of design esp. of this BW business of extracting, transforming and loading data because anything complex there means headaches for everyone who is supporting the system later. In fact it can be a pain in all the wrong places even for the person designing it, once trouble happens.

Primarily you need some transaction data for "raw" key figures like sales requirements (orders, deliveries, invoices) and some master data for characteristics (e.g. customers, locations, products and their attributes). You need to be clear on what data your data sources must contain at the bare minimum. So check if there are already replicated data sources in BW, that can be readily used. This saves time on re-inventing the wheel. Do not extract junk that will never need for demand planning.

Like all mechanical systems, the best DP solution should also have the following characteristics

-Failsafe

-Easy to understand by a third person

-Easy to use.

-Easy to FIX if a problem arises.

-Easy to read labels of any objects you create.

-FAST- less key figures, larger buckets.. month is faster than week and week is faster than day.

-Less clutter - no un wanted key figures. just because a demand planned wanted it (A + Bxc + D + E - Yxz can be done with just 3 key figures or you can have 10)

-Less clutter - No unwanted characteristics

-Role based data views . access to only so much data as in needed

-Do what SAP best practices on solution management for demand planning says. I presume you may not have access to the same. Please google and you will land on one.

Save enough time for concentrating on improving the quality of demand plan. A sound knowledge of statistical models is definitely important if you want to leverage statistical forecast and spend time reviewing the time series of each and every product/char you want to forecast. Most consultants spend no more than 2% of the total time on this and leave it to poor users to figure it out who will later complain that statistical forecast is BS and hence excel is better... I will if I were on the receiving end.

Hope this helps

Regards

Borat.