A common issue I encounter on projects is that QlikView has been sold to the users as the cure to all of their BI problems in one solution. Subsequently the project brief I receive is this sprawling morass of requirements, data sources and user groups. Not good.
with new QlikView implementations I always prefer to start with a relatively simple application. Preferably single source and single fact. This may not be the solution the user ultimately wants, but it is what I would like them to start with. QlikView is not a natural tool for new usersbrought up on their daily dose of Excel, so you must guide them onto the learning curve of using QlikView and what it can and cannot do. A simple starter application is the best way to approach this. It can be functionally rich, so long as the data model is simple. It is that data that is the crucial bit. Simple data means (hopefully) trusted, reconcilable data. This in turn makes QlikView appear trusted and reconcilable. If the first application delivered is based on complex data that cant be reconciled, where can the trust come from?
You may need to rein in your user from racing to complex data too fast, but you will reap benefits in the future as Qlik builds a solid reputation as users climb the learning curve.