Don't improve things you're not asked to

S-CurveImage by 96dpi via FlickrRecently I worked as a BI consultant for a big bank. I was responsible for their metadata warehouse solution.

We would like to integrate our data dictionary with your solution, they told me.

Data dictionary? I asked and they explained.

Oh I see, I replied, you're talking about business nomenclature. And I explained what business nomenclature means in terms of metadata warehouse, the CWM standard etc.

Then there was a presentation where I explained everything once more. I believed my presentation was quite good but I missed a point. I was strict in using "business nomenclature" because hey, I was right, wasn't I?

I understand you but where is our data dictionary? That was the first question.

So don't try to narrow paths that are given. Especially when you're not asked to do it.

Mark Madsen writes about the same issue although his issue is the ideology of bad non-centralized Excel data in BI:

We're facing the incomplete data problem because of another piece of BI ideology: all the data must be centrally managed. This is unrealistic. We can't possibly house every last bit of data. Because of this reality, BI tools like Business Objects added the ability to bring outside data into reports. Other vendors moved the BI processing to the PC.

Our ideology has failed us by setting up a paradox. If we do use these features or tools, then we contribute to our biggest complaint about Excel — manipulation of data outside the centrally integrated view. If we don’t use them then users will continue to circumvent BI tools.

Mark is right, people use and will use Excel. Don't try to convince them they're wrong because they are not. Just take it as a fact and build on it. Put the twisting path to use.

No comments: