Although some experimental evidence showed that an obstacle placed in front of a door allows making people's evacuations faster, the efficacy of such a solution has been debated for over 15 years. Researchers are split between those who found the obstacle beneficial and those who could not find a significant difference without it. One of the reasons for the several conclusions lies in the variety of the experiments performed so far, both in terms of competitiveness among participants, geometrical configuration and number of participants. In this work, two unique datasets relative to evacuations with/without obstacle and comprising low and high competitiveness are analyzed using state-of-the-art definitions for crowd dynamics. In particular, the so-called congestion level is employed to measure the smoothness of collective motion. Results for extreme conditions show that, on the overall, the obstacle does not reduce density and congestion level and it could rather slightly increase it. From this perspective, the obstacle was found simply shifting the dangerous spots from the area in front of the exit to the regions between the obstacle and the wall. On the other side, it was however confirmed, that the obstacle can stabilize longitudinal crowd waves, thus reducing the risk of trampling, which could be as important (in terms of safety) as improving the evacuation time.