Service desk metrics that actually mean something
First response time is the easiest metric to hit and the least useful. These are the numbers that tell you whether a service desk is any good.
Every managed service report leads with the same three numbers: tickets raised, tickets closed, and first response time against target. All three are green in almost every report ever produced, including the ones from providers that are being replaced.
That is not a coincidence. Those metrics are easy to hit and they measure the desk’s process rather than the user’s experience.
Here are the ones worth putting in a contract instead.
First contact resolution, by category
The percentage of tickets resolved by the first person who touched them, broken down by type of issue.
This is the number that describes capability rather than speed. A desk that answers in ninety seconds and escalates 70% of what it receives is a switchboard. A desk resolving 65% at first contact is doing the work.
Break it down by category, because the aggregate hides everything. A desk can look excellent on password resets and terrible on anything involving the mobile fleet, and the aggregate splits the difference and tells you nothing.
Reopen rate
The percentage of tickets closed and then reopened within a fortnight.
This is the honest counterweight to resolution time. Closing tickets fast is trivial if closure is not durable, and a desk under pressure on speed metrics will optimise exactly that way — usually without anybody deciding to.
Anything above about 5% means tickets are being closed rather than resolved.
Time to resolution, at the 90th percentile
Not the average, and not first response.
Averages hide the tail, and the tail is what people remember. Nobody forms an opinion of a service desk from the ticket that took four minutes. They form it from the one that took nine days and needed three follow-ups.
The 90th percentile is where the actual experience lives. Ask for it. Providers who quote averages usually have a reason.
Repeat contact rate per user
How many users raised more than three tickets this month, and which ones.
This surfaces two different problems: an underlying fault nobody has connected, and a user who has been failed enough times that they now escalate immediately. Both are expensive, and neither appears in a volume report.
Ticket volume trend, by root cause
Rising ticket volume is usually reported as a workload problem. It is more often a signal that something upstream is broken.
The useful version groups tickets by cause rather than by category — this many because the VPN drops on the fourth floor, this many because the new starter process misses a step, this many because a specific application breaks after each update. A desk that is genuinely working reduces its own volume over time by fixing causes. One that is not will report growing volume as evidence of value.
Backlog age, not backlog size
Twenty open tickets is fine. Twenty open tickets where the oldest is from March is not.
Ask for the age distribution of open tickets. A healthy backlog is mostly young; an unhealthy one has a sediment of things nobody will ever resolve and nobody will close either, because closing them requires admitting they were never going to be fixed.
The two questions no metric answers
Who actually answers the phone? A desk with high staff turnover shows normal metrics and produces a poor experience, because nobody accumulates knowledge of the environment. Ask about tenure and about how many people know your environment by name.
What happens at 2am? After-hours performance is where the difference between providers is largest and where the reporting is thinnest, because the volume is too low to move any monthly average. Ask for out-of-hours tickets as a separate report.
What to put in the contract
- First contact resolution, reported by category
- Reopen rate within 14 days
- Time to resolution at the 90th percentile, not the mean
- Backlog age distribution
- Out-of-hours performance reported separately
- A monthly root-cause review with at least one item actioned
And one clause that matters more than any of them: the provider reports on these whether or not they are flattering, and a missed target is discussed rather than re-baselined.
More on what our managed service includes, and the questions worth asking before you sign.