Help desk KPIs: 6 metrics worth tracking (and what to drop from your dashboard)

Most IT teams don’t lack metrics. The harder problem is having numbers they can trust and use to show leadership whether support is actually improving, especially as more requests are handled through automation or AI.

Take ticket volume. If it falls, more requests may be auto-solved. Or employees may simply have stopped using the help desk. If volume rises, the team may be receiving more requests because recurring issues are creating extra work, or because employees are using the official support channel more consistently.

A useful KPI gives you a clearer indication of what’s changing. For internal IT, that means understanding whether employees are getting their issues resolved, how long they’re waiting, and how much manual support work is being removed from the team.

This guide covers six metrics worth prioritizing as KPIs, guidance to help you set realistic targets, and four commonly reported metrics that are better kept off your KPI dashboard.

TL;DR

• A help desk metric becomes a KPI when you use it to track progress toward a specific priority, usually against a target or threshold.
• For internal IT, start with six metrics: auto-solve rate, first contact resolution, reopen rate, mean time to resolution (MTTR), backlog age, and employee satisfaction.
• Use benchmarks to sense-check your KPI targets. Your targets should reflect the types of requests your team handles and its current performance.
• Ticket volume, tickets resolved per agent, agent utilization, and average handle time are useful operational metrics, but they make poor standalone KPIs.
• Don’t judge a KPI in isolation. Check it against a related metric, such as auto-solve rate with reopen rate, or MTTR with backlog age. This helps to make sure that improvement in one area isn’t hiding a problem elsewhere.

A help desk metric isn’t automatically a KPI

Before we go into the weeds, let’s clarify the difference between a metric and a KPI. ITSM metrics measure an aspect of support performance. A KPI is a metric your team has chosen to track against a specific objective, with a target or threshold attached.

For example:

• Metric: Auto-solve rate is 14%.
• KPI: Increase auto-solve rate for eligible Tier-1 requests from 14% to 20% in the near term, with a longer-term target of 50%.

This is important because the same metric won’t be equally important to every IT team. A metric can be useful for monitoring or diagnosing support performance, but it only functions as a KPI when you use it to measure progress toward a specific goal.

Your KPIs should therefore reflect your current priorities. If the goal is to reduce repetitive Tier-1 work, auto-solve rate might become a KPI. If employees are waiting too long for support, MTTR or backlog age may be more useful.

Build your help desk KPI set around four performance areas

For most internal IT teams, the six metrics below provide a practical starting point. Together, they cover four areas.

1. Automation and work removed from IT

If the goal is to reduce manual Tier-1 support, auto-solve rate shows how much of that work is being resolved without human input.

Auto-solve rate

Auto-solve rate is the percentage of eligible support requests resolved end to end without an IT team member taking over.

You can calculate it as: 

📊Auto-solve rate = auto-solved eligible requests ÷ total eligible requests × 100

Unlike ticket deflection, auto-solve only counts requests that are actually resolved. Effective AI ticketing automation needs to complete the request end to end; sending an employee to a help article or automating part of the process doesn’t count if an IT specialist still has to finish the job.

Before setting a target, define which requests are eligible for auto-solve and what counts as a fully resolved request. Every request should still create a ticket so you can cleanly compare automated and human-handled requests. Otherwise, requests that never enter the ticketing system can shrink the denominator and make the auto-solve rate look better than it is.

There’s no universal auto-solve target because the achievable rate depends on the types of requests your team receives. Across Risotto customers, auto-solve rates typically range from 20–70%, with an average of around 50%. Gusto, for example, now auto-resolves 55% of tickets on average.

2. Resolution quality

Resolution quality is about whether the issue was properly resolved the first time. First contact resolution measures how often requests are resolved in the first interaction, while reopen rate measures how often tickets are reopened after being marked resolved. We’ll look at both in more detail below.

[H4] First contact resolution

FCR measures the percentage of requests resolved during the employee’s first interaction with the help desk, without follow-up or escalation.

You can calculate it as:

📊FCR = requests resolved on first contact ÷ relevant requests × 100

The 2025 Freshservice Benchmark Report found that roughly three in four tickets in its dataset were resolved during the first interaction with the employee.

It's important to remember that FCR varies significantly by request type. A password reset or straightforward “how do I?” question has far greater FCR potential than a complex incident that requires investigation. Requests dependent on manager approval may also require multiple interactions even when support is working exactly as intended.

Rather than using one average FCR across all requests, calculate FCR for similar request types. You can then set appropriate targets for each group, such as a higher FCR for routine troubleshooting than for tickets requiring specialist input.

Reopen rate

Reopen rate measures the percentage of resolved tickets that are later reopened. It helps show whether issues are actually being resolved the first time or whether employees need to come back for more support.

The calculation is as follows:

📊 Reopen rate = reopened tickets ÷ resolved tickets × 100

For chat-based support, define what counts as a reopen before using it as a KPI. An employee might reply in an old Slack thread, start a new request, or contact IT separately about the same issue. One practical definition is repeat contact about the same issue within a set number of days.

Where possible, track repeat contact across channels so reopen rate doesn’t undercount issues that weren’t fully resolved.

3. Employee time lost

An unresolved IT issue can leave an employee waiting or unable to work. HappySignals’ 2026 Global IT Experience Benchmark report found that employees lose an average of 3 hours and 18 minutes of productive time per IT incident.

That makes the time employees spend waiting for support an important part of help desk performance. Two useful KPIs here are MTTR and backlog age. We’ll discuss them in more detail below. 

Mean time to resolution

MTTR measures the average time between an employee submitting a request and that request being resolved. 

The calculation is:

📊 MTTR = total resolution time for resolved requests ÷ number of resolved requests

Freshservice’s benchmark report found an average resolution time of around 22 hours for IT tickets.

Because MTTR is an average, it can hide requests that take much longer than most to resolve. Track the 90th percentile resolution time alongside it to see the point by which 90% of resolved tickets were completed.

Then investigate what’s causing the longer resolution times. Reassignments are one useful place to start; HappySignals found that 13.3% of tickets were passed between two or more teams, with employee lost time increasing as the number of handoffs rises.

If AI handles part of your support queue, compare AI and human resolution times too. At Hazel Health, AI-resolved tickets average 57 minutes compared with 75 hours for human-handled tickets. At Gusto, the equivalent figures are 5 hours and 35 hours.

Backlog age

Backlog age shows how long unresolved tickets have remained open.

This metric tells you something different from backlog size. The number of open tickets shows how much unresolved work there is; backlog age shows how long that work has been waiting.

There’s no universal backlog-age target for internal IT. How long a ticket can reasonably stay open depends on its priority and the SLA attached to it.

Set backlog age thresholds against the relevant SLAs and group open tickets into bands so you can see which requests are approaching or exceeding them. Then split older tickets by who they’re waiting on, such as IT, an approver, or the employee. This makes it easier to distinguish a support bottleneck from requests that are stalled on an external response.

4. Employee experience

Resolution metrics tell you how efficiently IT handled a request. But ITSM experience also depends on how employees felt about the support they received. Employee satisfaction helps capture that side of performance.

Employee satisfaction

Employee satisfaction is commonly measured using post-resolution customer satisfaction score (CSAT) or another consistent employee experience score.

Calculate it as:

📊 CSAT = positive responses ÷ total responses × 100

Collect feedback as soon as possible after the request is resolved. For internal support, this can be as simple as a thumbs-up/down or emoji rating in Slack or Teams. Track the response rate alongside the score, as a high CSAT based on a small share of resolved requests may not represent the wider employee experience.

Don’t treat CSAT as a simple measure of whether the problem was fixed. HappySignals’ 2026 benchmark found that employees were more likely to cite speed of support, the support person’s attitude, and their skills than the resolution itself when describing their IT experience.

Satisfaction benchmarks are also difficult to compare directly because results depend on the question asked, scoring scale, and collection method. Keep these consistent so you can track changes in your own CSAT over time.

Metrics to drop from your KPI dashboard

Not every useful help desk metric belongs on your KPI dashboard. The four metrics below can still help IT understand demand, workload, and capacity, but they make poor KPIs because a higher or lower result doesn’t consistently mean support has improved.

Ticket volume

Ticket volume is the number of support requests received over a given period.

It can look like a straightforward way to judge help desk performance, but ticket volume can rise or fall for very different reasons. 

Higher volume might reflect more IT issues, or simply greater use of the official support channel. Lower volume could mean automation is resolving more requests, or that employees have stopped using the help desk.

That makes ticket volume a poor KPI on its own. If your target is to reduce ticket volume, a lower number can look like an improvement even when employees are simply bypassing the help desk.

Ticket volume is still useful for tracking demand, spotting spikes, and identifying high-volume request types.

Tickets resolved per agent

Tickets resolved per agent measures how many tickets each team member closes over a given period.

The metric doesn’t account for how difficult those tickets are. An agent resolving five complex incidents may be doing more work than someone closing 20 routine requests. As more straightforward Tier-1 work is automated, agents may also close fewer tickets because the requests left for them require more time and judgment.

That’s why it makes a poor KPI on its own. If agents are judged on how many tickets they close, they may be encouraged to prioritize quick requests over more complex work.

Don’t remove this completely from your dashboard though. Tickets resolved per agent is still useful for understanding how work is distributed across the team and whether staffing levels match demand.

Agent utilization

Agent utilization measures the percentage of an agent’s available work hours spent handling support requests.

A high utilization rate doesn’t necessarily mean the help desk is performing well. Agents may be spending most of their time on repetitive Tier-1 work, and if nearly all of their capacity is already taken up, there’s little room to absorb urgent incidents or escalations.

The reverse can also be true. Lower utilization may be a positive sign if automation has removed repetitive support work and freed up time for tasks that require human judgment.

Agent utilization is therefore more useful for understanding team capacity than as a KPI for support performance.

Average handle time

Average handle time (AHT) measures the amount of agent time spent actively handling a support request.

Setting a target to reduce AHT can encourage the wrong behavior. Agents may rush troubleshooting, close tickets too early, or escalate them simply to reduce their own handling time. Complex requests also legitimately take longer, so a lower AHT doesn’t necessarily mean better support.

AHT is more useful for identifying request types that consistently consume a lot of agent time and may be candidates for process improvements or automation.

Pro-tip: These capacity metrics discussed above can still help answer staffing questions, but they’re better used to understand how much work the team can support than to grade help desk performance.

Don’t judge KPIs in isolation

Even a well-chosen KPI only tells you part of the story. A KPI can improve without the overall support experience improving, so check it against a related metric to make sure the change hasn’t created a problem elsewhere. Here are three examples:

1. Auto-solve rate + reopen rate: If auto-solve rate is a KPI, monitor reopen rate to see whether auto-solved requests are later being reopened.
2. FCR + employee satisfaction: If FCR is a KPI, check employee satisfaction to make sure more first-contact resolutions aren’t coming at the expense of the support experience.
3. MTTR + backlog age: If MTTR is a KPI, monitor backlog age to make sure faster resolution of completed tickets isn’t happening while older requests remain unresolved.

The second metric doesn’t need to become another KPI. It gives you extra context for judging whether improvement in your KPI is actually producing the result you wanted.

How to report help desk KPIs to leadership

Operational KPIs become more useful to leadership when you connect them to the amount of work they remove. If auto-solve rate increases, estimate how much employee or IT time that represents, then translate that into the staffing or cost impact where useful.

Be careful with generic ROI assumptions. The value of an hour varies by team, so use your own staffing costs or internal rates rather than applying one dollar value across every department.

Keep your KPI dashboard focused with Risotto

A focused KPI dashboard should prioritize the metrics that show whether internal support is meeting the team’s most important goals. Other metrics, such as ticket volume, agent utilization, and AHT can still help IT investigate demand or capacity without being treated as KPIs.

That distinction becomes even more important as support is automated. Fewer tickets reaching agents, for example, can be a positive result if more requests are being resolved without manual IT involvement.

Risotto is an AI service desk for internal IT that uses Tier-1 ticket automation to resolve common requests in Slack or Teams while working alongside your existing ITSM. It gives lean IT teams visibility into auto-solve rate, AI vs. human resolution times, and hours saved, alongside their existing ticket data. The ITSM remains the system of record.