Of the five you listed, I'd say that Process and IP Scanning and Discovery are the most important planning aspects for scaling IPAM. IPAM has to be baked into your server deployment and change management processes. It should not exist as an isolated process; we all know what happens when you start isolating any part of your infrastructure.
Scanning is fine for the initial population of your IPAM solution. Otherwise, you're stuck importing data from all of those spreadsheets, which means the integrity of your data is suspect from the beginning (that's why you implemented IPAM in the first place, right?). After the initial scan, subsequent scans just validate the existing data, and catch any IPs that were allocated outside of your established processes.
I'd add one aspect: reporting. For organizations that are looking to reduce the complexity of their IP allocation, it's helpful to run reports to see which subnets are most and least utilized. I've heard that certain large civilian agencies have the ability to sell portions of their unused IP space (for agencies who bought a /16 20 years ago, for example). Of course, they first must determine which blocks are available, and IPAM (when properly implemented) can help answer that question.