Selective thinking

Business Intelligence has a series of complex definitions but they all boil down to a very simple one: BI is about extracting useful business information from data.

Most companies have several conventional database systems – finance, HR, marketing, sales and production. They all collect a mass of data at a truly paralysing level of detail, which is often far too much for most high-level decision makers.

Most finance directors don’t need to know that a sales person, Fred Smith, sold Sally Jones a widget at 11:03am on 23 April 2006. What they need to know is that sales of widgets by the northern sales force are falling by 3% a month. There, in a nutshell, is the difference between data and information – and information is what BI systems are designed to provide.

BI accepts that the existing database systems are doing a great job collecting the data and leaves them free to do just that. However, the useful business information that people need is often based on data held in multiple systems. For example, the sales system may know that Fred sold Sally a widget, while it’s the HR system that knows that Fred is part of the northern sales force.

On the face of it, the easiest solution is to move some data from the sales system to the HR system, but these systems are not designed to accept incoming data. In addition, the format of the data is usually incompatible. It turns out that it’s much easier to create an entirely new system (the data warehouse) which is designed, right from the start, to accept incoming data.

So the first job of a BI system is ETL – to extract the data from the source systems, transform it into a compatible format and load it into a centralised repository called a data warehouse.

Data storage

As a general rule, new data is moved into the data warehouse once a night, so the data warehouse gradually builds into a huge repository of data. This is a centralised store of all the data that can be used for analytical purposes in the organisation. Of course, the analytical needs of departments within an organisation vary hugely, so it’s normal practice to provide a ‘data mart’ for each department. Only the data that each department needs is transferred to its data mart and, within each data mart, the data is restructured so that it can be easily rendered into information.

Very broadly, BI systems come in two versions. You can build an entire system in-house and, given the complexity of the existing database systems within large corporations, many take this route reasoning, quite correctly, that off-the-shelf software may not match their requirements.

There is an alternative. If you type ‘Business Intelligence’ into a search engine you will soon become aware of a new world of specific BI-type products out there. These products lean more towards the off-the-shelf end of the BI market.

Unfortunately, they are often to be found floating in an alphabet soup and you can rapidly find yourself drowning in a sea of seemingly meaningless acronyms such as DSS, EIS, ESS, MSS, EPM and BPM.

Worse, you track down what BPM is and then discover that it actually stands for two things: “Business Process Management” and “Business Performance Management”. When you decide on performance, you find that it’s defined as: “consolidation of data from various sources, querying and analysis of the data, and putting the results into practice.” Which, infuriatingly, sounds exactly the same as the description given here for business intelligence.

Marketing speak

The bad news for unsuspecting finance directors is that BI has become very fashionable and as such has been adopted by the marketers. As we are all aware, it is important for all products to have a USP (unique selling proposition). This is most easily achieved by inventing a new TLA (three letter acronym) and bolting it on to the product. Some of these are genuinely useful because they do help to distinguish specific groups of products (such as customer relationship management).

For many others, I struggle to see any good reason for the acronym at all. Of course, we must remember that marketing is simply that. Many perfectly good, respectable products have suffered the ignominy of acronym marketing, so I am not suggesting that the addition of a TLA is an indication of a poor product. I am suggesting that if you elect to go for a BI solution from this end of the market, buy it because it provides the functionality it needs, not because of the acronym.

Either way, building a BI system can be a relatively complex, costly and time-consuming experience. But the return on investment can be huge as your decision-makers’ decisions are then driven by real information rather than gut-feel. ROIs of 600% are not impossible.

Although BI has been a long time maturing, it has now reached a stage where most companies cannot afford not to adopt it.

Dr Mark Whitehorn is a database and business intelligence expert. He regularly provides BI consultancy to corporate clients and is the author of several books.