Our honours class had a fascinating guest lecture today from Thomas Pogge, who at one point suggested that NGOs/charity groups do a lot less good than they could, if only they would get their priorities straight.
Of course, some such groups are merely self-serving bureaucracies who couldn't care less about achieving real good in the world. But let's put them aside and consider more sincere organizations. Pogge briefly mentioned two major flaws that are apparently very common, even in the best of them.
Firstly, they are too reluctant to specialize, or focus their resources where they could do the most good. Many people are apparently under the illusion that "fairness" requires them to spread their aid across a range of geographic areas. But that's silly; a needy person is however needy they are, regardless of where they live. If you could save either 2000 Indian lives or else just 1000 lives spread out across the world, there's really nothing at all to be said for preferring the latter. I think this is an especially pernicious case showing how intrinsic concern for "group welfare" leads us astray. Just because you've already helped a bunch of Indian people, doesn't mean that the next Indian person in need is any less deserving of moral consideration than someone from a different group. NGOs need to be more consistently impartial in this respect, seeking to maximize wellbeing without regard for group affiliations, or any misguided desire to be "egalitarian" in one's aid across groups. (Note: giving equal aid to groups who would benefit less from it is NOT the way to show equal concern for all people.)
The second problem is that they tend to commit the "sunk cost" fallacy. That is, once a project is begun, NGOs are very reluctant to abandon it, even when it is obvious that it is going worse than expected, and that they could do much more good elsewhere. Apparently many NGOs will only withdraw after it becomes undeniable that their efforts are doing significantly more harm than good. But in fact they should be far more flexible on this front. They should withdraw even if their efforts are doing net good, so long as they could do more good by reallocating their resources elsewhere.
Really, someone ought to create a charity organization that is explicitly utilitarian in its priorities. They would do careful research into how best to allocate their resources (including, of course, the question of how much such research itself is worth, for future reference), and then follow through accordingly. Most meta-charity "watchdogs" apparently don't track outcomes at all well. They might report how much money is wasted on administrative overheads, but that's not enough. It isn't enough to know how much money was ultimately spent charitably; we also need to know how effectively it was spent. Hence the need for a more accountable business of beneficence.