Ranking strikers by total goals rewards availability as heavily as finishing. Dividing by minutes played corrects that, and introduces a different distortion that has to be handled deliberately.
Totals conflate playing and scoring
A forward who plays every minute of a season will out-score an equally good forward who missed two months, purely through exposure.
Rotation compounds it, since a striker at a club competing on several fronts may be rested regularly while a rival plays continuously.
The total is still a real achievement, because staying fit is part of the job, but it is not a measure of how good the player is at scoring.
Rate measures isolate the scoring
Goals per ninety minutes, or minutes per goal, expresses the same information at a common scale so players with different workloads can be compared.
It immediately surfaces forwards who are highly effective in limited roles and would be invisible in a table sorted by totals.
It also flatters nobody for simply being fit, which makes it a better description of what happens while a player is actually on the pitch.
Small samples are the obvious failure
A player with a few hundred minutes can post a spectacular rate on a handful of goals, and the figure carries almost no predictive weight.
Any honest use of the measure sets a minutes threshold below which a player is simply not ranked, rather than ranking him on noise.
The threshold is a judgement call, and where it is set determines which surprising names appear near the top of a list.
Substitute minutes are not equivalent minutes
Minutes played late in stretched matches produce chances at a higher rate than minutes played in the opening half hour.
A rate built mostly from those minutes describes a favourable context rather than a superior player, and the difference is invisible in the number.
Separating starting minutes from substitute minutes fixes most of this, and it usually moves the flattering names down considerably.
What the measure is genuinely good for
It works well for comparing established forwards across a full season, where both have played enough for the sample to mean something.
It is also the fairest way to assess a striker returning from long injury, since it does not punish him for absence he did not choose.
Used alongside totals rather than instead of them, it separates the question of who scored most from the question of who scores most often.