We model the widespread intuition that more bureaucratic management can lead to less effort and quality in a principal-agent framework. Introducing a procedure aiming at codifying and easing verication of effort can be socially inefficient yet chosen by the principal. In the model with three performance levels, the agent can either shirk, work or go the extra mile, standard eort decreases the occurrence of bad performance, while the extra mile increases the one of the best and is harder to codify. The usual monotone likelihood ratio property naturally breaks down in such a setting. We show that the procedure makes the extra mile both cheaper and less implemented than without the procedure. The introduction of the procedure has implications for organizational design where the principal faces a trade o between incentive gains from task bundling and verication.