What it means
A customer can ask a question without leaving the page or making a call, and the team can share links and keep a written record. A chat window is only helpful when response expectations are clear, so set operating hours and tell visitors when agents are offline.
A fictional company that closes chat at 6 p.m. shows that messages will be answered next business day, and it does not display "agent online" after the shift ends. Zendesk discusses live chat support practices, and Intercom documents workload controls for team inboxes, which shows why routing and staffing matter alongside the widget itself.
Route conversations by subject, language or account status where appropriate, because a wrong assignment adds transfers and delay. In a fictional refund case, the first agent hands the chat to billing with the order details, so the customer does not have to repeat the story from scratch.
Concurrent chats can improve capacity, but too many open conversations reduce attention. Intercom's workload guidance describes assignment and capacity controls, and limits should reflect the complexity of the work.
A fictional support worker can handle several simple delivery questions at once, whereas a disputed charge requires focused review and should not count as an identical work unit. Bots can answer routine questions or gather context before a handoff, but it should be clear when the user is speaking with automation and there should be a path to a human for exceptions.
A fictional travel site asks for a booking reference in chat and routes to an agent if the automated lookup fails, rather than looping through the same question indefinitely. An automated acknowledgement is not a human response, and a promised timeframe must be one the team can meet.
Authenticate before discussing private account details, because being in a chat window does not prove identity. Avoid asking for passwords, full card details or unnecessary personal information, and follow the company's approved verification process even when a customer supplies an account email.
A chat transcript can help quality review and continuity, but retention and access should follow privacy rules and policy, and a fictional clinic using chat for appointment administration redirects urgent medical concerns to the appropriate clinical route. Measure first response, resolution and customer feedback together, since a quick greeting followed by long silence is not good service.
A fictional team reports a ten-second first response from a bot while human replies take twenty minutes, so its dashboard separates the two measures. Integrate chat with tickets and CRM records carefully, assign ownership of unresolved chats and confirm the case was handed over, because ending the chat window does not end the problem, and live chat is a service promise as much as a technology choice.
In practice
Real-world examples.
Example
An online shopper asks about delivery while checking out. An agent answers with the current delivery policy and the shopper completes the purchase. The conversation is saved against the order for later reference.
Example
A bot gathers an order number and hands the case to an agent. The agent sees the details already collected and does not ask for them again. The customer receives an answer in a single conversation.
Example
An unresolved chat becomes an owned follow-up ticket. The support team records the case owner and the date by which the customer will hear back. The customer is updated through the agreed channel after the chat window closes.
Formula
Calculation
Human first-response time = time of first human reply - time the customer started the conversation, under a stated operating-hours rule.
Worked example: a customer opens a chat at 10:00 and a bot greets them at 10:00 and 10 seconds. A human agent replies at 10:20, so human first-response time is 10:20 - 10:00 = 20 minutes, not 10 seconds. If the team handles 200 chats in a day and 150 receive a human reply within 20 minutes, 150 / 200 = 75% meet that target, which is a more honest service measure than the bot's speed.Case study
Seen in the real world.
In this fictional case, Oak Retail launches chat and sees many short questions answered quickly. Refund cases still wait too long. It routes refund requests to trained staff, sets honest availability and measures resolution rather than celebrating bot greetings alone. Oak then reviews its transcripts each month and finds that most long waits come from questions that need an order lookup.
It adds the order number to the first screen so agents can see it immediately, and it trains two more agents on refunds. After three months, the team reports human first response and resolution separately, so management can see where customers still wait. The company is invented, and the story is illustrative only.
Watch out
Common mistakes.
- Claiming a bot greeting is a human response.
- Opening chat without staffing or clear offline expectations.
- Sharing account data before proper verification.
Questions
People also ask.
Is live chat always staffed by a person?
No. Some services use bots or are offline at certain times.
Can agents handle several chats at once?
Sometimes, but capacity should match issue complexity.
What should be measured?
Human response, resolution, transfers and customer feedback.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%