8 ways BI projects are different from IT initiatives

8 ways BI projects are different from IT initiatives

Many organizations, including healthcare, continue to try to approach business intelligence projects like they would approach a project to update their network hardware, according to Analytics8, a BI data warehousing enablement and optimization consulting firm. These BI efforts often are unsuccessful because BI and IT projects are very different. BI and analytics projects require an entirely different approach to ensure success.

Subject matter experts need to be involved

Subject matter experts (SMEs) should be involved during the requirements and testing phases of a project. Today, SMEs are among the primary members of the delivery team involved from day one. There should be more prototyping and even faster project iterations than in past years. Projects move so fast and requirements change so quickly today that 100 percent involvement from business experts is needed throughout the entire project; otherwise the project will drag waiting on their input.

Intensive IT business analyst support is required

Other types of IT projects now are done with closer alignment to the business than previously. There are a lot fewer large-scale, force-it-down-their-throats IT projects now. Businesses will work around IT if that happens, and IT will get a reputation of being unhelpful. It is still a challenge for those in large organizations and in certain industries like healthcare and insurance, because they are more risk-adverse.

Requirements for BI projects can’t be defined in advance

This is the most fundamental difference between BI projects and IT projects. For an IT project, a healthcare organization can completely define its needs for a network upgrade project, or it can almost completely define its needs for an accounting system or a CRM system upfront. That’s not the case with BI and analytics projects—there’s no way to know what the data is going to reveal until executives start asking and answering questions. Today, organizations have much better tools that enable changes throughout a project. Data discovery tools from Qlik, Tableau and others have revolutionized the delivery of these projects. And the big vendors have followed suit with their own tools that enable fast prototyping and data profiling.

A different project management approach is needed

An agile methodology is a requirement for a successful BI or analytics project. This is true across a range of organizations and projects, not just in BI and not just in IT. It’s crucial to focus on fast time-to-value—that means short iterations, which increases confidence that a feature left out now will be incorporated into the next iteration within a few weeks.

The BI solution requires extensive testing

The testing we do is a lot more pervasive throughout a project. We actually have shorter testing phases because we’ve done testing throughout the project. The rapid prototyping we now do allows a better understanding of the data at earlier stages of our projects which allows us (and our clients) to fix things before any dedicated testing phase.

Change management is critical

Users are attached to their current toolset. The BI and analytics pendulum has swung away from neat and orderly “one-version-of-the-truth” data warehouses to a more chaotic world where business users have powerful tools and access to more data. Although almost all organizations will still have, and will still need, a data warehouse or something that serves as the data warehouse function, the need for change management is key because users like the tools they frequently use and don’t want to move to the new tool implemented with the data warehouses.

With powerful and decentralized analytic tools comes a need for centralized data stewardship so the organization doesn’t lose trust in the data.

Existing systems and business processes require tight integration

Healthcare organizations and other companies are less focused on the need to extract data from existing systems without affecting performance because modern operational systems and back-end databases are much better today. However, they should put a lot more emphasis on correcting the data in operational systems and even changing business processes as a result of findings during an analytics project—even during discovery or prototyping activities.

BI is a program, not a project

When organization executives wonder about how long it takes to finish a BI effort, it’s important to note that it’s a process that will never be finished. For an organization to really use data as a strategic asset, it will need to put a long-term program in place.