Measuring problem management

How best to measure your Problem Management practice?

A while ago Simon Higginson wrote a good article for The ITSM Review. While I agreed with most of it, I disagreed over some of the proposed KPIs for Problem Management. The article suggested four "SLAs":

1. Provision of Problem Management reference number
2. Time to get to the root cause of the issue
3. Measurement of provision of Root Cause Analysis documentation. To be provided within X working days of initial notification
4. Measurement of progress on root cause analysis actions as agreed (Target dates not to change more than twice)

First a pedantic point: An SLA is a contract between two people or human groups. An SLT is a target within that SLA. These four targets are SLTs not "SLAs". To be really pedantic, they may often be within OLAs not SLAs in which case they are not even SLTs. So let's just call them KPis: metrics used to measure people's performance. SLTs are a special sort of KPI.

A couple of less-pedantic points:

• There are many causes to a problem. Nominating one as the "root" cause is highly subjective. So if a timeframe is set, i'll always comply. "oops we're out of time. Yup that one looks like the root to me".
• Time to diagnose? time to fix? it is worth measuring these to spot trends. But how can we reasonably set them as targets? "Doctor, you must diagnose all patients within two hours" "Fireman, you must extinguish all small fires within three days, all medium ones within eight hours, and all enormous ones within two hours".
  SLTs are not just metrics. They are thresholds that people are held accountable for. If they are not fair then they won't work, and i don't think some time-based SLTs are fair.
  #1 and #3 are fine: they are defined standardised sequences of steps that can be repeatably performed, and improved to meet the SLT. #2 and #4 are anyone's guess: each problem is unique and some are unsolvable. By all means measure the trend, but the time for a particular problem is out of any person's control. This is a classic Standard+Case distinction.

I'd rather see KPIs like:

• Mean time to first respond to problems, by priority
• Satisfaction of the service desk and incident management team(s) with problem management
• Reduction in a composite risk index for the problem portfolio
• Percentage of problems resolved
• Percentage of recurring problems
• Percentage of problems "standardised": documented so they can be dealt with in a pre-defined manner when they recur
• Costs and resource usage in problem management
• Percentage of shelved problems ("cold cases")
• Number of knowledge articles produced - especially workarounds - by the problem management technicians

What are your prefered Problem Management KPIs?