Planning guide

How many live chat agents do you need?

Too few agents and visitors wait. Too many and you pay people to watch a quiet queue. You can get a sound first estimate from five numbers, most of which your chat reports already hold. This guide explains the formula, gives a calculator, and shows how to check the answer against what really happens.

The short answer

Agents on shift = chats per hour × average chat length in hours ÷ chats per agent ÷ target occupancy.

For 20 chats an hour lasting 12 minutes each, with agents holding 2 chats at once and a 75% occupancy target, the answer is 20 × 0.2 ÷ 2 ÷ 0.75 = 2.67, so 3 agents on shift. Allow for breaks and leave and you need 4 on the schedule.

Live chat staffing calculator

Enter your own numbers for your busiest hour. The result updates as you type, and nothing is sent anywhere.

Agents needed on shift

3

Agents to schedule

4

The five numbers

NumberWhat it meansWhere to find it
Chats in the busiest hourPlan for the peak, not the daily averageYour chat reports or an hourly heatmap
Average chat lengthFrom the first agent reply to the end of the chat, while the agent is holding itYour chat reports, or a sample of 50 transcripts
ConcurrencyHow many chats one agent handles at the same timeYour own limit. Start with 2
OccupancyThe share of capacity in use. Spare capacity absorbs burstsA planning choice. Start with 70% to 80%
ShrinkagePaid time away from chat: breaks, meetings, training, leaveYour rota. Start with 30%

These starting values are planning assumptions, not industry benchmarks. Replace each one with your own measurement as soon as you have it.

Worked examples

All four rows assume 12-minute chats, 2 chats per agent, 75% occupancy and 30% shrinkage.

Chats in the busiest hourChats open at onceAgents on shiftAgents to schedule
10222
20434
40868
80161116

"Agents to schedule" covers one shift pattern for that hour. If you are open for longer than one shift, multiply by the number of shifts you need to cover.

Choose concurrency carefully

Concurrency looks like free capacity, but it has a cost. A person holding three chats takes longer to answer each one, so chat length rises and part of the gain disappears.

  • Simple, repetitive questions: agents can often hold 3 at once.
  • General support: 2 is a sensible starting point.
  • Technical, billing or sensitive issues: stay at 1 or 2.

Change it in small steps and watch first response time and satisfaction for two weeks after each change. If either gets worse, go back.

Why small teams need more than the formula says

The formula assumes chats arrive evenly. Real chats arrive in clusters, and a small team has less room to absorb them. That is why low occupancy targets suit small teams: at 10 chats an hour the formula gives 1.33 agents, but one agent cannot cover a burst, a break or a sick day.

If chat must be answered during opening hours, keep at least two people able to take chats, even when the numbers say one would do. The second person can do other work and pick up chat when the queue builds.

Reduce the work before you add people

  • Publish better answers. Your most common questions should be answered on the site, before a chat starts.
  • Use saved replies. Canned responses shorten the most frequent chats. These live chat scripts are a good starting set.
  • Route chats to the right team first time. A transfer adds length to the chat and repeats work.
  • Let AI take simple questions. Measure the result honestly, because a chatbot that frustrates people only moves the work. See live chat vs chatbot.
  • Move shifts to match demand. Often the gap is timing, and no extra hiring is needed.

Check the estimate against reality

Run the plan for two to four weeks, then compare it with what visitors experienced:

  • Median first response time during the busiest hour
  • How many visitors left the queue before an agent answered
  • Satisfaction scores in busy hours compared with quiet ones
  • How often agents reached their concurrent-chat limit

In DChat, reports show first response and resolution times by agent and department, and the dashboard shows when chats arrive and the load on each agent. These reports are free. DChat's Community edition has no per-agent fee and allows unlimited agents, so you can add cover at peak times without changing your licence cost. Read more about measuring response time.

Common questions

How many chats can one agent handle at once?

Two is a sensible starting point for general support. Simple repetitive questions may allow three, and technical or sensitive conversations are better at one or two. Measure response time and satisfaction after any change.

What is a good occupancy rate for live chat?

Start with 70 to 80 percent. Higher targets leave little room for bursts, so visitors wait longer. Small teams usually need a lower target than large teams.

What is shrinkage in live chat staffing?

Shrinkage is paid time when agents are not available for chat, such as breaks, meetings, training and leave. If you have no figure of your own, plan with 30 percent and replace it with your real number.

Should I staff for the average or the peak?

Staff for the busiest hours. A daily average hides the times when visitors wait longest. Calculate each part of the day separately if your traffic changes a lot.

Does an AI chatbot reduce the number of agents needed?

It can, if it fully resolves simple questions. Count only the chats it resolves without a follow-up, because the conversations it hands over still need a person.

Self-hosted customer support software

Deploy live chat on your own terms, not on someone else's pricing model.

DChat gives you the installable server, web dashboard, website widget, and desktop agent tools in one self-hosted product, with live chat free to run, and an AI agent that answers, acts with the approvals you set and hands off to your team when you want one. Run it on infrastructure you trust, on the model you choose.

Deployment

Install on Windows or Linux, behind IIS or Nginx, in a VM, or in Docker if that fits your stack.

Commercial model

Live chat is free with unlimited agents. Premium, with the AI agent, is billed per agent per month, never per conversation.

AI Flexibility

Run the AI agent on a local Ollama model, or connect OpenAI and Anthropic with your own provider accounts.