There is good coherence between the ISI stage 2 signals and the cavity length.  I used the nominal calibration for the GS13s to nm (3 poles at zero, 2 zeros at 1Hz Q ~ 0.7, 1 nm / (2 pi f) above 1Hz), and it appears to match the VCO calibration for the cavity length quite well below the SUS resonances.

I also include a couple of driven measurements.  The current motion of the optics is at the limit of our ability to keep the lock (due to alignment flucutuations and dominated by ITMY), so it is hard to drive hard enough to get a good measurement without breaking the lock, but I was able to get some coherence.  I wonder if the drop-off below 200mHz in the driven measuements is due to some tilt-horizontal coupling in the drive, or possibly due to pickup in the GS13s.  The suspension resonances seem to amplify the motion by a reasonable amount, given that we can't expect to achieve very low Qs with the damping all at the top stage.