News Stay informed about the latest enterprise technology news and product updates.

Eight steps to BI success

The Data Warehousing Institute has a suggested program for reaching business intelligence (BI) goals. It involves taking eight careful steps, and lowering overall expectations.

When it comes to selecting business intelligence (BI) tool sets, even the best option will almost always require a buyer to compromise, according to Wayne Eckerson, director of research at The Data Warehousing Institute (TDWI).

"Everyone wants one tool to satisfy all users," said Eckerson, speaking at the TDWI 2003 World Conference last week. "That's like the Holy Grail of BI."

Julie Mae Longgood, vice president and manager of internal support at Banner Bank, a midsized regional bank in Walla Walla, Wash., arrived at the conference looking for expert advice on selecting a BI vendor. She is part of an IT team that is implementing a BI project to manage sales tracking for new accounts, as well as for branch management.

The Banner Bank team has narrowed the field to three vendors, she said, but the decision remains difficult. "We need an end-to-end solution," she said, "but what we are finding is that vendors strong in ETL (extraction, transformation and loading) are not necessarily strong in BI."

That's a common problem BI shoppers face, Eckerson said. So how can you find the best BI fit for your enterprise? Eckerson and conference co-presenter Cindi Howsen, an independent consultant and author of Business Objects: The Complete Reference, suggested these steps toward success.

Form a selection committee. Include a cross section of people who will eventually interact with the product. These should be IT users who will deploy and support the product, and understand the business requirements.

Define user groups. Not everyone will be equally satisfied, so Howson advises that you prioritize your user groups by determining who needs to be kept the most satisfied and who will deliver the most value as a result of using the product.

Define information requirements. What sort of information do you need most? Different tool sets analyze and present information in different ways.

Rank your selection criteria. Determine priorities by answering questions such as "Do the company's financials matter?" and "What is our data warehouse structure like?"

Create a vendor short list. Examine vendor market strategies, technology architecture, reporting capabilities, OLAP strategies, and ease of administration and pricing. Then compare those notes to your priorities. This is no simple step. Aside from pricing, the vendors may have little else in common.

Get scripted demos. Narrow your list to three vendors for this step.

Based on the demos, cut one vendor from your list. This is a fairly easy step, but the next one isn't.

Conduct test runs. At this point, you should be choosing between two vendors. Now the work is very time consuming. You have to get end users to try out the competing systems. Those tests should determine the winner.

Even following those guidelines closely does not ensure easy decisions in the end, as Longgood knows too well. "The initial users of the system need only simple analyses and want a simple interface, but those needs won't be the same down the road," she said. "It will definitely be a compromise."


Check out a Featured Topic on Analytics.

Check out our Best Web Links Data warehousing and business intelligence.

To provide your feedback on this article, contact Sara Cushman.

Dig Deeper on SQL Server Business Intelligence Strategies

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.