Retire five service desk metrics before they mislead you
Retire five service desk metrics before they mislead you
EXECUTIVE OVERVIEW
THE SITUATION
The monthly service desk report runs to twenty pages and nobody reads past the first chart. The numbers are accurate. They just describe the queue, not whether people got help.
THE CALL
Report on what changed and what you need, not on everything you counted.
SAY THIS
“We are replacing the monthly report with one page. It shows how long people wait for help, how many come back with the same problem, and the one thing we need from you.”
AVOID
Averages that hide the long waits Ticket counts as a measure of effort Targets nobody chose on purpose
TRY THIS WEEK
List every number in last month’s report and mark who acted on it. Drop the five that nobody used. Draft the one-page version and show it to your biggest critic first.
THE FULL STORY ·
7
MIN
Open last month’s service desk report and count the numbers. Twenty, thirty, maybe more. Now ask which of them changed a decision. Most reports have two or three that did and a long tail that nobody has looked at in a year.
That tail is not harmless. Every metric on the page asks for a little attention, and attention is the one thing your readers do not have. The useful numbers get buried, and the report trains people to skim.
Why good numbers mislead
A metric can be correct and still point the wrong way. Average resolution time looks healthy when most tickets close in an hour, even if a handful sit for two weeks. Tickets closed per analyst looks like productivity, until you notice it rewards closing fast over fixing properly.
The pattern is the same each time. The number measures what is easy to count, and over time people manage the number instead of the work behind it.
The five to retire
1 · AVERAGE RESOLUTION TIME
Replace it with the share of tickets resolved within the time you promised, and show the oldest open ticket next to it.
2 · TICKETS CLOSED PER ANALYST
Retire it. It rewards speed over quality and quietly punishes people who take the hard calls.
3 · TOTAL TICKET VOLUME
Keep a trend line if you must, but split out the repeat issues. Volume alone cannot tell you whether things are getting better.
4 · FIRST-RESPONSE TIME ON ITS OWN
An instant automated reply passes this test. Pair it with how long until a person actually helps.
5 · SURVEY RESPONSE RATE
It tells you about the survey, not the service. Read the comments instead and report the three themes.
If a number has not changed a decision in a year, it is not a metric. It is decoration.
What goes on the one page
Three measures and one ask. How long people wait for real help, measured as the share inside your promise. How often the same problem comes back within thirty days. What the people who used the service said, in their own words. Then a single request: a decision, a budget line, or a nudge to another team.
Keep the page stable from month to month so trends are visible at a glance. If something new belongs on it, something old has to leave.
Getting the change past your readers
Do not announce it. Show it. Send the one-page version alongside the old report for a month and ask what is missing. Most of the time the answer is nothing, and the old report quietly retires itself.
If someone does miss a number, give it back, but only if they can name the decision it supports.
The short version
Fewer numbers, chosen on purpose, each tied to a decision. Report what changed and what you need. The rest can live in a dashboard for the people who want it.
