A method includes measuring a first performance metric of a network comprising a plurality of virtual network functions (VNFs). The method also includes executing tasks to implement the software change on a first VNF set. The method also includes measuring a second performance metric of the network after at least one of the tasks has been completed and comparing the first performance metric to the second performance metric to determine a recommendation for whether to deploy the software change on the plurality of VNFs. The tasks are based upon a change management workflow created using a graphical model, the graphical model comprising modular building blocks selected from a change management catalog.