Leaving Room for Mystery in Measurement — Resolution Time

15 Dec, 2016

Question: Who is the most annoying person in the world?

Answer: Possibly a cocksure IT metrics ideologue.


When you work in business intelligence, everyone expects you to be THAT person, the one that has all the answers immediately, with complete certainty and leaves no room for interpretation or discussion. The ole’ smartest person in the room. They interrupt your perspective with cold hard facts in every meeting that leave everyone feeling left out of some of the most important discussions a company makes about performance management. To make it worse, performance management is one of the most exciting topics for all involved, because we all want to be better… if we could just have that perfect definition of “better”.

So, we had an interesting week at Northcraft Analytics with one of our new customers in a discussion about one of the most tried and true Key Performance Indicators in Incident Management, Average Resolution Time (or MTTR).

First of all, a little backdrop from our benchmarking database to set the stage:

  • Average Resolution Time for all of our customers (over 100,000,000 Incidents) which includes every type of incident you could imagine happening in a large IT shop is @ 135 Hours (actual elapsed time, not business time).
  • These 100,000,000 Incidents are spread across most of the major industry verticals.
  • These incidents range across the entire spectrum, including desktop, NOC, DOC, enterprise applications, storage, networking, custom applications, database, etc…
  • A significant portion of the incidents are what ITIL people would actually call requests, to the tune of about 35%.
  • There is no “auto-close” time factored into this number. As an aside, many ITSM systems have an auto-close feature that will add time to the resolution duration, which can make the measure less clear.
  • Average Resolve Time (MTTR) is one of the more “external” metrics that we offer (and package) because of the lack of human intervention possibilities in an ITSM system for the fields used in this particular calculation.

So, we arrived at our customer site “knowing” that average resolution time was probably going to be in the aforementioned standard range. It’s a little running magic trick between the founders/brothers here at Northcraft to predict resolution time right before we bring up our dashboard to show new customers post-installation. We do like to look glamorous, after all.

What happened this week that was different? Well, we bring up the dashboard that covers a company that has over 250,000 Incidents/year (we brought up 3 this week, so this is anonymous, all of them are in that range). The average resolution time for this particular customer was a mere 38 hours across all IT Incidents.

What do you think happened?  Join the discussion on LinkedIn.