FREE email newsletters 
Five Common Pitfalls to Avoid in BI Projects, Part 1
BI Analyst Take
Information Management Online, January 18, 2007
DM Review welcomes Lyndsay Wise as a new monthly online columnist. She will share her expertise and research materials gained as a BI analyst for the Technology Evaluation Centers in this column that will appear the third week of each month.
Annually, more than 70 percent of IT projects fail to meet expectations or become shelfware. Reasons for failure range from business intelligence (BI) projects not being completed on time and within budget to not being completed at all. Alternatively, BI initiatives may satisfy IT's needs, but may not address those of end users. What should organizations do to avoid the common pitfalls and ensure success? Organizations should benchmark companies who have had successful implementations and learn from their mistakes during the preparation phase so as not to repeat them.
Advertisement
Part 1 of this article explores the pitfalls in defining the business problem and anticipating BI tool use. Part 2 examines challenges in data delivery, training and solution choices. These pitfalls are by no means exhaustive of the list of issues to address; however, they provide an overview of items that should be identified before commencing on BI projects within an organization or business unit.
Defining the Business Problem
Defining the business problem during the preparation phase substantially increases the success rate of projects. Many organizations get caught up with overall initiatives and lose sight of BI's benefits regarding performance management, collaboration, workflow, process improvement, etc. This lack of focus often translates into scope creep and the subsequent expenditure of time, effort and money without positive results.
Business problem definition helps tie an organization's BI project to ROI. By identifying a direct need or pain point of an organization, the cost savings and implementation benefits can be identified up front. Simply implementing an ad hoc query tool, OLAP cube or dashboard to meet a business initiative will not produce ROI unless the BI implementation corresponds to an actual issue. Consequently, implementing BI successfully means going beyond defining how the organization will use BI and extends to identifying the pain points within the organization and how BI will alleviate that pain.
Similarly, the concept of "build it and they will come" does not usually work in BI projects. Many organizations implement an enterprise-wide data warehouse or BI solution to accommodate the needs or use of a specific department or group of users. They assume (or hope) that other end users, irrespective of their requirements, will adopt the existing solution rather than seeking another BI solution. This is unrealistic at best and disastrous at worst in terms of development time and potential waste. End users generally are reluctant to adopt BI solutions unless they see the benefit and have been involved in the process.
Attaining buy-in for BI should include the concept of need within the end-user community. Organizations should involve end users during the process of problem identification. Although the main decision might follow a top-down model, using a bottom-up approach. At the very least, collecting input from end users regarding their requirements can make the difference between the implementation of a tool that works as a value proposition and an implementation that no one adopts. Basically, the decision to implement BI should follow an assessment of organizational requirements as opposed to choosing the BI solution first and attempting to fit its tools to the organization.
Anticipating Use
Best practices indicate that the actual usage of BI solutions typically outstrips initial estimates. Many organizations implement BI to accommodate one department or business initiative within the organization. However, as the BI tool comes online or matures, additional end users begin to leverage it for data, and its adoption expands within the organization. Consequently, BI's actual utilization and, in turn, system requirements are underestimated.
Because platforms often are based on initial implementations, the additional BI users place tremendous stress on the organization's IT environment. As additional applications are built and data warehouses expanded, the original platform may not support the additional activity and queries. As a result, the environment slows or crashes, and end users lose faith in the BI tools as they are unable to use them.
When BI tools cause system problems, the end-user community tends to identify the instability with the BI tool not the general IT environment. The effort required to rectify the situation may not seem worth it because many organizations consider the time required to create a stable environment to be a liability. Alternatively, the consequence of not rectifying the situation results in substantial money and time wasted. By building a platform that anticipates a greater number of BI users, the organization will ensure a stable environment and smooth rollout of the BI solution.
To implement BI successfully, organizations should not neglect the preparation phases of the project. This means identifying the business problem and targeting BI applications, the extent of their use and the appropriate stakeholders. The main driver for a BI project should be the use of BI tools to increase performance as opposed to the implementation of BI based on an organizational initiative that is not tied to identifiable and actionable benefits.
Lyndsay Wise is an industry analyst for business intelligence. For more than seven years, she has assisted clients in business systems analysis, software selection and implementation of enterprise applications. Wise also conducts research of leading technologies, products and vendors in business intelligence, marketing performance management, master data management, and unstructured data. Check out her blog at myblog.wiseanalytics.com.
For more information on related topics, visit the following channels:
