Managers are trying to manage and have to predict results in order to get resources to be able to achieve them. It is a game. And there are several methods (or strategies if you want) to win the game. At certain probability at least. Maybe. Maybe not.
Heavy-weight manager Jack Welch sports the "If you can't measure, you can't manage it" formula.
Metrics, however, are just tools not the end in itself. In spite of that "toolism" is so catchy: the millennium dot-com business model revolution/bubble metric was HITS. That now means How Idiots Track Success. What you measure is what you get. So replacing randomness by bad direction you may run into a bee trap which is of no threat to a fly.
And the problem of course is that we often have to manage quality (e.g. loyalty, happiness or brand value) but metrics may only be a quantitative proxy to it. The concept of managing through metrics is further put in perspective by Theory of Constraint - which postulates that either local optimums (implied by many metrics) OR global optimum (implied by one corner stone companywide metric) is attainable.
To make long story short: metrics are a useful tool. Nothing less, nothing more.
Bibliography & Inspiration:
Goldratt, E.M. & Cox, J. The Goal.
Mintzberg, H., Ahlstrand, B. & Lampel, J., 1998. Strategy Safari: The Complete Guide Through the Wilds of Strategic Management. Financial Times / Prentice Hall.
Rock, D., 2007. Quiet Leadership. New York: HarperCollins Publishers.
Welch, J. (with Welch, S.), 2005. Winning. London: HarperCollinsPublishers.
WebExpo Prague 2010 conference. http://webexpo.net/