The battle of trying to apply traditional waterfall software development life-cycle (SDLC) methodology and project management to Business Intelligence (BI) has already been fought — and largely lost.
These approaches and best practices, which apply to most other enterprise applications, work well in some cases, as with very well-defined and stable BI capabilities like tax or regulatory reporting. Mission-critical, enterprise-grade BI apps can also have a reasonably long shelf life of a year or more.
But these best practices do not work for the majority of BI strategies, where requirements change much faster than these traditional approaches can support; by the time a traditional BI application development team rolls out what it thought was a well-designed BI application, it's too late. As a result, BI pros need to move beyond earlier-generation BI support organisations to:
- Focus on business outcomes, not just technologies. Earlier-generation BI programs lacked an "outputs first" mentality. Those projects employed bottom-up approaches that focused on the program and technology first, leaving clients without the proper outputs that they needed to manage the business. Organisations should use a top-down approach that defines key performance indicators, metrics, and measures that align with the business strategy. They must first stop and determine the population of information required to manage the business and then address technology and data needs.
- Put business in charge. Anecdotal evidence indicates that there's a direct correlation between business ownership of BI initiatives and success. Yet technology professionals still staff and run most traditional BI organisations — 55% of BI support organisations are staffed mostly with technologists and 67% report to the CIO. As a result, even with strong business executive sponsorship and support, business stakeholders often think of BI as yet another technology cost center, and BI requirements and priorities get lost in translation. The stigma associated with BI as a cost center can be eliminated by moving the BI support organisation under direct or partial control of the business; 27% of Forrester clients already do this, indicating that they're thinking outside the box.
- Drop the technology-centric mentality. The reality is that BI needs to be business-centric — but even the most modern SDLC and project management (PM) methodologies that call for heavy business involvement foster technology centricity. Yes, these methodologies are constantly evolving, but their roots are in the 1980s and 1990s, when we lacked the luxury of BI commodities like semantic data access layers, user-friendly graphical user interfaces (UIs), report-building wizards with hints and prompts, and tight integration with desktop office applications. Few BI pros should spend their time designing BI dashboard layouts. Why? All modern BI tools offer drag-and-drop and search-like UIs where any trained business user can pick the appropriate metrics from a menu, arrange them on a dashboard screen, and share the results with colleagues — or the entire enterprise. More modern UIs that include natural-language processing interfaces require even less training for business users to become self-sufficient.
- Move beyond project silos. BI pros need an agile system if they're to deliver business capabilities fast enough to ensure that BI is truly a strategic investment in decision support rather than a tactical, after-the-fact reporting system. Tactical, project-based BI efforts can lack a longer-term strategic vision and development approach for BI capabilities; many of the decisions made about the requirements, architecture, and development of the BI solution in such efforts inhibit agility, making subsequent development more difficult when the next set of requirements inevitably comes around. BI requires an iterative and flexible development life cycle that recognises one simple fact: Once you answer one business question, 10 other business-critical questions immediately follow.
- Focus on functional reporting capabilities while not ignoring the data. It's irrelevant how flexible and user-friendly BI tools are if the underlying data these tools depend on remains incomplete, untrustworthy, or simply unavailable to end users. But some BI programs continue to ignore — or at least not prioritise — capabilities for data quality, master data management (MDM), and metadata management that provide consistency, trust, and transparency of the enterprise data most critical to delivering business insight. BI support centers (BISCs) must closely coordinate efforts with — and often evangelise and drive — a broader enterprise data governance program. Only with effective data governance can a BISC ensure that quality data is coming in so that BI delivers trusted information to business stakeholders.
These and other best practices will result in business and application development organisations that are flexible enough to support today's BI realities. Otherwise, when business stakeholders can't get what they need in time, they go ahead and build it on their own, using homegrown BI applications based on spreadsheets or desktop database management systems or by procuring departmental or project-based BI capabilities as software-as-a-service.
While this may solve a problem in the short term, it moves the organisation further away from an enterprise-grade BI strategy.
In our latest Build An Agile BI Organisation research report we propose and research various approaches to building an agile BI organisation with multiple current state data points such as
- Whether organisations provide BI and data management suppport by the same or different teams
- Whether organisations provide BI support for client facing vs non client facing analytical apps by the same or different teams
- Whether organisations provide BI and advanced / predictive analytics by the same or different teams
- Whether organisations provide BI and big data analytics by the same or different teams
- How do organisations staff their BI COE BI CC (business vs. technology professionals, virtual vs. physical organisation, and others)
- 27 typical BI roles, their responsibities, what organisation they report to
Posted by Boris Evelson