There is no doubt that any agent has certain standing duties which we might call agentic duties, or the duty to learn, the obligation to know. So, I agree that that might be a fruitful line of inquiry. (Just recently made a post that made a bit of use of the idea in a rejoinder to Elizabeth Nowicki and the interpretation of “not good faith”.)

But my (threefold) reply started, first, with the epistemic objection: that we can’t really know what makes others happy except for their basic needs, because people just desire different things. Any should-statement must be possible, and it’s just impossible to know the preferences of 6 billion persons. This doesn’t release the obligation to find a way to provide universal access to basic needs; it compels it, really. The only question is a practical one, having to do with the best means towards achieving that end (essentially, the goal of maximum human economic and social development with minimum of externalities), given the nature of social systems, human nature, and all that.

In other words, it turns out that our agentic duty, our obligation to know, is the obligation to learn more about these social systems and human nature in order to engage in those optimal reforms: for the utilitarian, enaging in a hobby of social science is mandatory, if we’re serious about optimizing utility in the widest scope.

Of course, one may learn about the preferences of everyday strangers, and ask oneself whether or not one may help in achieving their preferences. This demands one more distinction and argument. Preferences can be either alien to a person, or familiar. Which is just to say, I may like something, and you may find it unfathomable, in which case my preference would be alien to you. We need to know the consequences of some alien preference in order to help it along. And if we don’t know the consequences, and we aren’t in a position to know them, then that seems to be the end of the road: for the uncertain must take a backseat to the certain when it comes to action, and since we’ve already established that we’re dealing with strangers, we’ve established that we’ve got little knowledge to go on concerning consequences. And if the preferences aren’t alien, so to speak, then all other things equal, one is obliged to help. But often, all things aren’t equal: the panoply of social facts still demand attention, and often present reasons not to satisfy some preference (introducing, for instance, notions of merit which may be inherant in the meaning of a preference).

Since the above arguments are hardly novel, and since the emphatic manner by which utilitarianism has been rejected shows that they have not made enough headway in convincing others, I made other arguments. The second and third parts of my (quick) threefold rejoinder indicate that, respectively:
a) The pursuit of happiness can be understood, analogously, like the friction of a moving car wheel against the road; too much mass, and the car will slow to a crawl (i.e., release people from the possibility of enjoyment); too little mass, and it will flip around and crash (i.e., enjoy yourself by inhibiting the happiness of others). Without both egoism and helping-behavior, value-directed action can’t succeed in the long-run. (Edit: I see you’ve formulated an explanation that is consistent with mine at here. We seem to disagree most clearly about where the “ceiling” may lie.)
b) In principle, enjoyment isn’t something that can be known or predicted, because (for psychologically functional persons) it is an unpredictable and ad hoc decision. In this sense, and only this sense, we’re not in a position to help.