|
This is purely a situational question.
An ideal combination for a Data-Mart project will be a higher importance business theme and a single (or few) and robust source systems. This combination will provide a high criticality and high feasibility story. However, if the situation is not ideal and you have to make a choice, you should select high-importance business theme over the 'simplicity' of implementation. In that scenario, you can improve upon the level of feasibility, by:
- Reducing the scope of business theme: Check on all sub-components of business theme. For example- a business theme of 'maximizing sales profitability', will include flavors of sales revenue management and sales cost management and sales compensation management .. As some of the sales cost elements come from relatively 'loose' field systems (sales process management cost, training cost, hiring cost..), you may keep them out of the scope in the first implementation.
- Be astute on the level of acceptable data quality vis-a-vis complexity of ETL: For example- Say, if by allowing 3-5% of data in-accuracy, you are able to reduce your ETL by 50%, you may let it pass. This allowance will happen if you feel that this differential will not impact the decisions you will take on this data and also you can compensate for this in-accuracy by doing some smart extrapolations.
|
