The SKA will deliver orders of magnitude increases in sensitivity, but most astrometric VLBI observations are limited by systematic errors. In these cases improved sensitivity offers no benefit. The best current solution for improving the accuracy of the VLBI calibration is MV VLBI, where multiple simultaneous observations around the target are used to deduce the corrections required for the line of sight to the target. We have estimated and quantified the applicability of MV from real-world ionospheric studies, making projections into achievable astrometric accuracies. These predict systematic measurement errors, with calibrators separated by several degrees, of $sim$10uas with current VLBI facilities. For closer calibrators, that are in-beam for single dish VLBI facilities, we predict systematic measurement errors of a few uas. This is the ideal combination, where the sensitivity of the SKA will provide the precision and MV will provide the accuracy. Based on these results we suggest that the SKA design should increase the number of VLBI beams it can form from four to as many as ten.