AI-generated editorial illustration.
Service desk KPIs are most useful when they explain one another. SLA achievement describes performance against agreed targets; backlog shows unfinished work; MTTR describes a defined time interval; and first contact resolution helps examine how work is resolved. None of these measures, on its own, proves that the customer experience improved.
Build the review around a consistent ticket population, visible calculation rules and a few exceptions that need a decision. That is more useful than adding another chart to a report nobody can interpret consistently.
1. Define the ticket population first
Separate incidents from service requests when their workflows or targets differ. Record whether a measure uses tickets opened during the period, resolved during the period or still open at the reporting date. Keep priority, service and team changes visible when comparing periods.
A resolution metric based on this month's closed tickets cannot be compared casually with a workload figure based on this month's arrivals. Both can be valid, but they answer different questions.
2. Read SLA achievement with the eligible population
For a resolution-SLA measure, a clearly documented calculation could be eligible resolved incidents that met their resolution target / eligible resolved incidents with a valid target × 100. State how paused clocks, exclusions and missing targets are treated.
Illustrative example: 180 of 200 eligible resolved incidents met the agreed target, producing 90% achievement. Before drawing a conclusion, check whether the ten percent includes a small set of high-impact incidents or a large number of lower-priority delays.
Show the count alongside the percentage. Also check changes in ticket mix: an easier reporting population can improve the percentage even when a particular service becomes harder to support.
3. Use backlog as a workload conversation
Backlog is a point-in-time view of unfinished work. Add ageing bands, priority and the reason a ticket is waiting so the team can distinguish blocked work, unassigned work and activity awaiting a customer or supplier.
Illustrative movement: opening backlog is 80 tickets, arrivals are 420 and closures are 400. With no other adjustments, closing backlog is 100. Reopens, cancellations and transfers would need their own reconciliation if present.
SLA achievement could improve at the same time that this backlog grows. The review should ask whether capacity, intake quality, dependencies or work selection explains the difference. Closing easy tickets faster is not the same as reducing the oldest unresolved risk.
4. Specify exactly what MTTR means
The acronym MTTR can refer to different intervals. Atlassian's incident-metric guide distinguishes repair, recovery, response and resolution measures. Name the interval in your report and document whether the clock uses calendar hours or business hours.
For example, a team may measure total opening-to-resolution hours for eligible incidents resolved in the period / number of those incidents. Define how reopened incidents and waiting time are handled before comparing teams or months.
Read the average alongside the distribution. If five incidents take 1, 1, 2, 2 and 14 hours, the mean is 4 hours and the median is 2 hours. That small illustrative sample shows why one long-running incident can change the average. Review the exception rather than assuming every ticket took four hours.
5. Make first contact resolution a quality measure
First contact resolution, or FCR, needs a local definition that matches the support channels. Decide what counts as the first contact, whether transfers disqualify a ticket and how a qualifying reopen is detected.
One operational definition is the share of eligible resolved incidents completed at first contact without transfer or a qualifying reopen. If the team uses an observation window after closure, allow that window to mature before reporting the final result. A seven-day window, for example, would be a local choice rather than a universal rule.
Read FCR with reopen rate and customer feedback. A quicker closure is not an improvement if the same customer must return because the original issue was not resolved.
6. Run a 30-minute exception review
- 5 minutes: confirm data quality, populations and changes in ticket mix.
- 10 minutes: inspect SLA breaches and ageing backlog by service impact.
- 10 minutes: discuss resolution-time outliers, repeat contacts and dependencies.
- 5 minutes: assign the next action, owner and evidence checkpoint.
This suggested agenda keeps the discussion connected to operating choices. An action might be to fix routing for a recurring request type, remove a supplier dependency or improve a knowledge article, depending on what the evidence shows.
Service desk KPI checklist
- Each measure has an agreed population and calculation.
- Counts and percentages are shown together.
- Backlog movements reconcile, including adjustments.
- MTTR clock rules and time units are explicit.
- FCR is checked against repeat contacts and resolution quality.
- Actions name an owner and a follow-up date.
Explore the Service Desk KPI Toolkit for a reusable review structure. For the wider leadership discussion, see the monthly IT health review guide.
The definitions and numbers above are worked examples. Align final calculations and targets with your service agreements and operating model.