on ‎2010 Jan 25 1:00 PM
I want to create a customer segmentation report on FI-AR data. I want to drill down by Company Code in the rows and view the 0DEB_CRE_LC amount.
In the columns I want to create a structure which contains the 0DEB_CRE_LC value for the top 10 customers per company code, then the value for 10% to 20% customers, and so on.
0COMP_CODE____TOP10%____10TO20%____BOTTOM10%
1000_____________£1,091______£297________£105
2000_____________£23,546_____£15,021_____£3,030
The idea is that I can show for each company code how much of the total accounts receivable balance is due from the top 10% of customers, and cascading down to the bottom 10% of customers.
Any ideas how to create this?
Edited by: Khaled McGonnell on Jan 25, 2010 5:50 PM
I have defined a calculated key figure which is 0DEB_CRE_LC and have calculated single values as Olympic Rank Number. I want to use the rank number in another key figure. However, the rank number isn't used, the actual value is used. How can I make it so that the actual rank number is used?
Request clarification before answering.
Use of Virtual Key Figures in InfoProvider enables use of rank in other formulae.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Khaled.
You cannot do this in the query designer alone, because you want to display the key figure values of multiple ranks according to 1 characteristic (customer) and then break those values down by values of another characteristic (compcode). Thus, for each cell the selection for both characteristics changes (different compcode, different "set" of customers).
I would try to solve it by flagging records in a DSO before the cube, if your datamodel allows it. You would have to add "lookup" DSOs, where you stage the data for each rank, so you "just" have to check for each transaction datarecord, if it is part of any of the ranks, if yes, flag the appropriate infoobject (so you need one for each rank, 10%, 10-20%, etc).
Like this, you can build your report using the flags in each key figure selection in your columns and the comp code in the rows.
However, as more transactions come in, for each comp code the rank of customers can change, so already loaded records need to be changed, as customers go in and out of the ranks. Thus, you will have to reload either all data (because you can't know how far back in time the first record of all current incoming records is and maybe volume is low) or read the acc.doc.numbers from the last delta and use these as a selection in a full load to another DSO.
Also, you need to fix the time selection according to the business perspective (top 10% per which timebucket?), when you stage the data and hence, you cannot aggregate across this level of (time)detail, but you will be able to drill down on any other object and get a meaningful result.
At the end of the day, I would consider building a dataflow off the AR DSO I assume you have in place now. Handle all of the above in this dataflow, essentially one more dso and a cube on top with a number of lookup dso's on the side. You can provide ranks for different timebuckets like this, if you add a char to identify which timebucket it is and let users select or make separate reports.
At any rate, it can be done, but you need a separate dataflow-setup to handle it.
Br
Jacob
Edited by: Jacob Jansen on Jan 25, 2010 7:42 PM
P.S: Well, maybe you can make it work with multiple conditions on your key figure selections, but since you are not displaying the customers, I am not sure how/if you can make the condition work properly... maybe the "char assignment" when defining the condition will help, but I think you need to display the objects selected to make it take effect...
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Jacob,
Thanks for a very comprehensive response. I also found materials on the use of Virtual Key Figures to address this. Effectively I create a virtual key figure in the InfoProvider which is the rank value. When used in the query, this KF actually gives me the values of the ranking, not the monetary value.
i.e.
0DEBITOR________0DEB_CRE_LC________ZDEBRANK______Ranking_____FormulaA______FormulaB
00000001________£1.000.000,00___________1______________1___________1______________£1.000.000,00
00000002________£900.000,00_____________2______________2___________2______________£900.000,00
00000003________£500.00,00______________3______________3___________3______________£500.000,00
ZDEBRANK is a virtual key figure in the InfoProvider
Ranking is a calculated key figure created in BEx
FormulaA = ZDEBRANK * 1
FormulaB = Ranking * 1
Being able to use the ranking value enables the development of IF scenarios based on rank which is very useful and key to this report.
Unfortunately, and a sad end to a story where a solution seemed in sight, I am restricted currently to the data model provided (just the Business Content) so I have to use Excel formulae.
Edited by: Khaled McGonnell on Jan 27, 2010 10:07 AM
| User | Count |
|---|---|
| 7 | |
| 6 | |
| 5 | |
| 4 | |
| 4 | |
| 4 | |
| 3 | |
| 3 | |
| 3 | |
| 2 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.