I think with those the idea is to make the arms so long that the precision of the relative separation matters less, especially since LISA is trying to listen for much longer wavelengths than LIGO. On top of that, they likely just need to get the unpredictable component of the relative acceleration to vary (unpredictable component of the relative jerk?) slower than the time taken for a gravitational wave to pass through.

Probably more a matter of precisely measuring the gravitational fields and accounting for those than of controlling their relative position extremely precisely (besides of course the precision needed to target the lasers across such a large distance).