Why service desk numbers need care
A service desk is among the most measured parts of many organizations, because the ticketing tool produces numbers automatically. That convenience is a trap. Easy-to-count measures are not always the ones that reflect good service, and a single figure can be read in opposite ways.
Below are common measures, what each reveals, and how each can mislead. The examples are illustrative; the right set depends on what the service is for.
Volume measures
Tickets received. Shows demand over time and helps with staffing. A rise may mean growth, a new system causing confusion, a recurring fault, or simply that it has become easier to report problems. A fall may mean fewer issues, or that people have given up and found workarounds. Volume is context for other measures, not a verdict.
Tickets by category. Shows where demand comes from, such as password resets, hardware or a specific application. This is often the most actionable view, since a large category may be a candidate for a better knowledge base or a permanent fix to a root cause.
Time measures
Response time. How long until someone first acknowledges a ticket. It reassures the requester, but a quick automated reply can make this look excellent without any real help being given. It is worth knowing whether the figure counts human responses or automated ones.
Resolution time. How long from logging to resolving. It is the measure most people think of first, and it needs a precise definition: working or calendar hours, whether the clock pauses while waiting for the user, and how reopened tickets count. Resolution time is also usually skewed by a few very long tickets, so an average can be dominated by outliers. A median, or a view by percentage resolved within a given time, often tells a more representative story. Teams under pressure to reduce it may close tickets prematurely, so it is best read together with reopen rates.
Age of oldest tickets. Shows whether anyone is being left waiting for a very long time. Averages hide this, which makes it a useful companion measure.
Quality measures
First contact resolution. The share of issues resolved at the first interaction. It points toward efficient service and fewer repeat contacts. The difficulty is definition: does it count a ticket where the user did not reply? Does it count phone fixes but not chat? Aiming too hard at it may discourage agents from escalating issues that genuinely need specialists.
Reopen rate. How often resolved tickets return. A high rate suggests fixes are not sticking or tickets are being closed too soon. A very low rate may mean users find it too difficult to reopen anything.
Escalation rate. How often tickets move to a second-line team. It can reflect gaps in training or documentation on the front line, or simply a complex mix of work. Low is not always good: front-line staff sometimes hold on to tickets they should pass on.
Experience measures
Satisfaction scores. Typically collected from short surveys after a ticket closes. They capture how the service felt, which timings cannot. Cautions: response rates are often low, and people with strong feelings are more likely to respond, so the sample may not be representative. Scores can also reflect the issue itself rather than the service, since someone whose problem could not be solved may rate poorly regardless of how helpful the agent was. Reading the comments usually teaches more than the number. This ties into the wider idea of the help desk experience for users.
Workload measures
Backlog. The number of open tickets at a point in time. A growing backlog suggests demand exceeds capacity or work is stuck, but its meaning depends on age and type. A large backlog of low-priority items waiting for user replies is different from a modest backlog of urgent, aging tickets. Look at backlog alongside age and priority.
Tickets per agent. Sometimes used to judge productivity. It is easily misused: complex tickets take far longer than simple ones, and agents who also train others or update documentation do work that is not counted. It is better used for capacity planning than for ranking people.
Utilization and occupancy. How much time agents spend on tickets. Very high levels leave no room for training or improvement and tend to be hard to sustain; very low levels may indicate overstaffing or uneven demand. There is no universal ideal figure.
A worked illustration of misleading averages
The numbers are invented. Five tickets are resolved in 1, 1, 2, 2 and 44 working hours. The total is 1 + 1 + 2 + 2 + 44 = 50, and 50 divided by 5 gives an average of 10 hours. Yet four of the five were resolved within 2 hours, and the median is 2. Reporting only “average 10 hours” suggests slow service, when in fact one problem ticket distorted the result. Reporting both figures, and asking about that one ticket, is more informative.
Reading measures together
No measure should be read in isolation. Some natural pairings:
- Resolution time with reopen rate and satisfaction, to check that speed is not coming at the expense of quality.
- Ticket volume with category, to find recurring causes.
- Backlog with ticket age, to see whether the queue is healthy.
- First contact resolution with escalation rate, to check that agents escalate when appropriate.
Frameworks for IT service management usually describe categories such as incidents, problems and requests, and ITSM basics can help with consistent categorization. Measuring each type separately, rather than blending them, avoids many misleading averages.
Targets and behavior
Once a measure becomes a target, behavior changes. Tickets get closed quickly or split into several small ones; difficult cases are left to age. These patterns are covered in the article on target side effects. A mix of measures, regular sampling of actual tickets for quality, and conversation with staff and users help keep the numbers honest.