I quite like
this chart as a measure of closeness (the minimum number of changed votes that would be required to swing the election). It's hardly a perfect measure of how plausible a different result could have been, though. If swings are close to uniform, then it underestimates the closeness of elections where one candidate wins a number of states by tiny margins (e.g 1948). However, in some cases it may overestimate the closeness of states when candidates succeed in heterodox states (e.g. the New Deal coalition). If Mississippi and Massachusetts are both close, for instance, adjusting your platform to try to win the former might cost you votes in the latter. A final problem with these results is it overestimates small states (due to the Senatorial EVs). The 1984 "Mondale wins" map, for instance, has him winning Wyoming, despite the fact Reagan won it by 52 points. This is because the number of votes it would take to swing Wyoming per electoral vote is small. In reality, it would be much harder to swing all those votes in Wyoming than in a larger state.
A simple way to reconcile these might be an average of the uniform swing rank and the minimum votes rank.