What Are the Best Customer Service KPIs?
Customer service KPIs (key performance indicators) are the numbers that tell you whether your support team is fast, effective and leaving customers satisfied. If you can only track three, start with:
- CSAT - are customers satisfied with the support they received?
- First contact resolution (FCR) - is the problem solved the first time?
- First response time - how long do customers wait for a first reply?
Add average resolution time and ticket volume per customer as soon as you have the data. The other KPIs below matter as your team grows or adds AI. This guide gives the formula, what the number tells you, and how to improve it for each KPI. For the broader customer-outcome side (NPS, churn, retention, lifetime value), see our guide to customer KPIs and how to choose them.
| KPI | Formula | Tells you |
|---|---|---|
| CSAT | Satisfied responses / total responses x 100 | Quality of individual interactions |
| CES | Average rating on an ease-of-effort question | How hard customers had to work |
| FCR | Tickets resolved in first contact / total x 100 | Effectiveness |
| First response time | Average or median time to first reply | Speed and availability |
| Average resolution time | Total time to resolve / tickets resolved | End-to-end speed |
| AHT | (Talk + hold + after-contact work) / contacts handled | Efficiency and staffing |
| SLA compliance | Tickets within SLA / total tickets x 100 | Reliability |
| Contact rate | Tickets / active customers (or orders) | Whether support demand is growing faster than the business |
| Repeat contact rate | Customers contacting again about the same issue / all contacts | Whether fixes stick |
| AI resolution rate | AI conversations closed without handoff / all AI conversations | How much AI actually handles |
Turn Instagram DMs into booked calls
AI that qualifies leads, handles objections, and follows up - so you don't have to.
1. Customer Satisfaction Score (CSAT)
What it is: a short survey after a support interaction, usually "How satisfied were you with the support you received?" on a 1 to 5 scale.
Formula: CSAT = (number of 4 and 5 responses / total responses) x 100.
Why it matters: it is the most direct read on how customers felt about a specific interaction, and it is easy to segment by channel, topic and agent.
How to improve it: send the survey right after the interaction closes, keep it to one question plus an optional comment, and read the comments from low scores weekly. Look at CSAT by topic as well as overall, since an average hides weak spots. Watch the response rate: a CSAT based on a handful of replies is noise.
2. Customer Effort Score (CES)
What it is: how easy it was for the customer to get their issue resolved, asked as "The company made it easy for me to handle my issue" on a 7-point agree-disagree scale.
Formula: CES = average of all responses, or the percentage of respondents who answered 6 or 7.
Why it matters: high effort (repeated contacts, channel switching, long forms) predicts frustration even when the final answer is correct. It is useful for finding friction in your process rather than judging an agent.
How to improve it: reduce handoffs, let customers resolve issues in one channel, pre-fill known information, and fix the steps customers complain about most.
3. First Contact Resolution (FCR)
What it is: the share of issues fully resolved in the first interaction, with no follow-up or reopened ticket needed.
Formula: FCR = (issues resolved on first contact / total issues) x 100. Decide in advance what counts as "resolved" (for example, no reopen or repeat contact within 7 days) and use the same rule every month.
Benchmark: SQM Group's call center benchmarking puts the industry average at about 71% (measured by post-call survey) and treats 80% or higher as world-class. See SQM's explanation of a good FCR rate. Treat it as a reference point for phone-heavy teams, not a target for every channel.
How to improve it: give agents the information and permissions to finish the job, improve your knowledge base, route issues to the right skill group the first time, and review reopened tickets to see what was missing.
4. First Response Time
What it is: the time between a customer's first message and the first reply from a person or an AI agent.
Formula: the average, and preferably the median and 90th percentile, of (first reply time minus ticket creation time). Report business hours and total hours separately so weekends do not distort the picture.
Why it matters: the first reply sets the tone, and slow replies are one of the most common causes of dissatisfaction. Set targets per channel, since customers expect minutes in chat and DMs and hours in email.
How to improve it: use auto-acknowledgements, saved replies and an AI agent for common questions, balance staffing against volume peaks, and prioritize the oldest unanswered conversations first. For ideas, see how to improve customer service response time.
5. Average Resolution Time
What it is: the time from the moment a ticket is opened to the moment it is resolved.
Formula: total resolution time / number of resolved tickets. Use the median alongside the average, because a few very old tickets can skew the average badly.
How to improve it: shorten handoffs, set clear ownership, use escalation rules (see our ticket escalation process guide), and track which categories take longest so you can fix the root cause instead of speeding up agents.
6. Average Handle Time (AHT)
What it is: the average time an agent spends on one contact, including talk or chat time, hold time and after-contact work such as notes.
Formula: AHT = (total talk time + hold time + after-contact work) / number of contacts handled.
A warning: AHT is useful for staffing and capacity planning, but optimizing it directly encourages agents to rush. Read it next to CSAT and FCR. A longer AHT on complex issues is fine if the customer leaves satisfied.
7. SLA Compliance
What it is: the share of tickets handled within the response and resolution times you promised internally or to customers.
Formula: (tickets that met the SLA / total tickets) x 100, measured separately for first response and for resolution.
How to use it: define SLAs by priority, not one number for everything, and review breaches weekly for patterns (a certain topic, hour or queue).
8. Ticket Volume, Contact Rate and Backlog
What they are: the number of tickets created, the ratio to your customer base, and the number of open tickets waiting.
Formulas: contact rate = tickets / active customers (or orders) in the period. Backlog = open tickets, ideally tracked by age.
Why it matters: raw ticket counts rise whenever the business grows. The ratio shows whether support demand is growing faster than the business, which usually points to a product or process problem. To bring the number down, see how to reduce support tickets.
9. Repeat Contact Rate
What it is: the share of customers who contact you again about the same issue within a set window, usually 7 days.
Formula: (customers who contacted again about the same issue / all customers who contacted) x 100.
Why it matters: it is the honest check on FCR and on any automation you add. If tickets "close" quickly but customers keep coming back, the closes are not real resolutions.
10. AI Resolution Rate and Handoff Rate
If an AI agent handles part of your support, you need KPIs that describe its work separately from your human team.
- AI resolution rate = AI conversations that ended without a human handoff and without the customer returning with the same issue / all AI conversations x 100.
- Handoff rate = AI conversations passed to a person / all AI conversations x 100.
- Answer quality = the share of a weekly random sample of AI answers that a reviewer marks correct.
- Customer rating of AI conversations = CSAT segmented by AI-handled versus human-handled.
A caution on definitions: vendors define a "resolution" differently. For example, Intercom bills Fin per outcome and Zendesk bills per automated resolution, and the rules differ. Define your own measure of success (no human handoff plus no repeat contact within 7 days is a reasonable starting point) and use it consistently. See also our guide to analytics for chatbots.
Customer Service KPI Benchmarks
Searches for "best in class customer service metrics" rarely find numbers that apply to your business, because benchmarks depend on industry, channel, customer type and how each metric is measured. Use these as a guide:
| KPI | What you can reasonably compare against |
|---|---|
| FCR | SQM Group's call center benchmark: about 71% average, 80% or higher world-class (post-call survey method) |
| NPS | Bain's definition, and industry NPS studies for your sector |
| CSAT | Your own trend over time, and industry indices such as the ACSI for your sector |
| First response time | Your own targets per channel, based on what customers in your market expect |
| Resolution time | Your own median by ticket category |
| AI resolution rate | Your own baseline, measured with a consistent definition |
Many teams set an internal CSAT target of 80% or higher. That is a rule of thumb, not an industry standard. The most useful benchmark is your own trend: pick a baseline, improve it quarter on quarter, and compare like with like (same channel, same definition).
KPIs for Individual Agents
Common agent-level KPIs are CSAT per agent, FCR, tickets resolved, quality assurance (QA) review scores and adherence to SLAs. Use them for coaching, not ranking. Volume and speed metrics on their own reward rushing, so combine at least one speed metric with one quality metric, and review a sample of conversations rather than only the numbers.
KPIs for Instagram and Messaging Teams
If much of your customer conversation happens in Instagram DMs or WhatsApp rather than a ticketing system, the same ideas apply with a few channel-specific measures:
- First response time in DMs, since people expect quick replies in messaging.
- Qualification rate, the share of new conversations that turn into a qualified lead.
- Booking rate and show rate, if the conversation ends in an appointment.
- Handoff rate to a human, and what happens after it.
- Conversations per active customer, as a measure of demand on the channel.
If an AI agent handles the first replies, check your platform's analytics for these. FlowGent includes conversation analytics (basic analytics on the Hobby plan and advanced analytics on Pro), and its AI agent for Instagram and Instagram appointment setter handle first replies, qualification and booking in the DM thread, so these KPIs are the ones to watch.
How to Choose and Use Your KPIs
- Start with a goal. Faster replies, happier customers, fewer tickets and lower cost each point to different KPIs.
- Pick three to five. More than that and nobody acts on them. Include one speed metric, one quality metric and one demand metric.
- Define each one in writing. What counts as a resolution, which hours count, which channels are included.
- Set a baseline first. Track for a month before you set targets.
- Review weekly, adjust quarterly. Weekly reviews catch problems. Quarterly reviews decide whether the KPIs still fit your goals.
- Read them together. A falling AHT with a falling CSAT is a warning sign, not a win.
FAQ
What are the 3 most important KPIs for customer service?
If you can only track three, use CSAT (quality), first contact resolution (effectiveness) and first response time (speed). Add resolution time and contact rate when you have the data.
How do you calculate customer service KPIs?
Each KPI has a simple formula: CSAT is satisfied responses divided by total responses, FCR is first-contact resolutions divided by total issues, first response time is the average gap between a ticket opening and the first reply, and contact rate is tickets divided by active customers. The formulas are in the table above.
What is a good first contact resolution rate?
SQM Group's call center benchmarking reports an industry average of about 71% and treats 80% or higher as world-class. Definitions and channels vary, so compare against your own history first.
What are KPIs for AI customer service?
The core ones are AI resolution rate, handoff rate, answer quality from a sampled review, and customer satisfaction on AI-handled conversations compared with human-handled ones. Because vendors define resolution differently, set your own definition.
What is the difference between KPIs and metrics?
A metric is any measurement. A KPI is a metric you have chosen because it tracks progress toward a specific goal. You may measure dozens of metrics but should only treat a few as KPIs.
How often should we review customer service KPIs?
Review operational KPIs such as response time, backlog and SLA compliance weekly. Review quality and outcome KPIs such as CSAT, FCR and contact rate monthly, and revisit which KPIs you track every quarter.