I want to build a workbook that can be used to model different scenarios. I have a fairly robust product hierarchy and I want to be able to choose certain skus within the hierarchy, whether at the sku level, the series level, product line level, etc. and allow the user to make adjustments to the chosen skus (by increasing or decreasing the amount sold) in order to see how that affects costs, margin, etc.
I have already built using input fields and variables, but that makes the entire product line have the adjustment, not just the items that I want to adjust.
Inputfields are clearly the way to go for this kind of a simulation. If your adjustments apply to the whole category, than it means that you entered the adjusted value at the category level of detail, and not at the sku (s) level, or perhaps your formulas need to be verified for accuracy... Hard to say what's wrong with very little information...
You can load a dummy field with some dummy values attached to the SKU and declare it as inputfield in the script (see QV Reference Manual). This will allow the operator to change the values for this field for each SKU.
You need to load an Inputfield value associated with the lowerst level of detail that you care to differentiate. In your example, if you want to modify numbers for "server x", that that's the lowest level you need to differentiate, then keep the values at that level. QlikView will aggregate the data up to higher levels.
If you only associate your inputfileds to Categories, then every change in the inputfield value will apply to the whole category (what was your initial problem, if I understood you correctly)