From your link:
IS GIA AN ERROR IN GRACE DATA?
No, it is not an error. It is a signal of great scientific interest in itself. GRACE observations, in particular when combined with GPS measurements of vertical surface deformation, have provided new and more accurate estimates of GIA models, and have led to refinements of ice-load histories. For studies of contemporary surface mass changes, the GIA signal must be removed from the GRACE observations. This is particularly important for estimates of Antarctic ice mass changes. The GIA corrections add some uncertainty for surface mass trends over the GRACE period; a canonical (and rather heuristic) uncertainty range of 20% is often assumed for GIA models.
In short:
The error is in GIA, not in GRACE data. For short-term mass changes (annual) GIA corrections are unnecessary. Remove GIA and the 20% error/uncertainty range is eliminated.