GPS campaign measurements are frequently used in order to determine geophysical phenomena such as tectonic motion, fault zones, landslides, and volcanoes. When observation duration is shorter, the accuracy of coordinates are degraded and the accuracy of point velocities are affected. The accuracies of the geodetic site velocities from a global network of International GNSS Service (IGS) stations were previously investigated using only PPP. In this study, we extend which site velocities will also be assessed, including fundamental relative positioning. PPP-derived results will also be evaluated to see the effect of reprocessed JPL products, single-receiver ambiguity resolution, repeating survey campaigns minimum 3 days at the site, and eliminating noisier solutions prior to the year 2000. To create synthetic GPS campaigns, 18 globally distributed, continuously operating IGS stations were chosen. GPS data were processed comparatively using GAMIT/GLOBK v10.6 and GIPSY-OASIS II v6.3. The data of synthetic campaign GPS time series were processed using a regression model accounting for the linear and seasonal variation of the ground motion. Once the velocities derived from 24 h sessions were accepted as the truth, the results from sub-sessions were compared with the results of 24 h and hypothesis testing was applied for the significance of the differences. The major outcome of this study is that on global scales (i.e. over long distances) with short observation sessions, the fundamental relative positioning produces results similar to PPP. The reliability of the velocity estimation for GPS horizontal baseline components has now been improved to about 85% of the average for observation durations of 12 h.