First of all you don't have to use the Complex FFT, the Real FFT would work fine because your input time domain data are real-data. Both Real and Complex FFT VIs return a complex spectrum output.
The complex spectrum data represent the magnitude and the phase of your signal at the different frequencies. So if you use the 'complex to polar' primitive in your Numeric>>Complex palette you can convert your complex spectrum to magnitude and phase (radians) signals. Next you need to scale your magnitude for 'half the length of your FFT' (that is divide by N/2) in order to get the correct 'amplitude' of the different tones. Finally you need to correct your value at 'dc' with a SQRT(2) factor ...
When you are done with all this and optionally also have corrected
for the noise bandwidth of your window, the unit of your magnitude will be the same as your input signal (m/s^2 or g I assume).
So instead of all this I would recommend you to use the high-level VIs that will take care off all the needed conversions and scaling for you. If you build a Waveform using the 'Build Waveform' primitive, you can then pass your signal to one of the fighr-level Power Spectrum, FFT (Mag-Phase) etc... VIs. Check the Analyze>>Waveform Measurements palette.