You’re going to have to fix stuff.  Along with there being unforeseen defects in the system, there is going to be corrupt data that needs to be corrected.  There are essentially two kinds of corrupt data – the first is wrongly entered by the system, and the second is wrongly entered by the users.

The first will hopefully be kept to a minimum as a result of good requirements (and interpretation), development, and testing.  However, it’s not a perfect world, and in addition to the corrupt data being corrected, there will be defects in the system causing them.  It goes without saying that these defects must be addressed.

The second issue occurs when the users are not used to this new system or new changes to an existing system.  As a result, they may think that they can do something in a different order than the system allows or something of this nature, and mistakes can and will happen.  Instead of looking at this in a negative way (“oh, this user has no idea what he/she is doing and should’ve been trained better”), look at it as an opportunity for improvement.  By seeing what users commonly do wrong in a production system, it’s simple to see where changes can be made to improve the application.  It becomes very easy to notice trends where all users consistently perform the same process erroneously.

By analyzing the data administration that goes on in post-production maintenance, which can often have a less strenuous schedule than typical development, you can take advantage of a great opportunity to discuss further enhancements to the system with the client or possibly even propose a white paper on how to improve ‘the big picture.’  If you've been in an environment where you've done nothing but take suggestions from the client - now's the time to make recommendations of your own!