I have measured digital circuits as being especially susceptible ±2MHz around their clock harmonics, increased by ‘drift’ due to temperature changes and aging.
IEC 61000-4-3 and -6 test over wide frequency ranges by stepping single frequencies in increments of 1% of their previous values, e.g., at 100MHz the step size is 1MHz; and at 6GHz it is 60MHz. Large steps might ignore narrow susceptibilities by ‘stepping over’ them.
IEC 61000-4-3 and -6 use 1kHz sinewave modulation, but 1kHz squarewave creates a much ‘richer’ spectrum, and because of this I’ve detected even very narrow ranges of susceptibility of phase-locked-loops in some very accurate sensing devices, e.g. the figure below.
So: electronics that pass the regular tests could still be especially susceptible at certain frequencies, but it seems most are sufficiently reliable in real life because those frequencies aren’t yet very common.
Critical equipment shouldn’t rely on luck, and MIL STD 461G uses 0.25% steps of the start frequency for the range 1-18GHz. At 6GHz, instead of a step size of 60MHz it uses 2.5MHz, and it uses 1kHz square-wave modulation – making it a much more thorough immunity test than IEC 61000-4-3 and -6, valuable for critical electronics.