One of the most difficult challenges confronting an organization that wants to increase productivity by adopting Copilot is determining the work breakdown of a software engineer’s time, where the bottlenecks in the software production pipeline currently exist and how AI can reduce these bottlenecks. Agile methodologies contribute to the lack of meaningful data by emphasizing human interaction over documentation and favoring a communal team approach to overall output. Collecting metrics is difficult, costly, and will feel intrusive to technical staff. This chapter aims to provide tangible practices to facilitate the move to wider adoption of other AI technologies.