A: What is it?
“Implementation” covers the start-to-finish activities associated with any Business Intelligence project. That includes projects of every size, from replacing a single report to building a BI infrastructure from the ground up.
Implementation can be viewed in terms of four typical phases:
- Planning and discovery: Defining the need(s), deciding on solutions, detailing what to do, how to do it, and who will be involved.
- Project preparation: Risk assessment, budgeting, purchasing, staffing, establishing project management, etc.
- Development: Requirements, installation, coding, documentation, testing, etc.
- Deployment: Training, roll-out, support, etc.
The implementation of a BI project can also involve product evaluations, vendor management, project communications, retirement of legacy systems, and many other activities. In short: There are a lot of moving parts.
B: Why does it matter?
By some assessments, the majority of BI projects fail. Some only fail to meet the highest expectations, others crash and burn completely. Among the frequent problems:
- Users don’t like what’s been implemented and refuse to use it.
- What’s been implemented isn’t what was really needed.
- The project goes far over budget.
- The project takes too long to implement, and is obsolete or redundant by the time it rolls out.
Why do these unhappy outcomes occur so often? Top reasons:
- Insufficient planning and/or discovery
- Lack of executive support and/or user education
- Poor change management and/or communication
- Poor data quality and/or slow performance
- Poor design and/or wrong tool(s)
- Inadequate budget and/or unrealistic expectations
These problems can occur with in-house projects (often because employee resources are not up to the required tasks), and also with outsourced projects (often because the external resources don’t understand the business). In short: There are no safe choices or easy answers.
C: What’s next?
These potential pitfalls are not new. In fact, the same issues have been identified and discussed for years. There are many books, workshops, and consultants that promise to explain the best practices of BI implementation—and for the most part, they all offer similar and familiar advice. So why hasn’t the success ratio of BI projects improved?
One reason is that the good advice either isn’t understood or isn’t followed. Many organizations implement best practice “labels” without actually understanding and adopting the practices. Other organizations may follow some best practices and ignore others, which almost never works out well.
But an increasingly important cause of project problems may be that traditional BI implementation is no longer the best approach, at least for some projects and some organizations. Alternative solutions such as agile BI and self-serve or cloud-based BI may offer a better way forward–which means that a new understanding of best practices will be emerging.
