
How to Manage a Live Chat Queue
Quick answer: Find out when your queue builds before changing anything, then set each agent's chat limit on purpose. Show waiting visitors their place in line, and add an estimated wait only once your platform has enough recent data to make it honest. Write a waiting message that says what happens next, let the chatbot collect details or answer simple questions during the wait, offer an email follow-up for anyone who cannot stay, and count how many people leave the queue each week.
Step 1: Find when the queue builds
Pull two weeks of chats and look at them by hour of day and day of week. You are looking for the hours where chats waited before an agent accepted them. Most small teams find the queue is not all day; it is a lunchtime peak, a Monday morning, or the hour after a marketing email goes out.
Write down three things for each peak: how many chats arrived, how many agents were online, and how long the longest waits were. This tells you whether the fix is people (shift times, a second person at lunch), settings (limits, status) or fewer chats (better bot answers). If the first response time report on your platform mixes bot replies and human replies, separate them first; only the human wait is the queue.
Step 2: Set chat limits on purpose
On most platforms the queue starts when every available agent reaches their limit of chats at once. LiveChat says this directly: customers enter the queue when agents' chat limit has been reached, and the limit is set per agent in the team settings. That makes the limit the main dial you control.
- Too low and visitors queue while agents have room.
- Too high and nobody queues, but each person is juggling more conversations, replies slow down and mistakes rise.
No vendor page read for this guide recommends a number. Start at a level your team says they can handle well, then change it one step at a time and watch both the queue and reply quality. Agents who handle long, careful conversations, such as billing, may need a lower limit than agents answering quick questions. Also check status settings: an agent left on "accepting chats" while at lunch makes the queue look shorter than it is.
Step 3: Show position, then an estimate you can trust
Visitors cope better with a wait they can see. You usually have three options:
| Option | Strength | Weak spot |
|---|---|---|
| Queue position | Simple and always accurate at that moment | Says nothing about speed |
| Estimated wait | Most useful when right | Wrong after sudden spikes or with little data |
| Typical reply time | Sets expectations before the chat starts | An average, not this visitor's wait |
Vendor tools differ in what they offer:
- LiveChat lets you edit the queue message and include placeholders for the visitor's place in line and the approximate minutes to wait. The placeholders must stay in the message for them to show.
- Zendesk can show an estimated wait, a queue position or both in messaging. It shows a range, refreshes every minute, and hides the estimate for queues without at least one routed ticket in the last seven days. It warns that accuracy drops after sudden spikes or when agent availability changes.
- Intercom offers set reply times or a dynamic one based on the median first reply time over 7 days; teams using balanced assignment can show a live queue position instead.
- Microsoft Dynamics 365 shows "People ahead of you" or "You're next in line", and only shows an average wait after 20 conversations and when the average is more than one minute.
A safe default for a small team: show position from day one, and turn on an estimate once you have checked it against real waits for a week.
Step 4: Write the waiting message
The waiting message is the one thing every queued visitor reads. Keep it short and answer three questions: are you in line, roughly how long, and what happens if you leave.
You're number %number% in line. We usually reply within a few minutes. You can keep typing your question here — the next agent will see it. If you need to go, leave your email and we'll reply there.
The placeholder shown is LiveChat's; other platforms use their own. Two details from vendor docs are worth building on. LiveChat lets visitors keep sending messages while queued, and agents can see them, so inviting the visitor to describe the problem saves time later. If a queued visitor leaves after sending a message, LiveChat moves the chat to the unassigned section and sends replies to the visitor's email. Tell visitors that, so leaving does not feel like starting over.
Avoid promising a time you cannot keep. "Usually within a few minutes" is honest when it is true at your peak; a fixed number is not. Choosing a response target you can keep is covered in setting live chat SLA targets.
Step 5: Let the bot help during the wait
A chatbot can shorten both the queue and each conversation in it:
- Before the queue. Answer the questions that do not need a person, such as order status or opening hours, so fewer chats join the line. The handoff design guide shows why cutting arrivals shortens waits by more than you would expect.
- On entry. Ask for the order number, account email or topic, so the agent starts with the facts. A short pre-chat form does the same for chats without a bot.
- While waiting. Offer help articles that match the topic, but do not pull the visitor out of line if they keep reading.
- On handoff. Pass a short summary so the agent does not ask again; see summarizing conversations for handoff.
Check your platform first. LiveChat states that bots cannot take chats from its queue and suggests automatic assignment for teams using its bot product, so the bot's work has to happen before the chat is queued.
Step 6: Offer an exit and measure who leaves
Some visitors cannot wait. Give them a clear way out that does not lose the conversation: an email follow-up, a ticket, or a callback if you offer one. When the queue is long or the team is about to go offline, say so in the waiting message, and switch to your after-hours chatbot setup when nobody is coming.
Track three numbers weekly: visitors who left while queued, the longest wait in your peak hour, and chats answered by email after the visitor left. Compare weeks with similar traffic before deciding a change worked. For the wider set of numbers, see chatbot metrics.
Check the arithmetic
An invented example, not a measurement. A shop has three agents with a limit of 3 chats each, so the team can hold 3 × 3 = 9 chats at once. On a Tuesday lunchtime, 40 visitors asked for a person between 12:00 and 13:00, and 10 left while queued: 10 ÷ 40 = 25%. The shop adds a waiting message with queue position and an email option, and sets the bot to answer order-status questions before the handoff. The next Tuesday, 40 visitors asked for help again; the bot resolved 8, so 40 − 8 = 32 joined the queue, and 4 of those left while waiting: 4 ÷ 32 = 12.5%. Two lunchtimes do not prove cause, so keep watching for a few weeks and read the chats of people who still left. These numbers are made up to show the method.
Common mistakes
- Raising limits until the queue disappears, then wondering why replies got slower.
- Showing an estimate before checking it against real waits.
- A waiting message with no exit, so visitors who leave lose their question.
- Agents left on "accepting chats" during breaks.
- Measuring the bot's instant reply as if it were the customer's wait.
- Fixing the queue with settings when the real cause is one peak hour with one person on shift.
FAQ
How do I reduce live chat wait times? Find the hours when the queue builds, then cover them with staff, set chat limits deliberately and let a bot answer simple questions before the handoff.
Should I show customers their queue position? Yes, in most cases. Position is accurate and easy to understand. Add an estimated wait once your platform has enough data and you have checked it.
How many chats should one agent handle at once? No vendor page read for this guide gives a recommended number. Start at a level your agents can handle well, change it one step at a time and watch reply quality and waits.
What should a live chat waiting message say? That the visitor is in line, roughly how long it usually takes, that they can keep typing, and how they will get a reply if they leave.
What do I do when the queue is too long? Tell visitors honestly, offer an email follow-up or ticket, and let the bot answer what it can. If it happens at the same hour every week, change the schedule.
Can a chatbot take chats from the queue? It depends on the platform. LiveChat says bots cannot take chats from its queue, so the bot works before the chat is queued.
Sources
- LiveChat, Queue configuration — livechat.com/help/queue-configuration/, read 12 November 2026. Source of when customers enter the queue, per-agent chat limits, queue message placeholders, messaging while queued, unassigned chats after a visitor leaves and the note that bots cannot take queued chats.
- Zendesk, Displaying the estimated wait time banner in messaging conversations — support.zendesk.com/hc/en-us/articles/8945071149210, read 12 November 2026. Source of wait estimate and position options, range display, refresh rate, the seven-day data requirement and accuracy warnings.
- Intercom, Share your expected response time — intercom.com/help/en/articles/732436-share-your-expected-response-time, read 12 November 2026. Source of set and dynamic reply times and queue position for balanced assignment teams.
- Microsoft Learn, Show customers their position in a queue — learn.microsoft.com/en-us/dynamics365/customer-service/show-queue, read 12 November 2026. Source of the position messages.
- Microsoft Learn, Show customers their average wait time in a queue — learn.microsoft.com/en-us/dynamics365/customer-service/average-wait-time, read 12 November 2026. Source of the 20-conversation and one-minute thresholds.
- Chatbotscape evaluation methodology. /methodology (continuously updated).