It can be worth it. What matters is how much time it saves your users over the course of using the app vs the time it took to develop it. So, if:

#-of-users * total-time-saved-per-user > time-spent-optimizing

Then it's worth it. You can even multiply by cost of user per time unit and cost of developer per time unit, to see how much money was saved.

Even in cases where its the same person on both sides, it can still work out. There's an xkcd comic about it, even.