That does not make Germany look any better but I find the "percentage on time" not very useful compared to the "years of delay" metric. And arguable a average/median delay per train would be better? Also some delay volatility data would be interesting.

If you look at France for example, 80% of trains are not punctual but the "total delays" is actually on the low range, France being on the large side with lots of lines, I would say that it shows that the delays (20% of the time) are actual shorts.