Fact table's changing measures

I have a business process that I'm modeling. The process is "Service Installation" and the measures are the amount of time it takes for various sub processes within the main process.

Example:

• Time to greet customer: 2 minutes
  
• Time to establish installation location: 5 minutes
  
• Time unpacking tools: 4 minutes
  
• etc.
  

Typically I'd design this like so:

fact_service_installation_time

• id
  
• date_id
  
• customer_id
  
• greet_customer_time
  
• establish_installation_location_time
  
• unpack_tools_time
  
• etc.
  

Unfortunately, these sub-processes measured may very well change. New ones added, old ones removed, etc. and at an unpredictable rate.

This makes me want to model this as such:

fact_service_installation_time

• id
  
• date_id
  
• customer_id
  
• installation_sub_process_id
  
• time_spent
  

This is somewhat shunned however http://www.kimballgroup.com/data-warehouse-business-intelligence-resources/kimball-techniques/dimensional-modeling-techniques/measure-type-dimension/

What are some good / manageable ways to handle such a situation?

I still like option 2 given the variety and variability of events. If you go wide how do you add up the total time or for that matter, which events were even applicable? Option 2 filters down to only applicable events and summing total time is easily accomplished. Performance can be addressed via partitioning or an aggregation fact table.

I agree with B&L that your second option is probably the best choice. You may also consider an aggregation that presents a total and the more common measures as columns. But getting agreement on what the latter should contain could be challenging.

Thank you both - you have each come through for me on several occasions and your input is highly valued.