Customer support metrics: what to measure and what good looks like

We earn a commission if you buy through some links on this page, at no extra cost to you. It never affects the order. How we make money.

Customer support metrics show how quickly you help customers, how well you solve their problems and what support costs your business. A small team does not need dozens of reports; a handful of well-chosen customer support metrics, reviewed every week, is enough to spot problems early and improve. This guide explains the key measures, how to calculate them, what affects them, and how to use them without turning support into a numbers game.

Why measure support at all

Without numbers, support quality is a feeling. Managers hear about the angry customer and miss the hundred who waited too long and left quietly. Metrics give you an honest picture: how long customers wait, whether problems are solved, which issues keep coming back and whether the team is stretched.

The goal is not to chase numbers for their own sake. It is to find where customers struggle and fix it, whether through faster replies, better help articles, product changes or more staff. A help desk tracks most of these metrics automatically.

First response time

First response time is the time between a customer contacting you and the first reply from a person. It is the metric customers notice most, because it tells them someone is working on their problem.

Measure the median rather than the average, since a few very slow tickets distort averages. Track it separately for each channel: customers expect chat replies within minutes and email replies within hours. Automatic acknowledgements should not count as a first response; most help desks exclude them.

To improve it: set business hours and clear targets, route tickets to the right person automatically, use saved replies for common questions, and check staffing at peak times with the support team size calculator.

Resolution time

Resolution time is the time from a ticket being opened to being solved. It reflects the whole experience, including back-and-forth, waiting on other teams and fixing the actual problem.

Some help desks also measure full resolution time including periods when the ticket was reopened. Look at resolution time by category: a long time for billing issues may point to a slow refund process, while a long time for technical issues may point to a dependency on developers.

Improve it by giving agents the access they need to solve problems themselves, writing internal guides for complex issues, and agreeing response times with other teams that support depends on.

First contact resolution

First contact resolution is the share of tickets solved with a single reply, without the customer having to come back. It is one of the best signs of quality, because customers prefer one complete answer to several partial ones.

Calculate it as tickets solved after one agent reply, divided by all solved tickets, for a period. Not every ticket can be solved in one reply, so compare categories rather than aiming for a single target. Improve it by asking for all needed information up front (through forms or bots), writing complete saved replies, and training agents to anticipate the customer's next question.

Customer satisfaction (CSAT)

Customer satisfaction is usually measured with a one-question survey after a ticket is solved: "How would you rate the support you received?" with a scale such as good or bad, or one to five. CSAT is the share of positive responses.

Keep the survey short and send it right after resolution. Read the comments, not just the score; they tell you why. Response rates are often low and skewed towards very happy or very unhappy customers, so treat CSAT as a trend indicator rather than an exact measure. Review every negative rating, and look for patterns by category, channel and agent.

Net Promoter Score and Customer Effort Score

Net Promoter Score asks how likely a customer is to recommend you, on a scale of zero to ten. It measures loyalty to the whole business, not just support, so it is usually owned by leadership or marketing.

Customer Effort Score asks how easy it was to get the issue resolved. It suits support well, because reducing effort, such as fewer transfers and less repetition, is something the support team controls directly.

For a small team, CSAT plus occasional effort questions is usually enough. Adding too many surveys annoys customers and lowers response rates.

Ticket volume and contact rate

Ticket volume is the number of new tickets in a period. On its own it says little, because it grows with your customer base. A better measure is contact rate: tickets divided by active customers or orders. If contact rate rises, something is causing more customers to need help, such as a confusing feature, a delivery problem or unclear billing.

Break volume down by category using tags. The top three categories usually account for a large share of tickets, and each is a chance to reduce volume through better help articles, product fixes or clearer communication.

Backlog

The backlog is the number of open tickets waiting for your team at a point in time. A steady backlog is normal; a growing backlog means tickets are arriving faster than they are solved. Check it daily and look at the age of the oldest tickets. A small backlog with a few very old tickets usually means some tickets are stuck, such as waiting on another team or forgotten.

When the backlog grows, act early: pause non-urgent work, bring in helpers from other teams, or send a proactive message to affected customers if one issue is causing many tickets.

Self-service and deflection

When customers find answers in your help centre, they do not need to contact you. Self-service rate or deflection is hard to measure precisely, but useful signals include help article views compared with tickets, searches with no results, and AI bot resolutions where customers did not ask for a person.

Look at searches that return nothing and tickets on topics that have an article. Both suggest gaps or unclear content. Our guide on how to build a knowledge base covers writing articles that actually reduce tickets.

Service level agreement (SLA) compliance

An SLA is a target, such as "first reply within four business hours for email". SLA compliance is the share of tickets that met it. SLAs are useful because they turn vague goals into clear promises, and help desks show timers so agents can see which tickets are due.

Set targets you can actually meet with your current team, by channel and priority. Publish them to customers only once you meet them reliably. Many help desks include SLA tools only on higher tiers, which affects cost; see what help desk software is.

Cost per ticket

Cost per ticket is your total support cost in a period divided by tickets solved. Include salaries, software and any outsourcing. It helps you compare options, such as hiring another agent versus investing in self-service or AI, and shows whether support costs are growing faster than the business.

Use it with care: cutting cost per ticket by rushing agents usually hurts satisfaction and creates more tickets later. The healthier way to lower it is to prevent tickets in the first place. Our help desk cost calculator estimates software cost by agent count, a useful input.

Agent metrics: use carefully

Help desks report per-agent numbers such as tickets solved, response times and CSAT. These are useful for coaching and spotting someone who is overloaded, but dangerous as targets. An agent measured only on tickets solved may rush or close tickets too early; one measured only on CSAT may avoid difficult customers.

Look at agent numbers alongside quality reviews of actual replies, and compare agents handling similar types of tickets. Use the numbers to start conversations, not to rank people.

What good looks like

There is no universal benchmark, because good numbers depend on your industry, product and customers. A software company with complex technical questions will have longer resolution times than an online shop answering delivery questions.

Instead of chasing industry averages, set your own baseline: measure for a month, then aim to improve the numbers that matter most to your customers. If customers complain about waiting, focus on first response time. If they complain about repeating themselves, focus on first contact resolution. Review targets every quarter as the team and tools change.

A simple weekly support dashboard

Most small teams need only one screen, reviewed weekly:

• New tickets and contact rate, compared with last week.
• Median first response time by channel.
• Median resolution time.
• CSAT score and a few recent comments.
• Backlog and the age of the oldest open ticket.
• Top five ticket categories.

Spend the meeting on what the numbers suggest, such as an article to write or a process to fix, rather than on the numbers themselves.

Turning metrics into improvements

A practical improvement loop:

1. Pick the top ticket category by volume.
2. Read twenty tickets in it to understand why customers contact you.
3. Decide on a fix: a product change, a help article, a clearer email, a better form or a bot answer.
4. Make the change and watch the category's volume and CSAT for a few weeks.
5. Repeat with the next category.

Share findings with the rest of the business. Support data is one of the best sources of insight into what customers find confusing.

Metrics for chat and WhatsApp

Messaging channels need slightly different measures. Customers expect near real-time replies, so first response time targets are in minutes. Conversations often pause and restart, so resolution time can look long even when the customer is happy. Track wait time before the first reply, handover rate from bots to people, and satisfaction per channel. For WhatsApp specifically, see WhatsApp CRM.

Common mistakes

• Using averages instead of medians, so a few outliers hide the real picture.
• Counting automatic replies as first responses.
• Setting agent targets that encourage rushing or closing tickets early.
• Tracking too many metrics and acting on none.
• Ignoring contact rate and celebrating fast replies to problems that should not exist.
• Comparing yourself with industry benchmarks that do not fit your business.

How to set up metrics in your help desk

Most help desks track the main metrics automatically, but a little setup makes the numbers more reliable:

• Set business hours so response times are measured in working hours, not overnight.
• Exclude automatic replies from first response time.
• Tag tickets by category with a short, fixed list, so volume by topic is meaningful.
• Turn on satisfaction surveys after tickets are solved.
• Define SLA targets by channel and priority, if your plan includes them.
• Merge duplicate tickets so volume is not inflated.

If you are still choosing a tool, compare reporting features in the help desk ranking and Zendesk vs Freshdesk. Reporting depth is one of the main differences between plan tiers.

Example: using metrics to fix a problem

A subscription software company with three support agents noticed CSAT falling over two months. First response time was fine, so speed was not the issue. Looking at negative ratings by category, most came from billing tickets, and the comments mentioned being passed between people.

Reading the tickets showed the cause: agents could not issue refunds and had to ask the finance lead, who replied once a day. Resolution time for billing was several times higher than other categories. The company gave senior agents permission to issue refunds up to a set amount, wrote an internal article explaining the process, and published a clear refund policy in the help centre.

Within a few weeks, billing resolution time fell and CSAT recovered. The metrics did not fix anything themselves, but they pointed to exactly where to look. For writing those articles well, see how to build a knowledge base.

Metrics when you add AI support

If you add an AI bot or AI reply drafting, measure its effect directly. Track the share of conversations the bot resolves without a person, how often customers ask for a human, satisfaction for bot-handled conversations compared with agent-handled ones, and whether tickets that reach agents after a bot conversation take longer. AI features are often charged per resolution, so compare that cost with your cost per ticket. Our guide to AI tools for small business covers where AI tends to pay off.

Questions