Information Age: News, analysis & insight for IT & business leaders

Although business intelligence (BI) is a well-established market, one that has existed for decades, there is still a plethora of issues to be negotiated when formulating a comprehensive BI strategy. Whether these challenges revolve around how to deal with unstructured data, creating a corporate-wide BI strategy after years working in silos or embedding BI firmly in other business processes such as the supply chain, few organisations feel able to satisfy all of their BI requirements.

Traditionally, says Dr Ben Azvine, the head of the research group in the office of the CTO at BT Group, BI technologies have processed structured information - that which can easily be expressed in tabular form, for example - whereas many knowledge worker tools have dealt with unstructured, harder to classify information, such as emails.

"BI is now incorporating what used to be dealt with by knowledge management," he says. "This fusion between structured and unstructured data is something businesses have to address."

However, others may beg to differ at his definition of knowledge systems. For Steven Goddard, the deputy head of Warwickshire County Council's Strategic Information Management Unit (SIMU) for example, knowledge management structures data in anticipation of answering regular queries. By contrast, BI provides raw data sets with which to answer questions with.

Structured data has its uses for identifying trends, but what is particularly valuable is garnering knowledge from all sections of the business and consolidating it in one central location. Andy Pocock, the IT director at newspaper and magazine distributor Dawson News, enthuses about the benefits of having an extranet which allows Dawson's publishers and distributors to flag up problems when they find them.

Does the need to be underpinned by a strategy? Pocock is not sure. Certainly, any strategy has to be process, rather than technology-driven. Dawson managers simply want to be provided with the information they need, he says, although he confesses never to have sat down with the express intention of formulating a strategy. For him, BI is purely a means to an end.

Divining what that end is can be tricky. Warwick Council's IM unit regularly undertakes an information audit and, says Goddard, "part of the audit asks users 'what do you want to know?' That can be the hardest part." That can be difficult when what they want to know is spread across completely separate databases.

Like many other organisations of its size, retailer John Lewis Partnership is currently evolving from a 'silo-based' reporting structure to a corporate-wide one. And that is highlighting some aspects of BI that are still more difficult to master than others.

"At John Lewis we have certain products, such as newspaper sales, that are called 'black clouds' because they are so difficult to deal with [at an analytical level]," says Sue Maxen, a senior analyst at John Lewis. Those problems don't make it easier for her to deliver the high levels of sophistication her users are calling for: "scorecards that cascade down the whole business".

Those kinds of demands are seen elsewhere in the supply chain. Tesco, Dawson's biggest customer, makes demands for detailed information about impending deliveries which the news distributor is not yet able to satisfy. A real barrier, says Pocock, in that, "it'll cost us, and it'll come straight off our bottom line". Even as BI grows in sophistication, clearly new challenges come sailing over the horizon.