Business intelligence (BI) has evolved significantly in recent years - development tools are more sophisticated, business has become more refined, senior stakeholder sponsorship has increased, it goes on.…
With these changes have come opportunities for transformation, hence the current trend in assessing new delivery models. But those of us in IT should be wary. Before discussing new alternatives, it’s important to realise there is not a single approach that will fit all BI needs, despite what anyone is telling you. Moreover, “classical” delivery models are the outcome of years of experience from experts.
If this is the case, why not pick and choose, adapt proven concepts, to create the most adequate approach for a given BI project?
Let’s assume we have an existing data warehouse in place, alongside rapidly growing business needs and split responsibilities of delivery between various (internal & external) parties at different stages. In this example any BI project manager would face a number of key challenges: the urgency of business need, a series of risks based on the lack of knowledge of the new data that is about to be acquired, the risk of a possible gap between business wants and solution delivered and last but not least, the high governance effort needed between delivery steps.
If faced with this situation, why not take steps from the V and spiral model, RAD and iterative delivery to create a Hybrid Delivery Model?
The project could start development in parallel with design by acquiring and profiling source data, iteratively adding data transformations based on business needs.
The combination of real data limitations/opportunities and early business involvement should result in strong requirements, validated design assumptions and reduce risk. While the Proof of Concept is kept alive for the business to use and make decisions, the code would be brought up to Production standard following a more “classical” delivery approach.
The use of reusable components during the Proof of Concept would obviously reduce the effort of productionisation. Based on the specific situation of this project, the cost of the Proof of Concept activities would be balanced by the reduction of risks (hence less friction when transitioning effort from one party to the other) and the early access to data by the business for exploitation.
BI has reached a maturity at which project managers have the opportunity to be creative about possible approaches based on their particular environment, so let’s add a BI flavour to the “classical” delivery models.