The use case was a dumb one: forecasting of the future operations, in the retail bank application, with the very same accuracy as the historical operations (for instance, the application should forecast that you will buy some cake at Albert Heijn in ten days from now, on a Tuesday morning at 10:12, for an amount of 1.27 Euros). When I tried to transform the use case so that it makes more sense business-wise (for instance, a budget planning exercise, where we still keep the analysis of time series), the interviews were dumb and wanted to stick to the original formulation of the use case.
