This article is part of an Essential Guide, our editor-selected collection of our best articles, videos and other content on this topic. Explore more in this guide:
2. - Best practices for BI software: Read more in this section
- Follow these five steps for successful on-premises manufacturing BI projects
- Training key to successful BI implementation, experts say
- Be specific when building RFPs for manufacturing BI software
Explore other sections in this guide:
The No. 1 rule of maintaining successful on-premises business intelligence systems, according to experts, is maintaining a partnership between IT and the business from the earliest stage of requirements planning all the way to user training and continuous improvement.
Business involvement, whether through a formal, centralized Business Intelligence Competency Center (BICC) or a more ad hoc, decentralized approach, is critical for developing and promoting business intelligence best practices across the enterprise. In addition, ongoing collaboration between IT and the business ensures there's a commitment to delivering real business value around on-premises business intelligence (BI) projects, as opposed to promoting the technology simply because it's the latest hot trend, experts say.
More on BI implementation
Discover how to build a BI implementation checklist
Find out why BI training is essential
"When you try to do BI with a top-down approach, you end up with something that is too broad, and it limits its utility quite dramatically," said Howard Dresner, principal at Dresner Advisory Services, a consultancy focused on business intelligence and information management. "You end up satisfying some of the needs of everyone, but not all of the needs of anybody."
With that caveat in mind, here are a number of other best practices recommended by business intelligence and data warehouse experts to ensure on-premises BI success:
Facilitate ongoing business involvement. This cannot be overemphasized, according to BI experts. Getting representatives from the functional business units involved as extended members on the BI project team is the only way to prevent gaps between what is delivered and what was expected, experts say. BI's ultimate role is to serve the business, so the team responsible for the BI program needs to elicit regular feedback from the various business constituencies on everything, from which data and metrics are important to what types of charts and graphics are easiest to digest.
Take an architecture approach. Experts say it doesn't matter that many of the newer on-premises BI tools are accessible enough that nonexperts can quickly start building their own dashboards and reports. While self-service BI is definitely an essential component to a successful BI program, experts resoundingly advise building out the initiative as part of a larger information management vision. An enterprise data warehouse doesn't have to be part of that broad architecture. The idea is to have a larger view and an information management roadmap that will discourage the proliferation of independent data marts and foster reuse so the wheel isn't reinvented with every new BI project.
"You don't want to haphazardly patch BI onto an information environment," noted William McKnight, president of McKnight Consulting Group, a Plano, Tex.-based information management consultancy. "You want to make sure you have a vision for information management within the organization and that everything isn't a one-off project."
Attend to data quality. Don't assume the data used in BI projects is in top form, experts say. This is another area where IT and business should work together to identify the proper data types, determine what metadata is important, where the data is located, and what, eventually, it should look like specific to a particular BI campaign. "You run into a lot of risk if you skip this step," McKnight cautioned. "Remember, you're responsible for whatever you dish up, so if it's garbage in, it's garbage out. You need to make sure you have high quality data because decisions are made of this stuff and shared throughout the enterprise."
Start with something of universal value. Don't try to do too much with the initial BI project, experts warn. Pick a visualization method or report that has direct ties to the business and confirm that it's a metric or insight that has relevance across the enterprise. "Make sure the first BI task is something that everyone understands -- if your company makes donuts, make sure it's all about that," said John Myers, senior analyst for business intelligence and data warehousing at Enterprise Management Associates, an industry analysis and consultancy based in Boulder, CO. "You have to get people's attention to get their adoption. Once you have credibility and a following, you can then expand to do some more interesting things."
Don't forget about continuous improvement. Don't get complacent after one or two on-premises BI successes. Once you have users hooked and engaged, they are going to want to see more BI reporting on a regular basis, said Myers. "Add something once a month or once a quarter, because once you have that audience, they are going to get bored if they see the same thing all the time," he said. He added that regular communications around deliverables and ongoing education around BI best practices and goals will help set expectations and whet users' appetite for more BI going forward.
"Don't be complacent because people are going to want more," Myers said. "By continuing to give BI to them, you ensure higher adoption rates, more credibility, and you continue to drive value."
Follow SearchManufacturingERP.com on Twitter: @ManufacturingTT