E-Commerce

Don't get burned by a bogus BI implementation

A well-executed business intelligence (BI) application can liberate data trapped in data silos. Check out this advice on how to avoid a BI rollout debacle.


Business intelligence applications (BI) automate processes that gather, analyze, transmit, and report mission-critical information locked in an enterprise's many data stores. The primary objective of BI application suite is to use organizational data assets to allow for fast and accurate decision-making and opportunity recognition. But as with all enterprisewide implementations, complex technical issues invariably crawl out of the woodwork.

If you are “volunteered” to be the technical resource on a committee tasked with implementing a business intelligence (BI) application, here’s some advice to help you avoid some of the pitfalls and make it a smooth rollout.

Start with how the data is stored
Today’s BI system can be viewed as a series of repeating processes. These processes can be organized in a series of steps:
  1. Store—Corporate information will be organized in related "object" groupings that are multidimensional in nature.
  2. Collect—Data must be gathered from all corners of the company in some organized form that allows for meaningful analysis. This retrieval process will have the capability of analyzing and directing access to all necessary data stores.
  3. Analyze—Tools must allow the system to perform comprehensive what-if analysis on the data. Drill-down capability should be present, allowing for detailed scenarios.
  4. Visualize—Along with analytic tools, visualization tools should be available to communicate large volumes of information in meaningful ways. Examples are tools that build graphs, color-code data, draw territory maps, etc.
  5. Plan—Once analysis is complete, decide how you will pursue the information you have acquired.
  6. Act—Communicate or act upon this information to a successful conclusion.

Some of the vendors
  • Cognos—Offers Impromptu and PowerPlay
  • Business Objects—Offers tools by the same name
  • SAP—Offers Business Information Warehouse, Knowledge Management, and Strategic Enterprise Management
  • Computer Associates—Offers Clever Path suite of tools
  • Insightful—Offers StatServer and Splus tools
  • Oracle—Offers Darwin, Discoverer, and Reports, as well as traditional data warehousing and data mart tools.

Some vendors offer specialized applications in one of the process areas listed above, while others offer a complete suite of tools in a comprehensive packaged solution. Organizations implementing a BI solution will be faced with a wealth of choices. The right choice will depend on each enterprise’s specific requirements.


Keys to a successful rollout
The successful implementation of any enterprisewide solution requires careful planning. Outlined below are a number of critical elements that an organization must consider before proceeding.

Commitment
Enterprises pursuing BI solutions must make a serious commitment at senior levels, or the project is doomed to fail before starting. This commitment is not only a matter of money but of resource allocation and authority.

Implementation partners
One of the biggest mistakes organizations can and usually do make is to try to build a BI solution on their own. You must undergo an honest self-evaluation of your own capabilities. Executed correctly, BI projects are massive efforts requiring specialized skills and significant resources. By partnering with experienced consulting groups or vendors, significant cost savings can be appreciated, while at the same time producing better deliverables.

Goals and objectives
Your organization must ask itself, "What do I want my BI system to do for me?" Answering the question will mean hours of meetings with business stakeholders to gain an understanding of what is needed.

Scope
Many BI efforts fail to define scope. Ensure that project requirements include input from all organizational entities. Don’t fall for the low-cost approach of building targeted solutions aimed at providing integration with only some of the enterprises stakeholders.

Infrastructure analysis
While determining what it will take to build your solution, you must review your IT infrastructure to ensure that the proposed solution will leverage current applications.

Support and training
Issues of system support and end-user training are often overlooked, but they should be resolved. Support and training can be supplied in-house or by contracting with external vendors.

Additional or specialized resource needs
Not only do you have to build it but you must also maintain it. This will require acquisition of new technical talent with specialized skills.

Design
Hire a system architect with experience designing enterprisewide BI solutions. An architect can develop an accessible foundation so that corporate information can be rapidly extracted and delivered across the enterprise.

Tools
Acquire appropriate tools for querying and analyzing data, including visualization, communication, Online Application Processing (OLAP), etc. Many of the product suites listed above provide this level of capability.

Planning your implementation
This is stating the obvious, but create a project plan that documents outstanding incidents, schedule and project dependencies, and risks.

Security
Security is both a regulatory and legal issue for data access. In the financial sector, for example, the Gramm-Leach-Bliley Act sets clear requirements for information access and usage. Security should be considered early in the requirements phase. Failure to do so could cause serious financial cost and project goodwill.

Follow-up and fine-tuning
Post-rollout follow-up assessment offers an opportunity for business drivers, project managers, architects, and end users to voice their frustrations and successes. Archive this information when it comes time to upgrade or migrate to another version.

Conclusion
BI systems are major project efforts requiring careful planning and consideration. With the right organizational support, success can be significant. As with all projects, however, there will be many hurdles, both technical and political. External expertise is recommended for the majority of organizations, as the specialized talent necessary for implementation is rarely in-house.

Editor's Picks

Free Newsletters, In your Inbox