The Florida Panthers had perhaps the best goaltending duo in the NHL during the 2007-08 and 2008-09 seasons, with Tomas Vokoun and Craig Anderson splitting duties. As a result, in nearly double the playing time, Vokoun's even-strength save percentage with the scored tied is actually slightly worse than his backups:
Let's start with the save percentage column. If, over the course of his 191 decisions, Vokoun had posted a save percentage four points better, then Florida would have been approximately 1.5 wins better per season. Starts for backups can be gerrymandered somewhat, so it's not clear that Craig Anderson would retain his save percentage with a bigger workload (and indeed he didn't as the starter in Colorado, dropping from 956 to 945.)
So what about the winning percentage column? Florida was eight wins better per 82 games when Vokoun's backups were in goal than when he was. And why was that?
The Panthers, for whatever reason, had a higher shooting percentage when Vokoun was on the bench than when he was playing. This was true no matter what - even-strength, PP, PK. We've seen many examples on this site and others showing that even the best-shooting forwards have little control over their own team's shooting percentage when they're on the ice - and we can be pretty sure that a goaltender, particularly when he's standing around in his own end on the PP, has no impact on his team's shooting percentage at the other end of the ice. (The year-over-year correlation of 5v5 on-ice shooting percentage for goaltenders is 0.08 - we square that and find that one year's results explain less than 1% of the other year's results.)
If goaltenders controlled their forwards' shooting percentage - or, more specifically, if Craig Anderson did, then why did he allow his teammates to put up a poor shooting percentage in 2007-08 when he was putting up such good numbers himself? Even though his save percentage, even-strength or overall, was so much better than Vokoun's, he put up the same mediocre winning percentage.
Won-Lost records are a funny thing. Sometimes teams get extremely lucky or unlucky without their past performance having any predictive value in the future. If we want to know what Tomas Vokoun's future performance will look like, we should look at his save percentage, not what his teammates did in the offensive zone.