The live chat metrics that matter for a small business fit on one page: how many chats you get, how many arrive when nobody's online, how quickly visitors get a reply, how quickly offline messages get followed up, how many leads chat produces, how often the chatbot hands off to a person, and how satisfied visitors are. Track them monthly in a simple spreadsheet, compare each month with your own previous months rather than with industry benchmarks, and use them to decide one change at a time.
Here's what each metric tells you, how to collect it without a data team, and what to do with it.
Why fewer metrics is better
Large support teams track dozens of numbers. A small team doesn't need to. Every metric you track takes time to collect, and a number nobody acts on is just a chore. Choose metrics that answer three questions:
- Are visitors getting a timely answer?
- Is chat producing leads and solving problems?
- Is the chatbot helping or getting in the way?
The core live chat metrics
| Metric | What it tells you | How to get it without a data team |
|---|---|---|
| Chat volume | How much demand chat is handling | Count conversations per week or month in your inbox |
| Chats outside office hours | Whether your hours match when visitors arrive | Count conversations that started while nobody was online |
| Time to first reply | How long visitors wait for a person | Sample conversations and note the wait, if your tool shows message times |
| Offline follow-up time | How quickly you reply to messages left after hours | Note when you replied versus when the message arrived |
| Leads captured | Whether chat feeds sales | Count contacts with an email who asked about buying |
| Handoff rate | How often the bot passes chats to a person | Count chats that reached a person versus the total |
| Satisfaction ratings | How visitors felt about the conversation | Use your tool's ratings, if it collects them |
| Top topics | What people actually ask about | Tally the topic of a sample of chats |
You don't need all eight on day one. Volume, offline chats, follow-up time and top topics are enough to start.
Speed: the metric that matters most for leads
How fast you respond is closely tied to whether a lead goes anywhere. In a Harvard Business Review study of 2,241 US companies, firms that tried to contact web leads within an hour were nearly seven times as likely to qualify them as firms that waited even an hour longer.
For chat, track two speeds separately:
- Live response: when someone is online, how long does a waiting visitor wait?
- Offline follow-up: when nobody was online, how long until you replied by email?
The second is often where small teams lose the most, and it's the easiest to fix with a morning routine.
Bot metrics: is the chatbot helping?
If you use a chatbot, watch where conversations end:
- Handoff rate. A very low rate could mean the bot answers well, or that visitors can't find the way to a person. A very high rate may mean the bot isn't answering what people ask. Read transcripts to tell which.
- Where conversations stop. If many chats end at the same question, shorten, reword or remove it.
- Early requests for a person. If visitors ask for a human on the first message, your greeting may promise the wrong thing.
- AI handoffs, if you use an AI assistant. Each one is a hint about missing content.
Why we don't give benchmark numbers
You'll find plenty of "average chat response time" and "average satisfaction" figures online. They're rarely comparable to your situation: they mix industries, team sizes, definitions and tools, and many aren't traceable to a published method. Your own trend is a better guide. If time to follow up drops from "next afternoon" to "first thing next morning," that's real progress, whatever an industry average says.
A one-page monthly scorecard
Copy this into a spreadsheet. The values below are an illustrative example, not targets.
| Metric | This month (example) | Last month (example) | Change we'll make |
|---|---|---|---|
| Chats | 140 | 120 | None needed |
| Chats while offline | 45 | 50 | Consider covering one more evening hour |
| Offline follow-up time | Next morning | Next afternoon | Keep the morning routine |
| Leads with an email | 22 | 18 | Test a shorter qualification flow |
| Handoff rate | About half | About two-thirds | Add two answers to the knowledge base |
| Most common topic | Delivery times | Pricing | Add delivery times to the greeting buttons |
Fill it in once a month, in under an hour.
How to collect the numbers quickly
- Sample, don't count everything. Reading 20 conversations a month tells you more about topics and quality than any dashboard.
- Use your inbox filters and search to count by status and find topics.
- Export contacts to a spreadsheet if your plan allows it, and count leads there.
- Track chat opens in the analytics you already use. A developer can listen for the chat widget's events and record them, so you can see which pages lead to chats.
- Write down what you changed each month, so you can connect changes to results.
Turning metrics into changes
Numbers are only useful if they lead to a decision. Common patterns:
| If you see | Try |
|---|---|
| Many chats while nobody's online | Adjust office hours, or improve offline handling and chatbot answers |
| Slow offline follow-up | A first-thing-in-the-morning routine with a named owner |
| Visitors leaving at one bot question | Shorten, reword or remove that question |
| The same topic every week | Answer it on your website, in the bot, or in your AI knowledge base |
| Low ratings on a type of chat | Read those chats together as a team and agree on a better approach |
| Few leads despite many chats | Add a short qualification step and a clear email request |
Change one thing at a time, then check the next month's numbers.
How to do this in RetroChat
RetroChat's shared inbox shows each conversation as Bot, Waiting, Open or Closed, with filters by status and search by name, email or message, which makes monthly counts and topic sampling quick. Contact profiles show the answers the chatbot collected, the page the visitor chatted from and their location. On Team plans and above you can export contacts to CSV; on Business and Autopilot you also get reports and visitor ratings after closed chats. See Plans and billing for what each plan includes, or compare them on our pricing page.
For developers, the JavaScript API's RetroChat.on('open' | 'close' | 'message', callback) runs your own code on those events, for example to record chat opens in the analytics tool you already use. For keeping the day-to-day queue under control, see our guide to a shared inbox for live chat, and for the full setup, our guide to adding live chat to your website.
Start your scorecard this month
Pick four metrics, fill in the scorecard once a month, and make one change at a time. You can start collecting real numbers in a free 14-day RetroChat trial, no credit card required.
Frequently asked questions
What are the most important live chat metrics?
For a small business: chat volume, chats that arrive while nobody's online, time to first reply, offline follow-up time, leads captured, chatbot handoff rate and satisfaction ratings. Start with three or four and add more only if you'll act on them.
What is a good live chat response time?
There's no single right number; it depends on your business and staffing. Promise a response time you can keep, measure against it, and improve your own trend. For leads, faster is clearly better, since research on web leads found quick follow-up made qualification much more likely.
How do I measure live chat without expensive tools?
Use your chat inbox's filters and search, sample around 20 conversations a month, export contacts to a spreadsheet if your plan allows it, and record everything on a one-page monthly scorecard. That's enough to spot trends and decide what to change.
How do I know if my chatbot is working?
Look at where conversations end, how often the bot hands off to a person and why, and whether visitors ask for a human straight away. Read transcripts alongside the numbers, because the same handoff rate can mean very different things.