It comes as thousands of tables with codes for column names, split across boundaries that change every five years. Answering a simple question, like who lives within 3 km of a store, means joining tables, carrying old counts onto new boundaries and working out medians by hand.
A retail analyst working out why one store trails the others, or a developer deciding where to build next, needs the figure along with enough of its method to defend it, and building that from the raw tables can take longer than the decision allows.