The Project Race

Originally Posted on 6/1/2009

I've been a BI consultant since the beginning of my career. My experience ranges from small projects with a handful of users to large corporation level systems. I've been working in luxury offices with moquette and boiserie and in country settlements with the smell of the cattle coming in through the windows.

Despite this wide range of different environments, there is one element that harnesses them all: the anxiety to demonstrate that the project delivers actual value.

Better than anxiety, I'd say hurry, when not frenzy.
BI, opposed to transactional systems, is often hard to communicate when it is first introduced in a company environment.
From a commercial point of view it is relatively easy to rely on special effects and eye candy to sell the package.

Costs often require the approval from people whose job is to run the company as a whole and does not have the specific knowledge required to anticipate what a BI project is really like. So it is almost unavoidable to create unjustified expectations. If I told you that the bulk of the job is ETL and data quality, would you have bought all the same? If you had known the effort required to define a common intelligence framework, a common language, common data rules, would you have signed the contract? The answer is often a clear "no". You had kept going with your half baked reporting widgets.

So, what is the perceived way out? It is to think big and start small. Just start delivering a chunk of data, just start working with a group of users and bring them live as soon as you can. Just start delivering value and people will realize that this datawarehouse thing is helpful for sure.

But.

If development has been too fast, if users have not been properly involved, you are likely to commit a suicide that will permeate the entire project, if any remains. If results are unreliable and people did not buy in the project, you are doomed.
The initial stages of a BI project are a true minefield. An ERP project fails at go-live, BI fails when you sit at the table for the second time. What's good is that, once the initial stage is successfully over, everyone can breathe freely, and the following stages are easier to go through.

Of course, we BI professionals do not need this sort of reassurance: we are able to see how the system will look like, to assess advantages and the ROI. It is rather easy if you know what you are talking about. The point is that only few senior level managers have the culture to understand our point. They have no doubt about needing an ERP to manage daily operations, but often do not get at all why they may need this datawarehouse thing. Usually a message like "consolidation will be easier" or "you can have daily figures on your Blackberry" are effective but they subside as they are, giving at best a partial image of the project. Nothing is more embarrassing then a CEO asking the CIO for how the new consolidation system is going, and the CIO answers "what system?".

As usual there is no substitute for deep domain knowledge, understanding and trust. It is no longer cool to admit of knowing nothing about technology, a thorough knowledge of IT related issues should be part of a CEO knowledge no less than financials or marketing. Till that point our lives will keep being much more complex than necessary.