Originally posted onJuly 21 2011
This title sounds like a Bible study session but it is not, of course.
I recently had the “opportunity” to be part of the long and painful data matching process in a data warehouse project by a midsize international company. After a long technical phase, where third party extractors have been rewritten from scratch twice to get technically correct results, our relief was jeopardized by the notion that the figures they were used to were slightly different than those coming out from the data warehouse.
This was no surprise, matching is always a two steps process, first get “all the rows” then create the correct business data. Till then, the customer relied on built-in reports in the transaction system, and it was natural to aim at the same figures.
The shivers begun when the users, questioned on how to calculate revenues and quantities, naively replied “We don’t know, we just take what the system says. It can’t be wrong, doesn’t it?”
Things got worse when the brass was informed about “the new system which is all wrong”. They were so used to get canned Excel worksheets and so far from daily operations, that they do not even realized what the problem was. The one directive that was issued was “make it right!”
But the problem was “What is right, exactly?” Nobody was able to tell it but everyone criticized the system for being late and for “containing just rubbish”. We were trapped in a Catch 22 situation.
We had no choice than carefully rebuild those rules by induction, comparing thousands of figures to find a way through. It was a long and exhausting job that surfaced some implicit and not thought-of choices in the area of management accounting. When those rules were finally put on paper to be reviewed by the customer, the task was considered to be “not relevant”. The project, at last, was set back on track to a happy end, but it was one of the toughest jobs I did in my consulting life.
In this case I was particularly unlucky because the usual Business – IT divide came together with unprepared people; but it also stimulates few thoughts on the idea of right and wrong.
While, as I said before, each one has its own definition of measures or dimensions with the same name; sometimes this definition is unknown and imprecise.
It might be because of poor culture, plain ignorance or laziness but, sometimes, numbers are given for granted.
Financial life or death is tied to unknown parameters, or prone to arbitrary mistakes. This is actually like juggling with nitroglycerine. The lack of precise knowledge of how the numbers are formed makes every action a guess. Thinking that life and employment of hundreds of people depends on such and attitude is really astonishing.
The second point is that, no matter what you say, the right number is only the expected one.
In this project, the numbers could be right in several different manners (no, IAS are not the only way to look at numbers and they do not fix all the issues) but the users refused even to consider them if we couldn't replicate the results that they were familiar with. Actually we had to replicate some mistakes to get the “right” figures!
When I was younger, I thought that Business Intelligence was a science as Business Management. The more I grow old, the more it looks like an art.