These days a big chunk of the case studies I see floating around include a large "research" component, by which I mean things like personas, user journeys, affinity diagrams, etc.
On one hand this is great, and is certainly a more valid approach than what I used to do, which is fire up the old Photoshop and try to produce something that will both make the client happy and get a lot of likes on Dribbble.
But at the same time, is it really worth it spending that much effort trying to learn things from user interviews when you could use that time to ship faster and start gathering data from real-world usage?
And how much of this process is really used to justify the design choices you would have made anyway? If you were working on your own product, would you still go through all those steps or is it something you're mainly doing for the benefit of placating nervous clients?
Now I'm coming at this from the perspective of someone who doesn't practice any of this, and who would love to be proven wrong. So if you've made the switch from just winging it to adopting a more rigorous research process and you've seen a real difference in your output, I would be interested to know about it; and especially know which technique you found the most effective.
Edit: to give some more context, a few examples of cases studies with a research component: