Business intelligence (BI) is a broad category of applications and technologies for gathering, storing, analysing, and providing access to data to help enterprise users make better business decisions. Business Intelligence is a process for increasing the competitive advantage of a business by intelligent use of available data in decision making.

The five key stages of Business Intelligence:

1. Data Sourcing
2. Data Analysis
3. Situation Awareness
4. Risk Assessment
5. Decision Support

BI is a hot topic at the moment but many companies initially struggle to understand what it is, and how it can benefit them. As a starting point, there is a natural tendency for companies to rush out and investigate what BI products are available in the market place. They soon find out the product range is vast, which can leave them with a feeling that BI solutions are complex, disjointed and may not be suited to their needs.

The key to a successful BI strategy is to first stand back and understand the business drivers behind the need for a BI strategy. This is typically a lack of timely information needed to improve organisational performance. Empowering business to be the key owners of a BI strategy results in BI systems that are more likely to meet business objectives, as well as garnering long term business support. At this stage it is often a good idea to approach product vendors to demonstrate real life business solutions in order to generate ideas, and debate. However, the focus should not be on the underlying technologies, but on the value the solutions are offering.
 
Once a BI strategy is in place a business focused BI roadmap can be developed, this in turn drives the creation of a technology roadmap. BI implementations can start off as small solutions, but as long as they fit into an enterprise architectural plan, they will grow and soon become long term assets. This approach allows for a low cost of entry into the BI space.

The information which is required for BI solutions often reside in disparate sources inhibiting data analysis, and the creation of an enterprise wide view. One common approach is the creation of a data warehouse which pulls in data for the various sources, creating a master copy of all the organisational data. This has the advantage of reducing the strain on real time operational systems while the data analysis is being performed. These master copies are periodically refreshed, for most company needs this frequency is more than adequate. If the creation of a data warehouse is unnecessary, then some product vendors, such as Microsoft SQL Analysis Services, also offer the ability to connect your data analysis service directly to multiple vendor databases. This allows for easy access to live information, as well as the ability to quickly plug in new data sources.
 
The ultimate goal is to utilise the analysed information to produce a BI solution that not only allows exceptions and trends to be identified, but also be actioned upon. In order to achieve this, BI information needs to be presented in views that are at the correct granularity for the user in question. The user should be able drill down through the aggregated information to a point where an operational task can be performed in real time. Of course, the user can also start with the operational task's view and work their way up to dashboards. This gives a greater understanding of the wider impact of the task in question. If BI solutions are successfully integrated with existing operational systems this will generate actionable insights across the enterprise, leading to significant overall improvements in the decision making process.

The information requirements need to be identified up front with the help of business users. This will strongly influence which data sources are analysed and which data models to produce. The business doesn’t implicitly think in terms of tables and relationships, but they have a very strong idea of what information they need and how it links together.
 
A good place to start is to first ask business users which new reports and dashboards would they like to have access to which will quickly highlight exceptions in their current operations. This is a reactive approach, but this prevents potential issues from escalating, such as customer service level agreements being breeched.
 
Once the data matures and large amount of historical data has been gathered, then trends and forecasts can be extracted through data mining. A various mixture of customer facing employees, and traditional decision makers need to be interviewed to understand what information they need to anticipate opportunities and threats in a timely manner. 
 
The outcome from both these stages will be a logical data model which may not easily fit into the existing data structures, but will provide an invaluable input when analysing existing data sources and producing data models.
 
The challenges that need to be addressed by technologists are:

• What are the data sources that provide the required information?
• Who owns the data sources and how can the correct authorisation be obtained?
• Is there missing data from the data sources, and how will this be missing data be dealt with?
• How often does the data need to be updated?
• How much historical data should be archived for trend and forecasting purposes?
• Does non required data need to be clensed from the system, to reduce noise?
• How can be link the data together to produce an enterprise view of the data?