Last week I was asked: ‘What is the definition of business intelligence (BI)? Is it about a company’s own operations, that is, information about itself, or an understanding of what are your competitor’s strengths, weaknesses and market positioning?’

The second alternative is probably the most appropriate definition. But the first is how the term is regarded by most organisations, and until recently much of the vendor community.

Collectively we have bounded our understanding and expectations of BI, and if we wish our businesses to become more productive and competitive, with BI technologies as supporting tools, that has to change.

At its most basic, within this bounded view, BI is considered to be the analysis of operational data extracted from transactional systems such as enterprise resource planning (ERP) and customer relationship management (CRM), and reported through numerical or graphical analysis.

The tools from specialists such as Cognos, Business Objects, SAS, and the BI extensions to enterprise applications, such as those from Oracle and SAP, have helped to create and reinforce this bounded perception.

This extends further. There is an implicit expectation that BI tools are specialist and expensive products, requiring significant training for their use, and often requiring specialists to undertake interpretation of the output.

Juxtapose this bounded view of BI with the operational demands faced by organisations.

We have faster business cycles. There is a need for decentralised decision-making, alongside a requirement for the increasing automation of business processes to reduce costs and risks.

Finally, of course, there is the ‘info-glut’ – the rapidly expanding volume of information that needs to be considered to make the best business decisions. I believe in this context that the majority of the BI systems many of us have deployed in the past are no longer fit for purpose.

BI products are prominently regarded and marketed as reporting tools, and it is implicit in the word report that the information they deliver is not current. The very fact that BI tools have to load data from the source systems into a data warehouse before analysis can take place, and that there often needs to be an element of data cleansing before it is analysed, means that the information may not be timely enough to support decision-making, particularly where business processes and systems need rapid amendment.

More responsive organisations require their decision-makers to be close to their operations, be it the call centre, the factory floor or the supermarket.

The output from BI tools not only has to be physically accessible at those locations, something that has been addressed by web-enablement of such tools, but more importantly it needs to be usable at those locations.

BI specialists create standard reports, menus and interfaces to maximise the accessibility of the tools, but this very process creates two further boundaries by constraining the information the decision-maker can see, and the way she/he has to access it.

Decision-making within a business requires the breadth of information described by my questioner last week – not applying it to business rivals, but to our own organisations.

To get the best decision requires complete knowledge. Not all of that information will reside in the enterprise applications whose figures are included in the data warehouse.

Much, potentially the majority, will be in email, documents in the content management system, and even outside the organisation, for example the newswires on the internet.

As the volume of the information external to the data warehouse increases, the ability to have complete knowledge will obviously decrease, and decision-makers are likely to make poorer decisions.

BI tools have the potential to give significant improvement in operations and decision-making to smaller organisations, but many of these are put off by costs and perceived complexity.

For these reasons we need a change in how we regard BI, removing physical and perceptual boundaries. And in a series of evolutionary steps from vendors, we are starting to see this.

The need to increase the breadth of information presented with both speed of delivery and ease of access has not been lost on the BI vendors.

Cognos, SAS, Information Builders and Business Objects have all announced joint initiatives with Google, enabling the familiar search interface to be the access point for the relevant reports, dashboards and so on, alongside other content from within and outside the organisation. With the launch of its Office 2007 products Microsoft will embed many BI functions within the applications themselves.

Open source BI tools such as those from Pentaho and JReport are establishing themselves as credible, lower-cost alternatives to products from the mainstream vendors, offering organisations much of the standard functionality out of the box without large deployment and training overheads.

Consequently these attract attention from smaller firms. We are also starting to see the emergence of software as a service as an alternative means of accessing BI technologies.

But this can only be the start of the change. All the improvements in the technologies from the vendors will have minimal effect if organisations do not recognise that good decision-making becomes harder as volumes of information increase and cycle times reduce.

BI tools need to be embedded in business applications to the point where users do not even recognise they are there.

In other words, BI needs to develop into an operational tool and become part of the integration between the various systems in the organisation.

The nirvana of closed-loop decision support through BI may not be achievable in the short term, but it needs to be the consistent ambitious goal.

BI, as my questioner identified, is about having the holistic view of the organisation. And to improve our businesses we need business intelligence, not BI.

Mike Davis is a senior analyst at Ovum

Further reading: