I'm with you...
Lets break it down then instead of lumping everything under one umbrella. (2017/18 vs 2016/17)
Missed shots conceded per game 4.65 vs 3.5 (variance +1.15)
Total shots conceded from outside the box per game 4.29 vs 3.68 (variance +0.61)
Total shots on target conceded from outside the box per game 1 vs 1.55 (variance -0.55)
Total shots conceded from inside the box per game 6.32 vs 5.66 (variance +0.66)
Total shots on target conceded from inside the box per game 2.03 vs 2.5 (variance -0.47)
Conceded shots blocked per game 2.68 vs 1.87 (variance +0.81)
These hard facts are mostly within what you consider similar territory 9.2 vs 10.5 (Variance of +1.3), so how can you claim it's better this year when you've specifically stated that it's similar. I'm being deliberately obtuse here because this is what you've set out as your criteria.
Now lets look at the goalkeeping performance 2017/18 vs 2016/17
Saves per game from inside the box 1.58 vs 1.71 (variance -0.13)
Saves per game from outside the box 0.94 vs 1.47 (variance -0.53)
So saves per game according to your criteria is once again similar, in the case of saves from inside the box it is remarkably similar...
Saves as a percentage of shots on target from inside the box 77.77% vs 68.42%
Saves as a percentage of shots on target from outside the box 93.55% vs 94.92%
Saves as a percentage of total shots on target 82.98% vs 78.71%
So armed with the knowledge of these hard facts (because the probability of the xG is flawed) and using simple extrapolation based on ter Stegen's relative performances ceteris paribus then Barcelona will have conceded 20 goals from the on target shots faced after the 31 match days vs 16 currently or conversely Barcelona would have conceded 20 goals after 31 match days last season as opposed to 28.