The Naked CIO: Business misintelligence

Just because business intelligence sounds impressive and has spawned a whole industry doesn't mean companies know what it's really for, says the Naked CIO.

I have just been asked to talk at a conference about what not to do when implementing a business intelligence platform.

Exclusive column: The Naked CIO

See what this CIO really thinks…

The Naked CIO: The V-word

The Naked CIO: Price of panic

The Naked CIO: The skills drain needs fixing

The Naked CIO: Madness in the method

The Naked CIO: Process not bureaucracy

The Naked CIO: Is open source dead?

Given my views on business intelligence, they may not have realised quite what they're letting themselves in for.

Even the term 'business intelligence' confounds me. Apart from the two words often being contradictory, my particular beef is that business intelligence is no more than a construct.

It is a result of a workaround to systems and applications not doing what they were supposed to in the first place.

What's worse, our collective incompetence at making systems and applications work properly has spawned an entire industry of business intelligence software.

Even a piece of software that was designed as a workaround for integration is often not deployed properly. Am I am the only one to see the irony in this?

The first common mistake is that people look to business intelligence to solve a problem, rather than to enhance a system.

Business intelligence software cannot solve illogical problems. It uses a logical set of procedures to process source data from multiple systems to consolidate and analyse this data in a specific way.

Contrary to popular belief, business intelligence is very poor at solving data quality and confidence issues. You must purge these from your source data first to be successful.

Also, many organisations try to consolidate information that has no normalisation. Pieces of data must have a finite relationship with each other to be of any value in analysis.

On top of all that, companies - especially global ones - have varying applications for a particular area of the business, for example CRM. These organisations believe that by implementing business intelligence they will have one system for the purpose of analysis.

They should be wary of this belief. In one sense it is true but only in achieving the lowest common denominator. So with disparate systems this analysis hardly ever drills down to the level that lets business intelligence significantly improve the quality of information available.

Business rules define how data should relate and what type of analysis should be driven by any business intelligence system. It is important to have consistent and agreed business rules for data before moving forward with business intelligence.

It sounds simple. But in many companies these rules are either poorly defined or not defined at all and that can have a huge impact on successful business intelligence implementations.

Finally - and most importantly - never seek to understand your data through the implementation of a business intelligence system.

If you do not understand enterprise data, if you have no concept of an enterprise data model for your organisation, you have no right to spend large sums on any data analysis, data mining or business intelligence tools.