Business intelligence involves managed data infrastructure and business analysts who apply tools to selectively extract historical data from silos of databases, transform the data to a consistent format and load data into tables within a repository. An enterprise data warehouse is the source repository of data sets for all the tabular information collected in an organization. Organizations also build data marts, which are subsets of a data warehouse or standalone, purpose-oriented views of a particular set of information.
In either case, analysts tackle business problems by creating questions that query a mart or warehouse to return reports that can be examined in greater detail through the use of equation-based analytical and graphic visualization tools.
Operational business intelligence is a newer development tuned to performance management, which sets objectives and accountabilities for near-term business goals. Like a data mart, near-real time data might be federated or aggregated in an operational data store, with the goal of providing front-line information to caregivers or field workers. Operational BI seeks to establish key performance indicators that are displayed in dashboards and scorecards of current status for executives, transactional managers and field workers.
Any approach to BI first requires good data quality to ensure that information is not redundant across systems, incorrect in its current state or ambiguous across names, addresses, products or other parameters. Data quality often requires manual intervention to correct or reconcile inconsistencies, but can also apply sophisticated matching tools and external data services to automate the data quality process.
Metadata management refers to data about data. Metadata management requires an organizational approach that establishes data dictionaries and consistent definitions of terminology between groups and entities across business and clinical processes.
As BI matures and crosses lines of business, a requirement has emerged to establish data governance, a set of controls that identify who owns data, who has the rights for entering or changing data and a mechanism for managing the process. Data governance requires high executive or board level involvement for policy setting. Those policies are handed to cross-functional teams that identify data stewards, who are the persons best assigned as gatekeepers of organizational policy.