A good coach will make a bad team competitive, and a good team great. While it is difficult to rate coaches, a good start is to measure how well their team did relative to how well they did the previous season, once regressed to the norm by 35%. With sufficient data we can start to see which of history's coaches might have been the most effective.
It probably comes as no surprise to see Scotty Bowman far, far atop the list, with Ken Hitchcock right behind as the league's most accomplished active coach. It's also interesting to see how much better non-NHLers did as coaches relative to those who had played in the NHL, or how among those who played in the NHL it doesn't seem to make a difference if they made the Hall of Fame, but does seem to matter what position they played (defensemen make better coaches).
Download the attached data and make your own calculations, perhaps restricting the study only to the post-expansion era, or changing the degree of regression (possibly by era) - that variable is in the top righthand corner.
Update: Early in 2015, Josh Weissbock got me coaching data from other leagues, and I'll be adding them here as I process them. Be warned that these leagues are different beasts, as the rosters change from season-to-season as players are promoted and demoted, making each team's winning percentages almost entirely unpredictable from year to year. There's also sometimes some confusion about who the coaches were, and what their records were when there was more than one, so it probably requires some data clean-up.