Most chat dashboards show dozens of numbers. A handful of them tell you whether chat is actually helping customers and paying for itself. This guide covers those: what each metric measures, the formula, and a target that is realistic for a small or mid-sized support team.
The targets below are common starting points, not industry law. Your first job is to measure your own baseline for a few weeks, then improve against it.
Speed metrics
1. First response time
How long a visitor waits between sending their first message and getting a reply from a person. This is the number visitors feel most, and it decides whether they stay or leave.
First response time = time of first agent reply − time of visitor's first message
Target: under 60 seconds during staffed hours. Report the median as well as the average; a few very slow chats can hide behind a good average.
2. Average response time
The average gap between a visitor's message and the agent's reply across the whole conversation, not just the first message. It shows whether agents are juggling too many chats at once.
Average response time = total agent reply delay ÷ number of agent replies
Target: under 45–60 seconds. If first response is fast but this is slow, agents are overloaded mid-conversation.
3. Queue wait time
How long visitors sit in a queue before an agent picks the chat up. It differs from first response time when chats are routed to departments or wait for a free agent.
Queue wait time = time chat was accepted − time chat was requested
4. Missed chat rate
The share of chat requests nobody answered before the visitor left. It is the clearest sign that you are understaffed at certain hours.
Missed chat rate = unanswered chats ÷ total chat requests × 100
Target: below 5%. Break it down by hour of day; the problem is usually a few specific time slots, not the whole day.
Quality metrics
5. Resolution rate
The share of chats that end with the visitor's problem solved in the chat, rather than being pushed to email or a ticket. This is what makes chat worth running.
Resolution rate = chats resolved in chat ÷ total chats handled × 100
Target: 70% or higher. Tag each chat as resolved, escalated or abandoned at the end so the number is reliable.
6. Average handle time
How long a chat lasts from the agent accepting it to it closing. Use it for staffing, not as a goal to minimise: pushing it down too hard lowers resolution and satisfaction.
Average handle time = total chat duration ÷ number of chats
7. Customer satisfaction (CSAT)
A one-question rating at the end of the chat, such as "Did we solve your problem?" or a 1–5 score.
CSAT = positive ratings ÷ total ratings × 100
Target: 80% or higher. Read the low-rated transcripts every week; they are the fastest way to find a training gap or a broken process.
8. Transfer rate
How often a chat is passed to another agent or department. A high rate usually means routing sends chats to the wrong place, or agents lack the knowledge to answer.
Transfer rate = transferred chats ÷ total chats × 100
Efficiency and business metrics
9. Chats per agent
How many chats each agent handles per hour, and how many at the same time. Two or three concurrent chats is a common comfortable range; beyond that, response times and quality tend to slip.
Chats per agent per hour = chats handled ÷ agent hours online
10. Chat conversion rate
For sales pages: how many visitors who chatted went on to buy or sign up, compared with visitors who did not chat. This is how chat earns its budget.
Chat conversion rate = purchases by chatters ÷ visitors who chatted × 100
Compare it with the conversion rate of visitors who did not chat on the same pages. The difference is chat's contribution.
11. Cost per chat
What each conversation costs you. Compare it with the cost of the same question arriving by email or phone.
Cost per chat = (agent cost + software cost) for the period ÷ chats handled
12. Proactive chat acceptance
If you send automatic invitations ("Need help choosing a plan?"), this measures how many visitors accept. A low rate means the invitation fires on the wrong page, at the wrong moment, or with the wrong words.
Acceptance rate = invitations accepted ÷ invitations shown × 100
All 12 at a glance
| Metric | What it tells you | Starting target |
|---|
| First response time | Whether visitors stay | < 60 s |
| Average response time | Agent overload mid-chat | < 60 s |
| Queue wait time | Routing and staffing | < 60 s |
| Missed chat rate | Coverage gaps | < 5% |
| Resolution rate | Whether chat solves problems | 70%+ |
| Average handle time | Staffing needs | Your baseline |
| CSAT | How customers feel | 80%+ |
| Transfer rate | Routing and knowledge gaps | < 10% |
| Chats per agent | Capacity | 2–3 at once |
| Chat conversion rate | Revenue impact | Above non-chatters |
| Cost per chat | Efficiency | Below email/phone |
| Proactive acceptance | Invitation quality | Your baseline |
Measuring automated chat and chatbots
If a bot or automated flow answers first, track these alongside the human metrics:
- Containment rate: chats the bot fully resolved without a person ÷ all bot chats.
- Handoff rate: chats the bot passed to an agent ÷ all bot chats. A rising number is fine if CSAT holds.
- Handoff wait: time from the bot handing off to an agent replying. Visitors who asked for a person expect one quickly.
- Bot CSAT: rate bot-only chats separately, or a good agent score can hide a frustrating bot.
Vanity metrics to ignore
- Total chats: more chats is not better. Ten fast resolutions beat twenty redirects to email.
- Messages sent: a long back-and-forth often means confusion, not engagement.
- Agents online: presence is not coverage. Missed chat rate tells you whether you have enough people.
- Chat length on its own: short chats are often the best ones. Read it next to resolution rate, never alone.
Calculate your own live chat metrics
Enter one week of numbers from your chat software's report. The calculator works out the core KPIs and checks each against the starting targets above. Nothing you type leaves your browser.
Chats per agent hour above 4 to 6 usually means agents are juggling more than two or three chats at once, and first response time will start to slip.
A weekly dashboard in five numbers
If you only track five, track these each week:
- First response time (median)
- Missed chat rate
- Resolution rate
- CSAT
- Chat conversion rate on your pricing or checkout pages
Watch them for four weeks before changing anything. Then fix whichever one is furthest from target, and change one thing at a time so you can see what worked.
Questions
What is the most important live chat metric?
First response time. It is the metric visitors feel directly, and a slow first reply causes missed chats, low satisfaction and lost sales further down the line.
What is a good first response time for live chat?
Under a minute during staffed hours is a common target. Measure your own median first, then improve on it.
How do I measure live chat ROI?
Compare the conversion rate of visitors who chatted with those who did not on the same pages, and compare cost per chat with the cost of handling the same question by email or phone.
How is resolution rate different from CSAT?
Resolution rate is whether the problem was solved in the chat; CSAT is how the visitor felt about it. A chat can be resolved and still rated badly if it took too long.
Which metrics matter for automated chat or chatbots?
Containment rate, handoff rate, handoff wait time and a separate CSAT for bot-only chats.