so as indicated before, there are three possible reasons. First, is the datum shift... if the pins are supposed to be where the end of the error tip is, then is could be a datum shift ... but I don't know what the distance values are so I have no idea whether that is 100m or what. Second, since the differences are similar but not exact this could mean that the coordinates are close to what they are supposed to be, but the numeric precision is insufficient to get a 'true value'. This gives rise to the last possibility that I can think of... your coordinates are perfect but the image is slightly displaced with some distortion and a slight rotation.
So from those (albeit, not a complete list), can you rule out the ones that seem the least plausible and report any supplementary information you may not have thought relevant at the time of reporting the issue