Context. The bright-rimmed cloud IC 1396N is believed to host one of the few known cases where two bipolar CO outflows driven by young stellar objects collide. The CO outflows are traced by chains of knots of H2 emission, with enhanced emission at the position of the possible collision. Aims. The aim of this work is to use the proper motions of the H2 knots to confirm the collision scenario. Methods. A second-epoch H2 image was obtained, and the proper motions of the knots were determined with a time baseline of ~11 yr. We also performed differential photometry on the images to check the flux variability of the knots. Results. For each outflow (N and S), we classified the knots as pre-collision or post-collision. The axes of the pre-collision knots, the position of the possible collision point, and the axes of the post-collision knots were estimated. The difference between the proper motion direction of the post-collision knots and the position angle from the collision point was also calculated. For some of the knots, we obtained the 3D velocity using the radial velocity derived from H2 spectra. Conclusions. The velocity pattern of the H2 knots in the area of interaction (post-collision knots) shows a deviation from that of the pre-collision knots, consistent with being a consequence of the interaction between the two outflows. This favours the interpretation of the IC 1396N outflows as a true collision between two protostellar jets instead of a projection effect.