Customer Support KPIs Every Business Should Track
The customer support KPIs that matter most are first response time, resolution rate, average handle time, customer satisfaction (CSAT), and ticket volume by reason. Together they tell you how fast you reply, whether you actually solve the problem, how much each contact costs, and why customers reach out in the first place. Track those five and you can fix almost anything.
What are customer support KPIs?
Customer support KPIs are the numbers that show whether your support is working. They turn a vague feeling like "we're busy" into something you can measure, compare week to week, and act on. Without them, you're guessing.
A good KPI connects to an outcome you care about. First response time connects to whether a customer feels ignored. Resolution rate connects to whether they have to come back a second time. CSAT connects to whether they'd buy from you again. Pick metrics that link to behavior, not vanity counts.
The mistake most teams make is tracking everything and acting on nothing. A dashboard with thirty numbers is a dashboard nobody reads. Start with five KPIs you'll review every week, and only add more once those are stable and on target.
Think of your KPIs as a small story about the customer's experience. Ticket volume tells you how many people needed help, first response time tells you how long they waited, resolution rate tells you whether they got what they came for, CSAT tells you how it felt, and cost per ticket tells you what it took to deliver. Read in that order, the five numbers walk you through a customer's journey, and a problem in any one of them points to a specific place to look.
What is a good first response time?
A good first response time is under five minutes for chat, under one hour for email, and instant for an AI customer service agent that replies the moment a customer asks. The faster the first reply, the less likely a customer is to give up and leave.
First response time measures how long a customer waits before any human or system acknowledges their message. It's not the same as resolution time. A fast first reply that says "we're looking into it" still beats silence, because silence is what makes people anxious and angry.
This metric matters more than almost any other because it sets the tone for the whole conversation. We dig deeper into why speed drives loyalty in our guide to first response time. Bund AI posts a 1.8 second median first response, which means most customers get a real answer before they've finished re-reading their own question.
How do you measure resolution rate?
Resolution rate is the share of tickets closed without the customer coming back about the same issue. You measure it by tagging conversations as resolved, then watching how many reopen or generate a follow-up within a few days. A high reopen rate means you closed tickets, not problems.
There are two flavors worth separating. First contact resolution is the percent solved in a single interaction. Overall resolution rate counts everything eventually solved, however many touches it took. First contact resolution is the harder, more honest number.
Aim for a first contact resolution above 70 percent on routine questions. Bund AI resolves 86.7 percent of tickets with no human touch at all, because it doesn't just answer, it takes the action behind the request, like issuing a refund or changing an order. When the AI handles the volume customers actually have, your humans get the rare, genuinely hard cases.
Here's a concrete way to read the number. Say you close 1,000 tickets in a month and 120 of them generate a second contact about the same problem within three days. Your first contact resolution is 88 percent, and those 120 reopens are your worklist. Sort them by reason and you'll usually find a few recurring causes. Maybe the answer was technically right but didn't match what the customer could see in their account. Fix the top two or three causes and watch the reopen count fall the following month. That loop, measure, find the reopens, fix the cause, is how resolution rate stops being a report and starts being a roadmap.
What's the difference between leading and lagging support KPIs?
Leading KPIs predict what's coming, and lagging KPIs confirm what already happened. First response time and resolution rate are leading. They move first, and they shape the customer's experience while it's still in progress. CSAT, retention, and repeat purchase rate are lagging. They tell you the result after the customer has already decided how they feel.
You need both, but you act mostly on the leading ones. If you wait for CSAT to drop before you do anything, you're reacting to damage that's already done. If you watch first response time creeping up week over week, you can fix the cause before a single customer rates you poorly. The leading metrics are your early warning system, and the lagging ones are how you confirm the fix worked.
A practical setup pairs each lagging KPI with the leading one most likely to move it. Watch CSAT, but steer with first response time and first contact resolution, because those two are usually what CSAT is really measuring.
What is CSAT and how do you track it?
CSAT, or customer satisfaction score, is the percent of customers who rate a support interaction positively. You collect it with a one-tap survey right after a conversation closes, usually asking "How would you rate the support you received?" on a simple scale. The percent who answer in the top boxes is your CSAT.
Keep the survey short and send it fast, while the experience is fresh. A survey that arrives three days later measures memory, not the interaction. Response rates climb when you ask in the same channel where the conversation happened.
CSAT is only useful if you read the comments, not just the score. A 4 out of 5 with the note "took too long" tells you exactly what to fix. For the full method, see how to measure customer satisfaction with surveys that people actually finish.
Why track average handle time and cost per ticket?
Average handle time is how long an agent spends actively working a ticket, and cost per ticket is what each resolved contact costs you in labor. Together they tell you whether your support scales or quietly bleeds money as you grow. They're the financial backbone of any support operation.
Watch average handle time as a balance, not a race. Pushing it too low can mean agents are closing tickets fast and badly, which drives reopens and tanks CSAT. The goal is a time that's efficient and thorough, not just short.
Cost per ticket is where automation changes the math entirely. When an AI agent resolves most routine contacts at a fixed monthly cost, your cost per ticket drops as volume rises instead of climbing with it. You can see how the numbers play out across plans on the Bund AI pricing page, and how support spend connects to revenue in our breakdown of the ROI of customer support.
How do you tie support KPIs to revenue?
You tie support KPIs to revenue by connecting them to retention, repeat purchases, and churn. A customer who gets a fast, complete answer buys again. A customer left waiting cancels. Support isn't a cost center when you measure its effect on whether people stay.
Start by segmenting your churn data by support experience. Customers who had a slow or unresolved ticket usually churn at a higher rate than those who didn't. That gap, multiplied by their lifetime value, is the dollar cost of bad support.
Then run it forward. If improving resolution rate by ten points keeps even a small share of those customers, the revenue saved often dwarfs the cost of the tooling that got you there. KPIs become a budget argument, not just an operations report.
How often should you review these KPIs?
Review your core KPIs weekly and your trends monthly. Weekly catches problems while they're small, like a sudden spike in response time after a product change. Monthly shows whether the fixes are holding and whether your targets still make sense as you grow.
Set a target for each KPI, not just a number to watch. "First response time under one hour" is a target. "First response time: 47 minutes" is trivia. The target is what turns the metric into a decision.
Share the dashboard with the whole team, not just managers. When agents see the same numbers leadership sees, they make better calls in the moment, and the metrics stop feeling like surveillance and start feeling like a shared scoreboard.
One more habit pays off: pair every weekly review with one decision. It's tempting to walk out of a KPI meeting having admired the chart and changed nothing. Force the question instead. What's the one number furthest from target, and what's the single change you'll make this week to move it? A small change every week, tied to a real metric, beats a quarterly overhaul that never quite ships.
How do AI agents change which KPIs matter?
When an AI agent handles the front line, some KPIs get easier and a couple become worth watching. First response time nearly stops being a problem, because an AI replies the instant a customer asks. Bund AI posts a 1.8 second median first response, so the metric that used to take constant staffing pressure to defend now holds itself. Your attention shifts from "are we fast enough" to "are we resolving enough."
The number that earns the spotlight is automated resolution rate, the share of contacts the AI closes completely on its own. It's the cleanest measure of how much real work the system is doing versus how much it's just deflecting. Bund AI resolves 86.7 percent of tickets with no human touch, and the gap between deflection and resolution is everything here. Deflection sends a customer away. Resolution sends them away satisfied. Watch the reopen rate on AI-handled tickets the same way you'd watch it for human ones, because that's how you know the automation is solving problems rather than burying them. For how this connects to the bigger financial picture, our breakdown of the ROI of customer support shows where automated resolution turns into saved revenue.
Frequently asked questions
Which support KPI should a small team start with?
Start with first response time and resolution rate. They're easy to measure, they directly shape how customers feel, and improving them moves almost every other metric. Once those two are stable and on target, add CSAT to capture the human side of the experience.
What's the difference between a KPI and a metric?
A metric is any number you can measure, like total tickets received. A KPI is a metric you've tied to a goal and committed to acting on, like keeping first response time under an hour. Every KPI is a metric, but not every metric earns the attention a KPI gets.
Can AI improve support KPIs without hurting quality?
Yes, when it actually resolves issues rather than deflecting them. Bund AI improves first response time and resolution rate at once because it answers from your own knowledge and takes the real action behind the request, then hands off to a human with full context when a case needs judgment. Speed and quality rise together instead of trading off.
How many KPIs is too many?
More than five or six for a single team usually means none get acted on. A focused dashboard beats a comprehensive one. Add depth only after your core metrics are consistently hitting target, and retire any KPI you haven't used to make a decision in the last quarter.
Track the metrics, then fix them automatically
Knowing your KPIs is step one. Moving them is the hard part, and it usually comes down to answering customers faster and resolving more on the first try. Bund AI does both, with a 1.8 second median first response and 86.7 percent of tickets resolved with no human touch, live on your site and email in under a day. Start free with Bund AI and watch your support KPIs move in the right direction.