Photo, above, by egor.gribanov, CC-licensed
Many IT companies and consultants define three ways to present business intelligence data.
In an ‘unstructured investigative’ environment you build a database of information that is available to specialist staff. These analysts use the data to answer one-off or infrequent business questions. Analysis techniques include visualisation, data mining and modelling. The quality of analysis often depends on the skills and business knowledge of the individual analyst.
‘Structured investigative’ systems use regular reports, scorecards or dashboards to provide answers to recurring business questions. This information is designed for business and service delivery managers. The information is updated periodically (weekly, monthly, quarterly etc) and enables monitoring of existing processes and systems.
The third category is often referred to as ‘embedded’ intelligence. This involves pushing tailored information directly to managers or other staff. The information is continuously updated so that product quality, service delivery and business performance can be measured against objectives. In a manufacturing or engineering business, embedded intelligence reports might be fed data directly from instruments on the production line or from the field. In white-collar service-delivery environments the link between activity and generation of intelligence data is sometimes less clear.
These delivery channels are closely linked to the type of data they present. In an organisation just starting to develop its business intelligence capability, it may be useful to focus on identifying which recurring business questions could be answered by regular reporting on key performance indicators. This type of structured investigative approach is more likley to deliver early benefits than the more specialised unstructured investigative environment; and it will probably be easier to implement than the embedded intelligence model.
Hat tip: I understand the names of the three categories — unstructured investigative, structured investigative and embedded — are used by IBM. I’m not familiar with that company’s work in this area, so am relying on another analyst for this bit of jargon. The three categories themselves, however, seem to be quite common in the BI literature.
