The data world is slowly moving toward agile development life cycles that improve feedback time, which in turn improves experiment frequency and data quality.
Time-consuming upfront design, read waterfall, has been found to not necessarily correlate with increased return on investment. Designing the data structures upfront means that you're likely to miss out on opportunities that you can find by experimentation. Upfront design signals that your organization is most likely unwilling to explore newer technologies and handle some of the risk that accompanies them, and therefore, it can fail to grow. This philosophy of tried and tested runs, in part, counter to being on AWS and everything that it has to offer.
Having some idea of the kind of data that you might need helps to move toward agile development practices, which in turn reduces the requirement to fully design data structures/schemas upfront, making them iterative changes, as more elements/requirements of the application...