I don't agree with taking last year's results, evaluate the net additions/subtractions, then determine any incremental improvement. This "all things being equal" approach would be espcecially dangerous with relievers. The only relevant evaluation is projected 2012 Coco vs. projected 2012 Madson.
For example, last year, the Reds made the right call on projecting Arthur Rhodes and went a different direction. This year, I think history will show the same with Cordero. Madson is in the prime of his career and could reasonably be expected to match or beat his last two seasons' performance. Considering the impact on 2012, signing Coco to close could very likely yield below average results, while Madson should bring above average results. That will make a significant difference on the Reds' ability to win baseball games.