Chat queue· Support operations
Chat Queue — The Line Before an Agent Picks Up
Quick answer: A chat queue forms when a visitor asks for a person and nobody can take the chat yet. On most platforms that happens because each agent has a limit on how many chats they can hold at once, and every available agent is at it. The queue then holds the chat in order until a slot opens. What the visitor sees while waiting is a setting: a plain "please wait" message, their place in line, an estimated wait, or a bot that keeps helping in the meantime.
When a chat enters the queue
| Situation | What happens | Example from vendor docs |
|---|---|---|
| Every available agent is at their chat limit | New chats wait in line | LiveChat: customers enter the queue when agents' chat limit has been reached |
| The team picks chats manually | Every new chat waits until someone takes it | LiveChat's manual selection mode sends all chats to the queue |
| Nobody is online in the right team | Depends on the platform: queue, offline message or unassigned inbox | Zendesk does not show a wait estimate when the group or queue is offline |
| A bot is answering first | The chat joins the queue only when the bot hands it to a person | Zendesk shows the wait banner once a ticket is assigned to a group or queue |
The queue is a symptom of capacity, not a feature on its own. Raising chat limits empties it faster but stretches each agent across more conversations; adding people or answering more with a bot shortens it without that trade.
What the visitor can be shown
| Option | What the visitor reads | Vendor example |
|---|---|---|
| Queue position | Their place in line | LiveChat's editable queue message with a place-in-line placeholder; Microsoft's "People ahead of you" and "You're next in line" |
| Estimated wait | A time or range | LiveChat's approximate-minutes placeholder; Zendesk's range in minutes, hours or days |
| Typical reply time | How fast the team usually answers | Intercom's set or dynamic reply time |
| Nothing specific | A generic hold message | Any platform with the options above turned off |
Position is easy to compute and easy to trust. An estimate is more useful but can be wrong, which is why vendors hide it when they lack data.
How vendors estimate the wait
- Zendesk. A statistical model combines historical waits for the customer's place in line with recent waits for the assigned group or queue, using the last 10 minutes, then the last two hours, then the last seven days, whichever has data. Estimates are shown only for groups or queues with at least one ticket routed in the last seven days. The banner refreshes every minute, shows a range, and changes to "An agent will be with you shortly" once an agent accepts. Zendesk lists it for Suite Team plans and up, with messaging turned on, and does not show it when the agent is the only online member of the group.
- Intercom. The dynamic reply time is the median first reply time over the past 7 days, updated daily, and includes time spent with a bot. With fewer than 5 conversations in that period, customers see a generic "as soon as they can" style message instead. Teams using balanced assignment can show a live queue position instead of the estimate. See our Intercom review.
- Microsoft Dynamics 365. Average wait time is an exponential moving average of past connected conversations in the queue, recalculated every 10 minutes; abandoned conversations are excluded. It appears only after at least 20 conversations and when the average is more than one minute. Queue position updates every 30 seconds while the position is under 150.
- LiveChat. The queue message can include the visitor's place in line and the approximate minutes to wait. The article does not say how the minutes are calculated.
None of these pages publishes data on how much showing a position or estimate changes how many visitors stay.
Where chatbots fit
A bot changes the queue in two ways. Before the queue, it answers what it can and sends fewer chats to people, which shortens the line for everyone else; the handoff design guide explains why that effect grows as a desk gets busier. During the queue, it can tell the visitor what is happening, collect details so the agent does not have to ask again, offer an email follow-up, or answer a simple question while they wait. One limit to know: LiveChat states that bots cannot take chats from its queue. See human handoff and the after-hours chatbot for the case when nobody is coming.
How to tell if the queue is a problem
Watch three things: how long queued chats wait before an agent accepts (the human side of first response time), how many visitors leave while queued, and when in the day the queue builds. Long waits at fixed hours point at staffing. Long waits while agents report spare time point at chat limits or status settings. A queue that never forms but chats still feel slow points at agents answering slowly, not at the queue.
FAQ
What is a chat queue? The waiting line for chats no agent can take yet, usually because every available agent is at their limit of chats at once.
Why do customers end up in a chat queue? Because demand at that moment is higher than the number of open chat slots, or because the team picks chats manually.
Should I show customers their queue position or an estimated wait? Position is simple and reliable. An estimate is more helpful when it is accurate; Zendesk and Microsoft both hide theirs when they lack enough recent data. Many teams show position and add an estimate once they trust it.
How is the estimated wait time calculated? It differs by vendor. Zendesk uses recent historical waits for the queue and place in line, Intercom uses the median first reply time over 7 days, and Microsoft uses a moving average of past connected conversations.
What happens to a chat if the visitor leaves the queue? It depends on the platform. LiveChat moves a chat to its unassigned section if the visitor left after sending a message, and sends replies to the visitor's email.
Can a chatbot reduce chat queue times? Yes, by answering simple questions before the handoff and by collecting details while the visitor waits, so the agent's part takes less time.
Related terms
- Chat routing — decides which team's queue a chat joins.
- Live chat — the channel the queue usually serves.
- Chatbot first response time — the wait the queue adds to.
- Human handoff — the step that sends a bot conversation into the queue.
- After-hours chatbot — what to show when nobody is online.
- Typing indicator — another signal that tells a waiting visitor something is happening.
- Chatbot abandonment rate — the related measure of people who give up.
Sources
- LiveChat, Queue configuration — livechat.com/help/queue-configuration/, read 12 November 2026. Source of when customers enter the queue, manual selection, the queue message placeholders for place in line and minutes, the note that bots cannot take queued chats, and unassigned chats when a visitor leaves.
- Zendesk, Displaying the estimated wait time banner in messaging conversations — support.zendesk.com/hc/en-us/articles/8945071149210, read 12 November 2026. Source of the estimation model and data windows, the seven-day requirement, range display, refresh rate, plan list and when the banner is hidden.
- 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, the 7-day median, the 5-conversation fallback 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 and update intervals.
- 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 moving average, 10-minute recalculation, 20-conversation and one-minute thresholds and the exclusion of abandoned conversations.
- Chatbotscape evaluation methodology. /methodology (continuously updated).