Crossing the Chasm II - Departmental to Enterprise BI
In a prior blog, I discussed strategies for crossing the “Chasm” in TDWI’s five-stage BI Maturity Model. The Chasm represents challenges that afflict later-stage BI programs. In my prior blog, I showed how later-stage BI programs must justify existence based on strategic value rather than cost savings efficiency, which is a hallmark of early-stage BI programs.
But perhaps a more serious challenge facing BI programs that want to cross the Chasm is migrating from a departmental BI solution to one with an enterprise scope.
Departmental Solutions. In my experience, some of the most successful BI solutions in terms of user satisfaction are developed at the department level. One reason is because the scope of the project is manageable. A department is small enough so that the technical developers might sit side by side with the business analysts and subject matter experts. They may see themselves as belonging to a single team and may even report to the same boss. The BI project is smaller and probably draws from just one or two sources that are used regularly and well known by people in the department. Moreover, hammering out semantics is straightforward: everyone uses the same language and rules to define things because they all live and work in the same “departmental bubble.”
Chasm Challenges
Politics. Conversely, developing an enterprise BI solution or migrating a successful departmental one to the enterprise level is fraught with peril. The politics alone can torpedo even the most promising applications. The departmental “owners” are afraid corporate IT will suck the lifeblood out of their project by over-architecting the solution. Corporate sponsors jockey for position to benefit from the new application but are reluctant to pony up hard dollars to lay the infrastructure for an enterprise application that benefits the entire company, not just them.
Semantics. In addition, semantics are a nightmare to resolve in an enterprise environment. What sales calls a customer is often different than how finance or marketing defines a customer. Ditto for sale, margin, and any other common term used across the enterprise. In fact, it’s always the most common terms that are the hardest to pin down when crossing departmental boundaries.
Scope. Finally, scope is an issue. Enterprise applications often try to tackle too much and get stalled in the process. The number of data sources rises, the size of the team increases, and the number of meetings, architectural reviews, and sign offs mount inexorably. The team and its processes eventually get in the way of progress. Simply, the BI program has become a bottleneck.
Hybrid Organizational Model
To cross the chasm, BI programs need to maintain the agility of departmental scale efforts with the economies of scale and integration delivered in enterprise solutions. To do this teams must adopt a hybrid organizational model, in which the central or corporate BI team manages the repository of shared data, a common semantic model, and a set of documented standards and best practices for delivering BI solutions.
Departmental IT. The corporate team then works closely with departmental IT staff responsible for developing BI solutions. The departmental developers leverage the enterprise data warehouse, common semantic model, and ETL and BI tools to create subject-specific data marts and standard reports tailored to departmental users. The corporate team oversees their development, ensuring adherence to standards where possible.
Super Users. If no IT staff exists in the department, the BI team cultivates “super users”—technically savvy business users in each department—to be their liaison or “feet on the ground” there. The corporate BI team creates subject-specific data marts and standard reports for each department using input from the Super Users. It then empowers the Super Users to create ad hoc reports around the boundaries of the standard reports but not overlapping them.
BICC. A fully-functioning BI Competency Center adopts this hybrid model where the corporate BI staff work cooperatively with departmental IT professionals and super users The glue that holds this hybrid model together are the standards and best practices that empower departmental experts to build solutions quickly that leverage the enterprise resources.
This approach ensures both agility and alignment, a rare combination seen in most BI programs. The fear of giving control back to departments and proliferating spreadmarts is why many companies are stuck in the chasm. But unless the corporate BI team cedes development work via well documented standards and practices, it becomes a bottleneck to development and causes spreadmarts to grow anyway.
Posted by Wayne Eckerson on November 16, 2009