I am a huge fan of the GV design spring process. And I agree that the best data you get is from actually getting users to use the product or putting them in as realistic situations as possible to see behavior.

I think the problem with user research is not the research itself, but how much faith people put into it and how deliverable-oriented it has become. In most cases, humans are too complicated to understand what they would do in a given situation, especially when it comes to buying behaviors and behavioral change.

That said, I think some degree of user research ahead of time is useful simply to get a variety of perspectives and to gut check your assumptions about what the problem is or what complexities exist in the problem space. This can be done quickly and without putting too much emphasis on deliverables like personas and journey maps (most of the time, you need to talk to a lot of people anyway to make any sort of accurate persona.. so most personas these days aren’t even reliable because they are based on a tiny sample… I am guilty of this, too).