Z as the dimension of time is the convention, taken from digital signal processing, of letting a third axis carry time so that a measurement can be seen as it changes from one period to the next.
In signal processing, a time–frequency representation captures both how a signal evolves in time and what frequencies it contains. It is used for signals whose character shifts, where a single snapshot would hide the change, and the width of the analysis window trades precision in time against precision in frequency. A waterfall plot makes the idea visible by stacking successive spectra in depth: frequency runs across, amplitude runs up, and time runs along the z axis.
📝John Zdanowski carries the convention from electrical engineering into business data. In 📝Z Axis Labeling, transactional data from business applications is appended with end-of-week, end-of-month and similar period columns, giving each record a position on the time axis so the same metric can be read across periods. The idea pairs with 📝Sampling Frequency, the rate at which those periods are measured.