How Not to Measure Business Intelligence Performance

I’d like to paraphrase one anecdote.  

Long time ago a company introduced invoice verification performance measurement process. Each administrator was paid according to number of processed invoices. Soon after the process introduction, administrators came to an idea to split invoices into smaller invoices with smaller orders. Well, pretty soon introduced performance measurement processed showed that everybody is working hard and processes many invoices.

Huh, what to say… After introduction of improvement system additional time was spent and more costs were created. Great job!

Something similar can be aligned to advice of how not to measure business intelligence performance. Do not count jobs and tasks per tables or lists, with semaphore lights, comments, deadlines and etc.

These lists can be easily transformed into many smaller tasks. Lists can mislead for example it might look that nothing is happening while there are happening solutions on instances that are not mentioned (not on the list) but do make significant basis of complete reporting later. If there is no true and deep knowledge about what is going on in business intelligence processes it is easy to misinterpret value of performance on business intelligence solutions.