Ten golden rules of business intelligence
- 24 March, 2010 13:57
Rob Mills is vice president of sales, Asia Pacific, at Information Builders
In the past, companies have spent a lot of money on business intelligence (BI) software, but have not always achieved the expected results.
Evidence of this is usually revealed by users complaining about a lack of data quality and user-friendliness of traditional BI systems and tools, or reports that are incomplete or overloaded with information. This can seriously impact decision-making.
These deficiencies are mainly caused by functional and organisational weaknesses in the implementation of BI projects.
Here is a list of ten golden rules of BI to help organisations steer clear of past mistakes, and make the most of their software investments:
1. Define the functional requirements. KPI comparisons are at the centre of every BI application. Determine which information should be provided by BI applications, when it needs to be made available, and in what format, before doing anything else.
2. Define the user groups. Define who the users of the BI solution are. There are generally three groups of users: general users of reports; the producers and analysts who evaluate the data material; and finally the planners and managers who decide the objectives.
3. Involve the users at an early stage. Create a simple prototype of the solution at a very early stage. In this way, a review can be carried out to ensure that the core requirements are included from the very outset.
4. Get support from management. This is the only way the short and long-term corporate goals can be incorporated.
5. Identify the required KPIs. Operating figures are required for the management of company processes. Define them in consultation with the specialist department. In materials handling and production, for instance, KPIs such as "material costs per component part" or "turnover per employee" are proven variables. This makes it easy to determine whether the objectives have been achieved or not.
6. Ensure data integration and data quality. Data integration is a decisive factor in the success of the BI project. Identify the operational systems in which the required information is available, and how the data should be accessed.
7. Find out which BI tools are already available in the company. Many companies looking at BI start with nothing. In fact, it is likely that isolated applications exist within teams and departments, and as a result a variety of BI tools (the most common being spreadsheets) are already in use. Determine whether existing end-user tools should continue to be used or whether they should be completely replaced. In most cases, standardising on a single BI system is preferable to ensure consistency of information delivery across the enterprise.
8. Decide on the right BI software. Develop a proof-of-concept, to make a final decision on the most suitable software, based on specific brief. This ensures a greater degree of certainty that the software fits the business.
9. Limit project runtime. Here the old rule applies: "anything that takes longer than six months is no longer a project, but a problem." When implementing new BI projects, focus and proceed in clearly definable steps. Projects should be designed so that the first executable and operational results are available after two or three months.
10. A BI project is a constant process. The requirements of companies are constantly changing and the same applies to a BI application. All BI solutions must be continually developed and optimised on a permanent basis. This is the only way they can fulfil business requirements.
Rob Mills is the Australian managing director of business intelligence software company Information Builders