Blog · Board and management reporting · Customer service
A helpdesk KPI scorecard should reconcile ticket backlog and distinguish response, resolution, and reopen cohorts. Work through an internal employee helpdesk example, then change the opening backlog to see what moves.
A helpdesk KPI should show whether employees receive timely help and whether unresolved work is accumulating. Start with backlog, first human response, resolution time, target attainment, and reopens. Keep each measure's ticket cohort and service clock visible so a change in the report reflects a change in service.
An internal employee helpdesk handles access problems, equipment requests, and other workplace support. This scorecard evaluates that service. Broader customer KPIs and customer satisfaction KPIs cover different business questions and should not substitute for a ticket movement reconciliation.
Use separate populations for separate measures:
| KPI | Eligible population | Numerator or measured interval |
|---|---|---|
| Closing backlog | Tickets unresolved at the closing snapshot | Distinct ticket count |
| First response time | Created-cohort tickets with a recorded first human reply | Creation to first meaningful reply |
| Resolution time | Resolution movements during the period | Creation to that resolution milestone |
| Resolution target attainment | Eligible resolution movements | Movements completed within the applicable target |
| Reopen rate | Resolved-ticket cohort with a complete observation window | Distinct tickets reopening within that window |
These definitions intentionally distinguish a status movement from a distinct ticket. One ticket can be resolved, reopened, and resolved again. Count its movements when reconciling the queue, but count it once when measuring whether a cohort ticket reopened.
Service clocks also require a policy. Yale Health Sciences IT's service agreement distinguishes response and resolution targets, business hours and calendar hours, and circumstances in which resolution measurement may pause. Use your own approved service rules; those institutional targets are not a benchmark for the hypothetical business below.
The ticket file needs a stable ticket ID, queue, request type, priority, created timestamp, first human response timestamp, resolution timestamp, and status. Add status history if tickets can reopen or resolve more than once. Retain the applicable service target and any exported service-clock durations.
Take opening and closing unresolved-ticket snapshots at explicit timestamps in the same time zone. Include waiting-for-employee and waiting-for-supplier statuses if they remain unresolved under your policy. A paused service clock does not automatically remove a ticket from backlog.
Exclude duplicate imports, and record merges, transfers, and cancellations separately. Do not remove a ticket simply because its elapsed time is inconvenient. Florida Tech's service-level description explicitly excludes time outside its operational hours and holidays, illustrating why a calendar-duration export cannot silently replace a business-hours measure.
A hypothetical US distributor reviews one reporting week for its access and device-support queues. Both use the company's documented business-hours calendar. Resolution targets are assigned by priority in the export; attainment below uses each movement's applicable target.
There are no merges, cancellations, or transfers in this example. The movement table counts every transition into and out of the unresolved population, including reopens of older tickets.
| Queue | Opening unresolved | New tickets | Reopen movements | Resolution movements | Closing unresolved |
|---|---|---|---|---|---|
| Access | 20 | 100 | 4 | 94 | 30 |
| Devices | 30 | 60 | 6 | 56 | 40 |
| Total | 50 | 160 | 10 | 150 | 70 |
The controlling identity is:
Closing backlog = opening backlog + new tickets + reopen movements − resolution movements
For access, 20 + 100 + 4 − 94 = 30. For devices, 30 + 60 + 6 − 56 = 40. Combined, 50 + 160 + 10 − 150 = 70, which must equal the closing unresolved-ticket snapshot.
Backlog rose by 70 − 50 = 20 tickets. Reporting that the helpdesk resolved 150 tickets would hide this increase. Resolutions did not offset all incoming and reopened work.
All newly created tickets in this base example received a first human response within the reporting week. Thus the observed response cohort includes all 160 new tickets. Resolution movements include eligible tickets created earlier as well as during the week.
| Queue | First replies observed | Total first-response minutes | Resolution movements | Total resolution hours | Resolutions within target |
|---|---|---|---|---|---|
| Access | 100 | 900 | 94 | 188 | 88 |
| Devices | 60 | 1,500 | 56 | 336 | 42 |
| Total | 160 | 2,400 | 150 | 524 | 130 |
Durations are business-clock totals across their respective eligible observations. They describe elapsed service time, not the hours technicians spent handling requests.
| Measure | Access | Devices | Combined |
|---|---|---|---|
| Average first response, minutes | 900 / 100 = 9.0 | 1,500 / 60 = 25.0 | 2,400 / 160 = 15.0 |
| Average resolution, hours | 188 / 94 = 2.00 | 336 / 56 = 6.00 | 524 / 150 = 3.49 |
| Resolution target attainment | 88 / 94 = 93.6% | 42 / 56 = 75.0% | 130 / 150 = 86.7% |
Percentages are rounded to one decimal place and resolution averages to two decimals. Combined averages divide total duration by total observations. They do not average the queue averages.
The device queue warrants review because it combines slower response, longer resolution, and lower target attainment. Segment by priority and request type before assigning a cause: the device queue may carry more complex work.
For this example, the illustrative aging threshold is more than five business days since ticket creation, measured on the same business-hours calendar.
| Queue | Closing unresolved | Older than five business days |
|---|---|---|
| Access | 30 | 8 |
| Devices | 40 | 20 |
| Total | 70 | 28 |
The aged share is 28 / 70 = 40.0% of closing backlog. Retain the threshold in the report and inspect the old-ticket list; a low average resolution time can coexist with neglected unresolved work.
For reopen quality, use a separate, fully observed prior cohort:
| Mature-cohort input | Count |
|---|---|
| Distinct tickets resolved in the prior cohort | 100 |
| Those tickets reopening within seven calendar days | 8 |
| Reopen rate | 8 / 100 = 8.0% |
These eight tickets are not interchangeable with the current week's ten reopen movements. The cohorts and counting units differ. Compare rates only after each cohort ticket has a complete observation window.
Add the reporting week's labor cost alongside the resolution count:
| Current-week cost input | Value |
|---|---|
| Included helpdesk labor cost (USD) | 9,000 |
| Resolution movements | 150 |
Cost per resolution movement is 9,000 / 150 = $60.00. Label that denominator explicitly; repeated resolutions can lower this ratio without helping more employees. Use KPI examples to frame complementary measures rather than treating this cost ratio as the whole scorecard.
Keep every base-case input fixed except the device queue's opening backlog, which rises from 30 to 50 tickets. This variation represents more work carried into the week, not a change in the current week's service durations or new requests.
| Backlog calculation | Base case | Variation |
|---|---|---|
| Opening unresolved, both queues | 50 | 70 |
| New tickets | 160 | 160 |
| Reopen movements | 10 | 10 |
| Resolution movements | 150 | 150 |
| Closing unresolved | 70 | 90 |
The variation closes at 70 + 160 + 10 − 150 = 90 tickets. Closing backlog is 20 tickets higher, while the observed response and resolution measures remain unchanged. Management must distinguish inherited workload from the current week's processing performance; the variation alone does not establish the added tickets' age.
Bring a sample of your own ticket export, status history, and backlog snapshots to Covirage, with the service-clock definitions used by your team. Covirage maps the columns, and its tools compute the comparisons and check the movement totals; the external AI model explains the results with every figure traced to its rows. Create a free Covirage account, which takes about a minute, then set up your business and bring your file.
See Covirage for customer service teams for the supporting service-analysis context.
Response time measures the wait until a meaningful first reply. Resolution time measures the interval until the agreed resolution milestone. They use different endpoints, and neither necessarily equals the technician's handling time. State whether the clock counts business hours and whether any statuses pause it.
Start with unresolved tickets at the opening snapshot, add new tickets and reopen movements, then subtract resolution movements. Add explicit adjustments for transfers, merges, or cancellations where relevant. The result must match unresolved tickets at the closing snapshot.
Use the service commitment agreed for each priority and employee need. An urgent access outage and a planned equipment request can have different targets. Compare like priorities on the same service clock rather than applying one external benchmark to every request.
Choose tickets resolved in a defined cohort, allow a fixed observation window, and count the distinct tickets that reopen during that window. Divide by eligible resolved tickets. Keep this quality measure separate from the count of reopen status movements used to reconcile backlog.