Continue in 2 seconds

Can you help me with some resources to prepare a standard approach for requirements definition and analysis for BI projects?

  • Anne Marie Smith, Sid Adelman, Clay Rehm, Tom Haughey
  • June 13 2006, 1:00am EDT

Q: Can you help me with some resources to prepare a standard approach for requirements definition and analysis for BI projects?

Sid Adelman's Answer:

Look at Margy Ross's chapter on gathering requirements in The Data Warehouse Lifecycle Toolkit by Ralph Kimball, Laura Reeves, Margy Ross and Warren Thornthwaite.

Clay Rehm's Answer:

The Rational Unified Process (RUP) has a standard approach and can be used on almost any project including BI type projects. You will find many valuable resources by performing a search on the Internet. In addition, there are many fine publications in print.

Anne Marie Smith's Answer:

Data and Process Requirements gathering is part of many methodologies and consists of:

  • Interviewing subject matter experts and relating needs
  • Organizing complex information into understandable subject areas
  • "Translating" technical language into business language and vice versa
  • Ensuring stakeholder involvement at all levels of involvement
  • Drafting clear and concise written documentation for users and technicians
  • Working successfully with multidisciplinary teams

The structured information planning methodology has requirements gathering and documentation as its foundation, and supports the creation and enhancement of robust, scalable and stable applications. The same approach can be followed for business intelligence systems requirements as for transactional systems, recognizing that the types of data and sources and uses will differ. Consulting organizations that specialize in data warehousing and business intelligence projects can assist a company to develop and implement a solid BI requirements process. Also, some companies offer courses in requirements gathering for transactional systems and BI applications.

Tom Haughey's Answer:

The following is the method I use and has been highly effective in the situations I have had to deal with. The following figure contains an example.

  1. Identify business goals, such as reduce attrition.
  2. Get behind the goals and understand what the strategy it takes to achieve the goals.
  3. Degenerate these into the top ten questions or queries for every goal.
  4. Break the queries into the data needed to satisfy them and use these to populate the model.
  5. Identify typical scenarios that the DW must solve for each goal or tactic. A scenario is a typical business situation with alternate conditions and circumstances that the DW must satisfy. For example, using the goals identified below, describe a typical case for resolving policy and customer attrition.

Figure 1 is an example of the first three steps.
This is the general approach but some additional techniques are necessary to make them happen.

  1. Conduct interviews. Talk to business people and power users about their needs and understand them thoroughly.
  2. Standard queries. Do not accept the position, which so many people espouse, that there are no typical queries in a DW. There are. Even if there are none, instead of getting into a dispute over this, ask for some sample queries. There certainly are these.
  3. Conduct some smaller group information sessions. Group sessions give you a more synergistic answer.
  4. This next step is important. Take the scenarios identified above and have the business people compose some representative and detailed descriptions of those scenarios. Take the example used before and have business people or power users describe how they will deal with the problem of attrition. A typical example could be as follows: "We get a list of the customers that just lapsed and their major characteristics (which we determine from our business knowledge). We send these off to data mining who then prioritizes these characteristics, which are returned to us. We then take these characteristics and look for people who have them but have not yet lapsed. They are likely to lapse. We create a list of these and turn it over to a retention unit to call these customers in an effort to retain them."

These scenarios are useful for two things: to find data and to validate the model.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access