The popular 1D MAS exchange experiment CODEX suffers limitations due to signal loss during the finite recoupling periods, during which the magnetization evolves in the transverse plane. Here, we address the origins and possible improvements of this problem, aimed at (i) an optimization of the signal-to-noise ratio in the experiments, as well as harnessing intermediate-motion induced signal loss for obtaining approximate information on (ii) correlation times and (iii) potential distributions, where the latter are often found in polymeric systems. First, we show that the intensity of the signal is sensitive to the radiofrequency (rf) parameters of the carbon recoupling and proton decoupling, and care must be taken to gain optimal signal intensity. Optimum conditions are found for recoupling pulses being as short as possible for large chemical shift anisotropy (CSA) values, and approaching a ratio of 3 between the nutation frequencies for protonated carbons, calling for an individual adjustment in each case. Second, we demonstrate that the effect of intermediate motions can be studied semi-quantitatively by combining CODEX data with its constant-time modification CONTRA, which allows for a tuning of the signal loss due to intermediate motions. Third, for the case of samples featuring a distribution of correlation times, we propose a procedure to obtain an estimate of the proportion of molecular segments in the sample for which the CODEX data are representative, i.e., which share of segments moves truly in the slow-motion regime. The procedure involves the combination of CODEX data with a cross-polarisation (CP) reference experiment for an estimate of the full sample magnetization; it is demonstrated on the example of semi-crystalline poly(ethylene oxide).