Top ITSM metrics you need to track and measure

Most ITSM dashboards track traditional metrics such as ticket volume, SLA compliance, first response time, and resolution time. These metrics are still useful for understanding how much work passes through the service desk and how quickly it gets handled.

But once AI starts resolving a share of requests, traditional reporting can make it difficult to see how much work automation is completing and how AI-handled requests compare with those handled by the IT team.

That’s why IT teams need to pair traditional service desk metrics with measures that give them more visibility into automation performance and whether automated resolutions hold.

This guide covers the ITSM metrics worth tracking, which ones matter most for lean teams, and how to measure the impact of AI on support performance.

TL;DR

• Traditional ITSM metrics still matter, but they don’t show how support work is split between AI and the IT team.
• For AI-enabled support, prioritize auto-solve rate, AI vs. human time to resolution, first response time, employee CSAT, and escalation rate.
• Track AI and human resolution times separately so a blended average doesn’t obscure how quickly each type of ticket is resolved.
• Look at reopen rate and what happens before and after escalation to check whether automation is resolving requests effectively.
• If your ITSM doesn’t separate AI and human activity, you may need to calculate these metrics manually. Risotto captures both alongside your existing ticketing stack

Why most ITSM metrics fall short for IT teams today

Most traditional ITSM metrics were designed for service desks where humans handled most requests. That led teams to focus on metrics such as ticket volume, SLA compliance, and resolution time to track how much work agents handled and how quickly they handled it.

These metrics are still useful, but as AI takes on more common IT requests, traditional reporting can make it harder to see how work is split between automation and the IT team.

For example, overall resolution time can tell you whether support is getting faster, but it won’t show whether that change comes from AI-handled or human-handled tickets. Likewise, the number of tickets resolved doesn’t tell you how many were completed without an IT agent stepping in.

That lack of detail can make reporting difficult. One IT leader we spoke to said their dashboard couldn’t show how many tickets AI had resolved, so they had to paste screenshots of those figures into updates for their CSO.

That doesn’t make traditional service desk metrics obsolete. The table below shows what they still tell you and what additional context can help once AI becomes part of the support workflow.

The next section narrows this down to the five metrics IT teams should prioritize when AI and humans share the support workload.

The 5 essential ITSM metrics for lean IT teams

These five metrics give lean IT teams the clearest view of how well an AI-enabled service desk is performing, covering both automation performance and the employee experience.

1. Auto-solve rate

Auto-solve rate is the percentage of tickets resolved end to end without an IT agent stepping in. It’s a much clearer measure of how much repetitive work automation is removing for your team.

It also gives IT a more defensible measure than ticket deflection. Deflection can be vendor-defined and difficult to verify because the interaction may never create a ticket. If every request still creates a ticket, you can clearly separate automated from human-handled requests, recalculate the numbers from your own ticket data, and maintain an audit trail of what happened. That makes the results easier to stand behind when reporting automation performance to leadership.

This metric is also useful for deciding what to automate next. Break auto-solve rate down by request type or department to find where automation performs well and where requests still need agent involvement.

So, what’s a good auto-solve rate? In Risotto’s prospect conversations, starting baselines have often been around 0–15%. Teams have set short-term POC targets around 20–25%, with medium-term goals of 35–50%. More mature automation programs can exceed 50%, although the right target will depend on the types of requests your team handles and how much of that work can realistically be automated.

As that rate increases, the team can handle more support volume without adding the same amount of manual work.

For example, Risotto customer Gusto now auto-resolves 55% of tickets on average and supports twice the ticket volume with the same lean team.

2. Time to resolution, split AI vs human

Time to resolution (TTR) measures how long it takes to resolve an employee’s request, from their first message to resolution. In an AI-enabled service desk, we recommend tracking TTR separately for requests resolved by AI and those resolved by an IT agent.

Lumping the two together can hide the impact of automation. If AI resolves routine requests much faster than tickets that need an IT agent, one overall average won’t show that difference.

Alongside the AI vs. human split, consider whether the point when a ticket is marked as resolved matches when the employee actually has a working fix. From the employee’s perspective, the important measure is the full elapsed time from their first request to a working fix, including time spent waiting for replies, approvals, or other steps in the process.

Research by Anunta found that IT leaders often point to SLA compliance as evidence of fast support, while employees judge resolution by when they can actually get back to work. That makes elapsed resolution time especially useful for understanding the support experience from the employee’s point of view.

Risotto customer results show how much resolution times can vary depending on who handles the request. ThoughtSpot reduced average resolution time from 31 hours for human-handled tickets to 6.5 hours for AI-handled tickets. And at Gusto, requests resolved by AI average 5 hours, compared with 35 hours for those handled by an IT agent.

Tracking the difference between AI and human resolution times also gives IT a clearer way to quantify the time automation is saving and use it when calculating IT automation ROI.

3. First response time

First response time measures how long an employee waits between asking for help and receiving the first useful response.

SLA compliance tells you whether your team responded within an agreed target. First response time shows exactly how long the employee waited, which makes it more useful for tracking changes in the support experience over time.

With conversational ticketing, requests can be captured and acted on automatically in Slack or Teams instead of waiting in a queue for an IT agent to respond. For example, Fundrise has an average first response time of 3.8 seconds using an AI automation layer like Risotto. 

Keep an eye on this metric over time. If first response time starts increasing, break it down by channel, request type, or time of day to identify the cause of delays.

4. Employee CSAT

Employee (CSAT) measures how employees rate their support experience after a request is resolved. It adds direct employee feedback to operational metrics such as TTR and auto-solve rate.

Many IT teams don’t have a CSAT baseline yet. You can start by establishing one and track how satisfaction changes over time. CSAT survey response rates of around 10–15% can be realistic, and how you collect feedback can affect participation. Simple in-channel ratings, such as emoji-based feedback in Slack, let employees respond without opening a separate survey. 

Look at CSAT alongside your resolution metrics. If TTR falls while employee ratings improve, faster support is translating into a better experience. If ratings decline, investigate the tickets behind them to understand where the experience is breaking down.

Risotto captures this feedback directly in Slack. Employees can rate resolved tickets, and the Analytics view lets IT report on tickets by rating.

5. Escalation rate: instant vs eventual

Escalation rate measures how often a request still needs an IT agent to step in. For AI-enabled support, it’s useful to look at when that escalation happens, not just how often.

An instant escalation happens when a request is passed to IT straight away. An eventual escalation happens after automation has already tried to resolve it. Tracking the two separately helps you see which requests need a human from the start and where automation is getting partway through before handing off.

A high escalation rate can also point to gaps in knowledge coverage. That link shows up in the customer data: a team working with only around 30 knowledge-base articles was seeing an escalation rate of roughly 90%. If escalations are high, look at the highest-volume request types being handed to IT and identify which could be automated next by expanding knowledge coverage or refining workflows.

Risotto also tracks chat-with-team escalations, where an IT agent joins the conversation without taking over the request completely.

See your auto-solve rate, AI vs. human resolution times, and hours saved out of the box. Book a demo

How to validate whether IT support automation is actually working

The five metrics above give you a high-level view of support performance. A second set of checks helps you see whether automated resolutions hold and whether requests that still need an IT agent are handed over without unnecessary friction.

A ticket being marked as resolved doesn’t tell you what happened afterwards, such as whether the employee reopened it or continued asking for help. One team we spoke to worried that closed tickets could inflate its resolution numbers. Another team lead manually reviewed 61 resolved tickets to check whether the underlying issues had actually been fixed.

To validate how well your automation is performing, look at these three checks alongside the metrics above:

• Reopen rate of auto-solved tickets. Track how often tickets handled by AI are reopened. If a large share comes back, your auto-solve rate may be overstating how many issues automation successfully resolved. Break reopen rate down by request type to find automations that need attention.
• Interactions before escalation. Track how much back-and-forth happens before an employee reaches an IT agent. If automation can’t resolve a request, it should recognize that quickly and hand it over. A high number of interactions before escalation can mean employees are spending too long trying to get the help they need.
• Employee messages after resolution. Check whether employees continue the conversation after a ticket has been marked resolved. Follow-up messages can indicate that the original resolution didn’t fully solve the problem, while no further contact can be a useful signal that the issue stayed resolved.

Reported alongside auto-solve rate, these metrics give IT a stronger basis for showing whether automation is genuinely resolving work instead of relying on resolved-ticket counts alone. That becomes especially useful when leadership asks how the automation numbers were calculated and how reliable those results are.

Make sure your metrics are measuring what you think they are

Check the underlying ticket data before reporting. Poor categorization or inconsistent ticket statuses can skew metrics such as auto-solve rate and resolution time.

For example, onboarding, offboarding, and IdP status-change tickets can inflate solve-rate figures if they’re grouped with standard support requests. Transferred tickets may also appear as closed in the originating department, while resolved tickets left open by agents can distort resolution time.

Set clear rules for which tickets count toward each metric, make sure ticket statuses are used consistently, and spot-check the data before using it in internal reporting.

How to track ITSM metrics without building manual reports

Choosing the right metrics is only useful if you can track them consistently. Your existing ITSM may report traditional metrics without separating work handled by AI from work handled by your team. In that case, you may need to work out metrics such as auto-solve rate or AI vs. human TTR manually.

Some IT teams are already finding workarounds for those reporting gaps. One Head of IT we spoke to was even exporting helpdesk data into ChatGPT to get useful insights from their data.

Risotto removes that extra step by capturing AI and human support activity in the same reporting layer. The analytics dashboard shows auto-solve rate, AI vs. human resolution times, and hours saved automatically. This gives IT a clearer way to show leadership both how much work is being automated and what that translates to in time saved. 

You can also track first response time, ticket volume, backlog, ticket status, and escalation patterns, with breakdowns by department, category, channel, and time period.

From there, you can investigate changes without exporting the data elsewhere. Risotto lets you compare automation performance across workflows and departments, monitor where Tier-1 requests are handed to Tier-2, and ask questions about your support data in plain English.

The platform can also generate weekly or monthly operational summaries covering trends, automation metrics, and escalation insights for leadership.

You don’t need to replace your existing ITSM to get this visibility. Risotto can sit alongside tools such as Jira, Freshservice, ServiceNow, or Zendesk with bi-directional sync, so your existing platform can remain the system of record while Risotto adds automation and the reporting needed to measure it.

Measure the impact of IT support automation with Risotto

The right ITSM metrics give you visibility into both service desk performance and the impact of automation. As AI takes on more requests, that means understanding how much work it resolves, how quickly those requests are handled, and whether those resolutions hold.

Risotto gives you that view automatically, so you can measure the impact of automation without rebuilding your reporting process or replacing the ticketing software you already use.

FAQs about ITSM metrics

What are ITSM metrics and KPIs?

ITSM metrics are measurable data points that show how your service desk is performing, such as ticket volume, resolution time, SLA compliance, and employee CSAT. A KPI is a metric tied to a specific business or service goal, such as reducing resolution time or increasing auto-solve rate.

What is the difference between ticket deflection and auto-solve rate?

Ticket deflection measures requests that avoid human support, often through self-service or automated answers. Auto-solve rate measures tickets that automation resolves end to end without an IT agent stepping in. Auto-solve rate therefore gives IT a more direct measure of completed automated work.

Which ITSM metrics matter most for IT teams?

Lean IT teams should focus on auto-solve rate, AI vs. human time to resolution, first response time, employee CSAT, and escalation rate. Together, these metrics show how much work automation is removing, how quickly employees get help, and where human involvement is still needed.

How often should IT teams review their ITSM metrics?

Review ITSM metrics often enough to catch meaningful changes and act on them. For most teams, weekly operational checks and monthly trend reviews are a practical starting point. Watch for shifts in auto-solve rate, resolution time, CSAT, and escalations, then investigate the request types driving them.