Systematic vs random
Systematic errors repeat in a pattern—uncorrected scale, refractive index assumptions, or a bad prism constant. Random noise spreads shots but may average out. Good practice reduces both: calibrate, use redundant observations, and cross-check.
GNSS-specific factors
Sky view, multipath, baseline length, ionospheric activity, and whether you used PPK or RTK all change residuals. Compare rover shots to known control regularly.
Scanning & photogrammetry
For LiDAR and photogrammetry, tie to control and use check surfaces or independent targets. Report in the same units and datum as the contract.