Once the clip is opened in mocha an X-Spline layer is created of the phone and a planar layer is added. This is then tracked forward frame by frame automatically with some manual corrections, this then generates corner pin data to be used by After Effects.
The corner pin data in then copied into After Effects and applied to the app pre-comp, making it match the movement of the phone in the shot footage. Once it has been lined up the Keylight effect is then applied to get rid of a specific green channel, letting the app animation display through the shot footage:
Overall I'm really happy with how the track came out. Although the reflections on the screen made it far harder to track I think it vastly improves the realism of the final shot.