Somebody has to take the bad shot
Late in a possession, with the clock expiring and a defence set, some shot has to be attempted. Whoever takes it will convert it at a poor rate, because it is a poor shot by construction. That player's efficiency absorbs a cost that keeps everyone else's clean. Rankings that ignore this are ranking roles and calling it talent.
The comparison that would be fair
A useful comparison holds the situation constant: similar shot locations, similar time remaining, similar defensive pressure. That is more work than quoting a percentage, which is precisely why the percentage keeps winning arguments it should not win.
What I still use it for
None of this makes efficiency useless. Tracked over time for the same player in the same role, it is a good signal of form and of whether a change has taken hold. It is a bad signal for cross-player judgement, and an actively misleading one for cross-team judgement. Use it as a thermometer, not a verdict.