Topics: Coronavirus | AI | Data Science | Deep Learning | Machine Learning | Python | R | Statistics

KDnuggets Home » News :: 2013 :: Oct :: Publications :: Iterative-Incremental Approach for BI Implementation ( 13:n25 )

Iterative-Incremental Approach for BI Implementation


Despite big investments, BI projects often fail to deliver, and traditional waterfall methods have proven ineffective. The iterative approach proposed here outlines how to break large projects into more manageable pieces, and uses the idea of a "parking lot" of value-adding features.



Guest post by Jean-Stephane Faubert (InEdge), Oct 8, 2013.

Insurance decision-makers rightly expect to gain a competitive advantage from their Business Intelligence (BI) projects. Despite big investments being made, BI projects often fail to deliver expected outcomes. Traditional 'waterfall' methods have proven ineffective at supporting on-going change. And they often leave business decision-makers patiently waiting for results.

The iterative-incremental (II) approach to insurance BI projects explained in this article has two important structural characteristics. First, and foremost, it works according to the 'divide and conquer' philosophy. Large problems, such as the implementation of a BI platform, get broken down into more manageable pieces, called iterations, which bring incremental value to the business. At each three-month interval, for example, a business can expect to see concrete results flowing from its investments in BI.

Second, the II approach revolves around the idea of a 'parking lot' of value-adding features. The client identifies and prioritizes those features that are most valued. And each iteration focuses on the delivery of the selected subset that will bring the most benefits.

Iteration Zero

The first step in any II project is what is called Iteration Zero. Its purpose is to set the foundation for the iterations that are to follow. This initial stage, that should last no more than four to six weeks, should produce clear conclusions that will drive the rest of the project. It begins by clarifying intentions to make sure all stakeholders agree on the business drivers behind the investment in BI. A clear consensus is essential. Then, initial requirements are gathered in the form of user-stories explaining who needs what kind of information. Once this is done the target solution is designed. But not with the aim of having an ultimate solution that will address all present and future requirements forever.

The solution is expected to evolve from one iteration to the next as the business priorities change and the requirements are either refined or redefined. At the end of the Iteration Zero, a high-level implementation plan is prepared, covering all the iterations required to address the user stories defined so far, as well as a detailed plan for the first iteration that is to follow. At the end of any iteration, the position in accordance to business priorities is reassessed and the high-level plan adapted accordingly.

Iteration One

The first iteration is paramount to the success of any BI project because it shows the business community that value can be delivered in a relatively short time frame. If the first iteration fails to deliver value, decision-makers may lose confidence in the project and seek help elsewhere. The business community may even fall back on old processes.

A winning strategy in BI is to take over existing analytical and reporting functions and to improve upon them. The idea here is for the BI system to support the same needs, but with added value, such as providing ad-hoc querying capabilities to end-users in a sandbox environment of their own. In a nutshell, starting by focusing on current business processes, making sure that what is delivered also exceeds current abilities.

With this in mind, managing priorities is a fundamental challenge: choosing which features to implement next, accepting to leave some in the previously mentioned 'parking lot', etc. And there is also the question of precedent. By starting to implement a solution by focusing on new features the risks of project failure are increased. By keeping the focus on current analytical processes, there is a baseline against which to compare the new solution as well as readily available business knowledge.

Conclusion

In all iterations user experience is the top priority. The solution must be designed with the business community and from their point of view. To ensure success, design activities must focus on components, such as the data model, that users will see and work with on a daily basis.

Decision-makers are looking for tools that will give them a competitive advantage. Traditional approaches to BI projects take too long and fail to deliver adequate ROI. Older methods are ineffective at supporting on-going change.

Reaction to change is swift, direct and effective under the iterative-incremental approach, which yields results quickly. It is a proven methodology for maximizing value in BI projects. It is a field-tested roadmap for success.

For more details about this approach to BI implementation, read InEdge's white paper An Iterative-incremental approach to insurance Business Intelligence Implementation.


Sign Up

By subscribing you accept KDnuggets Privacy Policy