Get ofn-align-layers. Basically you create a path with 4 points, the first two on the reference layer and the last two on the adjusted layer. Then you call the script (the adjusted layer need to be the active layer) and the script scales/rotates/translates the layer so that the two pairs of point match. The great benefit of this method is that there is one single transform applied, a manual adjustment often implies several transforms so you have lore image degradation due to pixel interpolations.
However, if you are trying to match two hand-held shots, you will likely get a better result with Hugin (which will compensate for the changes in camera axis). But then the script is quick to install and check…