Imagine you have spent the last 9 months developing a sales Data Warehouse and an Analysis Services cube.  Your end users love what you have built for them so far but it’s never good enough right?  You have a new requirement to allow your users to enter in projected or target sales for future months.  You think about your options for a while and narrow it down to 3 possible ways to solve the problem.

Option 1:  Have the users enter the targets into an excel spreadsheet and then import the data into your warehouse using a tool like SSIS.

Option 2:  Have the users enter the targets into a SharePoint list and then import the data into your warehouse using a tool like SSIS.

Option 3:  Enable writeback on the measure group’s partition so the users can enter the data in Excel and it will automatically write the data back to the cube and a separate writeback table in a SQL Server database also.

Each of these options has benefits and disadvantages but for the purpose of this post I will focus on option 3 (even writeback has disadvantages).  The goal is to allow users to enter targets with the least amount of effort and maintenance and accomplishes this for the most part. 

To get started you need to first add the table, view or even a named query that stores the granularity that targets will be entered in at.  For example, sales are probably recorded at a daily level but budgeting for targets on sales are likely done at a month level.  So the data source view may look something like the screenshot below. 

The highlighted object is a named query (this could be a physical table or view) called TargetCategorySales that stores the proper grain that targets would be entered and a place holder column for the soon to be entered targets.  The other non-highlighted tables are the supporting dimensions and the actual sales.

Next add the TargetCategorySales tables will get added as a measure group to the cube.  After adding the new measure group in the Cube Structure tab in the cube designer make your way over to the dimension usage tab and ensure the relationships are properly defined between dimensions and the new measure group.

Notice here you will see the granularity is different for each measure group.  The actual sales can be tracked all the way down to an individual product and a day, while the targets are only tracked to a category and month.  It’s probably a good idea to go ahead and reprocess now to ensure everything is still working properly

Next, go to the Partitions tab and right click on the target measure group and click Writeback Settings as shown below. 

This brings up a dialog box that will create a new SQL Server table that will store the data changes made by the user in Excel.  Name the table whatever you want then click OK.  This table will not be actually created until the next time you process.  So go ahead and process and you will see the new table in the database you selected in the dialog box earlier.  This table will be empty until the user enters in targets in Excel. 

The user now will open Excel and create a pivot table with the proper granularity, in this case the month and category level.  It’s probably a good idea to create a template for the users that has the appropriate granularity already set in the pivot table.  After creating a proper Excel template as shown below you can enable writeback in Excel 2010 by clicking Enable What-If-Analysis under the PivotTable Options menu.  You’ll also probably want to go ahead and click Automatically Calculate Changes under the same menu, which will apply your changes in Excel as you go. 

Now you’re ready to start adding some targets!  All you have to do is type right on top on the existing numbers in the pivot table.  After typing in your target values go back to the What-If-Analysis menu and click Publish Changes.

With these changes in the data it will now appear in the cube for browsing and comparing to actuals.  You can go back and look at the writeback table to see all of the changes that were recorded as shown below.

Now after a while this writeback table can get very large so you will eventually want to integrate these targets back into the data warehouse using some kind of ETL process.