Context. Magnetic clouds (MCs) are observed in situ by spacecrafts. The rotation of their magnetic field is typically interpreted as the crossing of a twisted magnetic flux tube, or flux rope, which was launched from the solar corona. Aims. The detailed magnetic measurements across MCs allow to infer the flux rope characteristics. Still the precise spatial distribution of the magnetic twist is challenging, so debated. Methods. In order to improve the robustness of the results, we perform a superposed epoch analysis (SEA) of a set of well observed MCs at 1 au. While previous work was done using the MC central time, now we use the result of a fitted flux rope model to select the time of the closest approach to the flux rope axis. This implies a precise separation of the in-and outbound regions and to coherently phase the observed signals. We also search for and minimise the possible biases such as magnetic asymmetry and a finite impact parameter. Results. We apply the SEA to derive the median profiles both for the flux rope remaining when cross by the spacecraft and to recover the one present before erosion. In particular, the median azimuthal B component is nearly a linear function of the radius. More generally, the results confirm our previous ones realised without such deep analysis. The twist profile is nearly uniform in the flux rope core, with a steep increase at the border of the flux rope and with similar profiles in the in-/outbound. The main difference with our previous study is a largest twist by ∼ 20%.