Conductive domain walls (CDWs) in the uniaxial ferroelectric lithium niobate (LiNbO 3, LN) have attracted much interest as potential elements in 2D nanoelectronics due to their orders-of-magnitude larger electronic AC and DC conductivities as compared to the host material. On the way toward generating standardized CDWs into z-cut bulk LN crystals with controllable geometry and electrical properties, we have encountered setbacks recently: Although the first preparation step, i.e., the established UV-light-assisted liquid-electrode poling, reliably creates fully penetrating hexagonal domains with the domain walls (DWs) being aligned almost parallel to the polarization axis, the second step in the DW “conductivity-enhancement” process through post-growth voltage ramping, resulted in randomly shaped DWs as reflected in their different current–voltage (I–V) characteristics even after having applied the same process parameters. To clarify this phenomenon, we present here an in situ and time-resolved second-harmonic-generation microscopy investigation of DW samples of different sizes, monitoring the DW evolution during that critical voltage ramping, which allowed us to reconstruct the 3D DW shapes both prior to and after the enhancement process. As a result, we are able to map the temporal changes of the local DW inclination angle α and to quantify the DW velocity. As a consequence, we need to re-assess and re-think the origin of the DW conductivity (DWC) in LN: The hitherto assumed simple connection between α and the DWC cannot be generalized since point defects accumulating along DWs act as extra sources for charge carrier trapping/release, significantly contributing to the DW current.