The action is under consideration because it causes some benefit, worth $A.
The action may cause some public-relations fallout, which in turn will cause some amount of lost business (or analogous malus). The total losses caused by public-relations effects are a continuous random variable; and a complicated one, because the causality chain between action and the final losses involves multiple random events.
But, it has some expected value, $B > $0. (Disregard outliers, so as to better approximate the median outcome rather than the mean outcome.)
Finally, the action will inevitably cost some resources to implement, worth a total of $C.
(Again, all three quantities $A, $B and $C are nonnegative.)
We can then evaluate whether the action may be taken. That question is equivalent to the following:
"Is it legal? And if so, then does it hold that $A > $B + $C?"
(In the larger context, then, if multiple actions may be taken, then choosing which single one should be taken is a matter of maximizing that same inequality rather than truth-testing it.)
* * * * *
Why should we use any decision-making process other than the above?
Equivalently: why should we take into account any further "morality", beyond merely legality plus long-term expected-value for oneself?