Problems in the real world have much more detailed context which can be used to "guess" an answer to then test rationally. Even if it turns out the answer isn't adequate as it stands, and it often isn't, it usually does provide an "advanced base" to find a better answer.

Also real world problems rarely have an accessible "best" answer; you need to find one that is good enough; and you usually need to define "good enough" on the fly as well.