If salaries are a poor measure of worth, reviews at work can loom like a public examination of your soul. I have seen various tactics where the two discussions are split. The 'number' is given first so that an open discussion can be had. It is given last because the discussion is more important than the number. It is given at a wildly different time to the chat so that it doesn't dominate either way. It tends to dominate since it is such a clear signal of what the bottom line impression is.
I agree with the principle that if you discover anything that comes as a big surprise in an annual review, the relationship is dysfunctional. If a manager is relying on sitting down with a bunch of data to examine how someone has performed, the relationship is dysfunctional. The only form of feedback that I think we take kindly to is from someone we trust. Trust comes from a belief that someone has a good understanding of what you are doing, what you are trying to achieve and why you are doing what you are doing. Many of these things are ridiculously hard to communicate. The person who has best articulated this for me is Virginia Postrel. Which is why I call the heart of the problem the Postrel Problem.
The last century has seen the magic of the industrial revolution. Through division of skills and the introduction of machines, productivity has exploded and much of the world has been ripped from poverty. The last three decades in China are nothing short of miraculous. Many of the gains come from cutting out inefficiencies and standardising. If things are standardised they are easily communicable. Once you can do something well, you can roll it out. You can use a cookie cutter approach. Decisions can slowly but surely be made further and further away from the customer.
Except they can't. At some point we get beyond the stuff we have to do, and to the stuff we want to do. Jobs are not just jobs. They are an expression of personality. They give meaning. They are a source of relationships. They start conversations. This is the kind of stuff that can't be communicated through data or even written feedback. Sit down and try write just 100 words on the 10 most important people in your life and why that is the case. It is hard! It is ridiculously hard.
A lot of our knowledge is tacit. We don't even understand that we understand it. It comes through time, nuance, relationships, empathy, connecting and a web that can't be expressed except through experience. That is why Postrel argues convincingly that decisions should be pushed down not up. The role of a manager in a world they don't understand shouldn't be to try. The role should be to remove obstacles. To ask how they can stop getting in the way and how they can provide resources and perspective. Or they should just roll up their sleeves and work with people.
The reason reviews often end up being awful is because managers are typically busy. They don't work with people, people work for them. The best feedback comes from people you work with. It doesn't come in an annual review. It comes over time. It doesn't always come through words. And it certainly doesn't come in the form of a number.