KPI on dynamic data population - numbers are wrong

Question

OBIEE 12c

We have a subject area with Share of Market and Share of Segment (Product) measures where the measures are pinned to month level detail; detail level; and market type - which is a tier of the market dimension hierarchy and product type which is a tier in the product dimension hierarchy.

I used these to create some KPI analyses and all was okay until I put dashboard prompts on to restrict various other dimensions, like product / geography - you get the gist...

When the filters are on the measures do not produce the right results.

The measures are based on 'Internal Sum of Sales' for that Market / 'External total Sum of Sales' for that Market - for Share of Market

And parallel idea for share of segment.

My measures are logical measures, based on logical measures, based on physical columns - do I repeat myself, do I repeat my myself

Anyone help me out please?

Teresa Modesta · Answer

I'll try out your ideas and let you know, but it is no small box of wires, so I may be sometime...

Teresa Modesta · Answer

Keep it physical when multiplying, sound advice!! ;-)

Robert Angel · Answer

Hi,

I ran into this once and found the only way around it was to construct parallel logic to the measures in the body of the report and make sure the elements filtered on where in the presentation, but hidden.

This solved my problem, where the measures pinned to dimension hierarchy tiers were consistently wrong even when I was pushing back the logic to the server with the appropriate settings.

I later also discovered 'selection steps' and think this could also have helped at the time, but I had solved my issue by then.

You can still break your problem up by using the function based functionality to replicate your measures.

As a general rule of thumb though it must be physical when you are looking at multiplication or division, logical otherwise.