There is much talked about business requirements gathering phase in your BI initiatives. As part of this BR phase, you go to the users and ask for-

• Their information requirements- Reporting as well as analytics needs.
• Look at the current set of reports, dashboards and scorecards.
• Check on any additional potential information needs they might have

This phase will provide you with inputs for modeling and designing your BI platform. The tip here is that this input will not be the single driving factor for you BI plans. The reasons are that information requirements keep on changing. By the time you deliver the data warehouse as per the business requirements, you may realize that half of them have changed. Therefore, one has to model and design your BI environment, so that it can-

• Accommodate wider set of information needs, without changing the model
• Be Flexible in such a way so that you can add new dimensions and measures without re-writing your data-warehouse.
• Respond quickly to the changed information requirements.

This is achieved by following means

Educate your users on the different styles of business requirements

Though unconventional, but it helps to make users think in dimensional model mind-set. Instead of putting requirements only in form of exact reports and MIS (which should still be done), users can articulate their requirements in form of-

• Which dimensions they want to have the 'cuts' and reporting?
• Which measures they want to get reported?
• What are the business hierarchies they want to have their information on?
• What is the level of detail (some-what linked to a hierarchy level); they need to have the information on?

Users need to be made aware that once you have a schema, you can churn out million different reports or analytic views. The idea is to define that schema. Point to note is that we are not doing a holistic modeling here.

Use conformed and foundation dimensions

I think we have beaten this subject of foundation dimensions & measures to death (we can't help it as it is core to your BI success). Even if users are not asking for all attributes of customer (say) dimension, you should ideally have a holistic customer dimension from the day one and populate it as well. This is easy if the entire data is coming from a single customer master.

NOTE- As a separate note, foundation dimensions and measures are important, but if they are coming in the way of your first set of data-marts, you may skip them so to stay with-in the given investment boundaries. For example, if you have to make an ETL routine to pull a holistic customer data (to populate the foundation dimension) from five different sources, you may skip it, as sponsors will not buy this kind of spend for 'future flexibility'. However, if you are creating an enterprise data warehouse, don't skip the foundation dimensions and facts.

Use domain expertise to guess future information requirements.

There are some standard measures and reports which are used in a given industry or function. Even if your users have not asked for it, we would suggest that you should include it in your scope.