For Kant, morally good action conforms to the good will, and is done, on balance, out of regard for duty, i.e., whatever other motivations may exist, the agent finds the motive of duty sufficient for action (and, I presume, sufficient to overcome countervailing motivations should they arise). Duty gets identified through application of the categorical imperative procedure, which involves, inter alia, specification of the conditions in which the action will occur. (And some completely ad hoc generalization of factual conditions, but some other time.) So, for any given act or course of conduct, how does one tell if it is morally good? By assumption (and one hopes nto at all controversial as observation of human conduct), action may have multiple motivations or be the result of a multitude of desires and preferences. Most are irrelevant to the moral quality of the action, which is dependant on whether the act was done with a regard for duty. In other words, all acts are the same from the outside or observation. Moral qualities are entirely private to the agent (and God I would venture). Kant was a Christian believer and seems not to have had serious doubts about the existence of the Christian God. Anyway, his arguments about good will and the like presuppose God. Back to the point -- the ethics is quite private, it is entirely internal. There does not look to be any route to judgment of conduct by human beings.
I find that a telling deficiency. Fatal. It appears that the Kantian ethics cannot guide conduct because it has no public form or content. (Yes, I think Searle is right that Kant was a solopsist.)
I remember teaching this material by keeping the focus on the argument that the regard for duty only needed to be sufficient to motivate action as a hypothetical. One did not need to act against one's desires and inclination. Where they aligned with duty, the agent was a happy actor. That seems to me still the proper way of presenting the material. And the material is still wrong.