No Crystal Ball
There is no crystal ball for predicting the future data analysis needs of the enterprise. The tendency to compromise and build the enterprise data model based solely on current needs is much greater for those building denormalized enterprise database schemas than for those building fully normalized schemas.
Even when aware of the compromises being made, individuals undertaking a denormalized implementation of the logical model tend to discount the possibilities of those compromises working against future growth of the system because they believe they have already anticipated all future analytical opportunities.
However, as business discovery evolves, it inevitably becomes clear that those initial projections do not fully support the requirements and vision of the business. For such an enterprise database design, it is inevitable.