Light detection and ranging (lidar) have become a cornerstone technology for geodetic applications, enabling high-resolution mapping. The accuracy of such kinematic laser scans depends critically on precise trajectory estimation, which is often limited by outages in GNSS signal reception, IMU noise, or calibration errors. A promising way to mitigate these limitations is the use of so called lidar-to-lidar correspondences, geometric matches between overlapping part of point clouds acquired at different times and acting as additional, spatial constraints to improve trajectory estimation and sensor calibration. Integrated directly into a rigorous estimation approach such as dynamic network, these correspondences improve the estimation of both the trajectory, and the calibration parameters, significantly reducing sensitivity to possible sensor and calibration errors. We demonstrate their impact in laser scanning in mobile (car) mapping system in urban and/or mountainous terrain where the reception of GNSS is absent over few minutes. The impact of precise correspondences improves the point cloud georeferencing/registration 2 to 10 times with respect to the state-of-the art, depending on the setup.