Business Intelligence and Data Warehousing 
Consultant - specialising in COGNOS solutions 
   
Requirements Gathering

It needs to be remembered that business requirements are the reason you wanted to build a data warehouse in the first place. These requirements will affect what goes into the data warehouse, its organisation and how often it is refreshed.

Hopefully you’ll already have some idea of the project scope before you get round to requirements gathering. Someone must be sponsoring the project, and to get any funding at all must have defined some boundaries around it. From here we can begin to ‘flesh out’ the project and discover what is, and isn’t possible.

There are various techniques involved in requirements gathering:

- interviewing the user community to find out what their needs and expectations are. Often this involves translating their business needs into data needs IT departments can deal with

- interviewing the IT community to (start to) find out what is likely to be involved in gathering the information required by the user community. Is the data available at all? Often the data required is available, but not in the organisation’s main operational systems. Is the data reliable?

- Organising meetings involving reporesentatives of the users and the IT community

What to gather. In building a data warehouse we are aiming to collect all the information we are going to need for reporting purposes. Sometimes users know exactly what they want to see in their reports – often they don’t, so we need to find out what the organisation’s success metrics are. We need to know how they are getting these at present (if they are at all) before we can make a judgement on whether the new solution is better than the existing one.

Scheduled or ad-hoc reporting? So far we have mostly talked about data requirements. But we also need to look at reporting requirements. Usually some reports will be ‘canned’ – run by a scheduler and perhaps delivered by email. But a lot can also be gained from ‘ad-hoc’ reporting, analysing data in a drill-up/drill-down manner or data mining. These considerations will need a different approach to scheduled reporting and will need to be taken into account during the requirements-gathering exercise.