I never looked at the peer review process in that way ("stamp of approval"). For a journal article, as long as it's not a desk-rejection (because the study is not in scope of the journal or such), it's usually an iterative process involving the authors, the reviewers, and the editor, which results in an often greatly improved article. A key person in that process is often the editor, rather than the reviewers (if they do their job well).

Peer review as such does not provide much of a signal as to how important a paper is or a binary "good paper"/"bad paper" decision; ideally, it's a process that helps maintain a certain standard. Conference proceedings typically don't use this process, but (more or less) ask for a simple yes/no (or a 1-5 rating where everyone knows that it's going to be rejected if you don't give a high enough rating because of a 30% acceptance rate etc.).