Sample preparation is too often still an issue in EPR spectroscopy. Hence, focussing on X-band with its rather manageable dimensions is a good first bet. Furthermore, higher fields and frequencies often come with reduced robustness regarding data acquisition and overall stability of the setup, as do benchtop spectrometers. Cooling, while often necessary, adds a further level of complication that needs to be properly justified.
Robustness has several aspects. First, it means that an experiment should always yield identical results when repeated on the same sample using the same setup. This is particularly important for quantitative analysis. Second, robust measurements are straightforward to perform, regardless of their internal complexity, and reliably yield results. This is notwithstanding recording a standard in case a sample does not give an EPR signal to ensure that it is really the sample and not some problem with the setup or operator.
A third level of robustness is closely related to reliable analysis. The way EPR spectra are analysed needs to follow established and proven protocols and use robust fitting strategies for extracting simulation parameters from the data. This means that most of the complexity of data acquisition and analysis that can be automated should be handled by software that is at the same time fully modular, allowing to change every detail, and user-friendly and robust with regard to its operation.