To provide updates since the last time I posted here.   I couldn't be happier with the changes that we did.   Below are a few key items that are helping us move forward as an IT organization based on walking away from Categorization and moving to Service Offerings:

1. We now have very specific data for what services are generating the workload which is helpful to trend volume and evaluate staffing levels
2. We now have very specific escalation paths for all of our service offerings, making resolution on priority issues happen quicker and making it very clear on how to escalate vs. tribal knowledge from before.
3. We now have very specific change approvers for each of the Service Offerings (we do Service offering approvers + CAB (which is just BRMs) for approvals), this allows us to not push all changes through the CAB and have a more efficient/agile process.
4. We now have an owner for each Service Offerings, when we look at items from an Enterprise Architecture perspective this is helpful.   I created Service offerings dashboards which filter relevant data to all the Service Offering Owners (i.e. SLA met %'s, Change %s, Request and Incident volume, KB usage, etc.)
5. We assign all of our assignment groups a Level 1, Level 2 or Level 3 value.   Doing this and combining the Service Offering data I can focus where I need to on the Shift Left that is desperately needed here.
6. We do use Resolution Codes based on 7 different options; we have a simple pivot table report we run that shows top 20 service offerings pivoted on resolution code so we can make high-level decisions for problem management (e.g. Resolution code of User Training/Education in high numbers means we need to improve User Experience/Training).
7. Allows us to focus our limited resources on the proper things (i.e. on problem management and KB creation with the top incident generators) since all of the above let's us make data driven decisions.

As far as changing the organization's mindset, for #5 we were able to put an estimated amount of dollars we spend on support for each level which (i believe) has been eye opening for leaders in other groups.   This also (in our organization) shows that we have too many Level 3 people working incidents and need to improve our Level 2 team.   Re-allocating resources based on the data I gather above is a simple discussion since I have the data.  

I make decisions at a 1,000 foot level with the above data (GIGO) and often times we find the data can be misleading when i got down a level or two within the organization.   The model still works, it is just in a state of on going continuous improvement and it drives the right conversations.

Hope this helps!