Guide

Live chat response time: measure it, then cut it

A visitor who opens a chat is waiting with the page still in front of them. How long they wait for the first reply is the number they notice most. This guide shows how to measure it honestly, set a target you can keep, and bring it down, using DChat as the worked example.

1. Measure the right number

First response time is the time from the visitor's first message to the first reply from a person. It is the number to start with. Two others matter later: the time between later replies, and the time until the conversation is resolved.

Look at the median as well as the average. A few chats left waiting over lunch can pull the average up while most visitors were answered quickly. The median tells you what a typical visitor saw; the average tells you whether some were left far too long.

In DChat, the Reports page shows average and median first-response time, with resolution time and CSAT beside them, broken down by agent, department, inbox and label, for any date range. Every report exports to CSV. These reports are free.

2. Count only the hours you are open

A chat that arrives at 11 pm and is answered at 9 am was not a ten-hour failure; you were closed. Set your operating hours first, so the clock only runs while you are open, and so the visitor is told when to expect an answer.

DChat lets you set business hours per inbox, with their own time zone, split shifts and holidays. Outside those hours an automatic away reply goes out on every channel, once per closed period, and a waiting visitor can leave an email address so the answer reaches them later. All of that is free.

3. Set a target you can keep

Published averages for chat response time vary widely, because each vendor measures differently. Do not copy one. Measure your own median for two to four weeks, then set a first-response target a little better than it, and track how often you meet it. Raise the bar once you meet it most of the time.

DChat's first-response SLA target is free on every edition. It respects your operating hours, marks conversations that missed it in the list, and has its own report: the share of conversations that met the target, with misses by agent and department. Next-response and resolution targets per department and priority, with breach alerts, are part of Enterprise.

4. Find where the time goes

Slow first replies usually have one of four causes. The report tells you which.

  • Nobody noticed. Agents work in other tabs. Browser push notifications and the Windows desktop app alert them to a new chat.
  • The chat sat in the wrong queue. Departments decide which queue a visitor waits in; teams let you hand a conversation to the right group from any channel. Both are free.
  • One agent had too many chats. A concurrent-chat limit stops one person being handed more than they can hold. One limit for everyone is free; a limit per agent, and auto-assignment to whoever is free and under their limit, are part of Premium.
  • Not enough people at busy hours. The dashboard heatmap shows when chats arrive. Move shifts to match, or cover the peak with the AI agent for common questions.

5. Make the first reply faster to write

The first reply is often the same few sentences. Saved replies make it a few keystrokes. DChat's canned responses are free and can insert the visitor's name or other details; a field with no value is highlighted for you to fill, never sent blank. The guide to canned responses covers how to write ones that do not sound canned.

A quick acknowledgement followed by a real answer is better than a long silence followed by a perfect one. Say hello, say you are looking, then look.

6. Let the AI take the simple questions, carefully

An AI agent answers at once, at any hour. It helps most with questions your help articles already answer: opening hours, delivery, returns, how to reset a password. It should hand anything else to a person, with the conversation attached, so the visitor does not repeat themselves.

DChat's AI agent is part of Premium, at $19 per covered agent per month, on a local Ollama model or your own OpenAI or Anthropic key. It is off until you turn it on. Read when not to use an AI chatbot and how to measure one before you rely on it.

7. Review it every week

Response time drifts as traffic, staff and products change. Look at it weekly, not only when someone complains. DChat can email a daily or weekly team summary with response and resolution times, CSAT and SLA misses, at each recipient's own hour, through your own mail server. It is free.

Common questions

What is a good live chat response time?

There is no single right number, and published averages are measured in different ways. Measure your own first response time for a few weeks, then set a target a little better than your median and track how often you meet it. A target you meet most of the time is more useful than an ambitious one you miss.

Should response time count hours when we are closed?

No. A chat that arrives at night and is answered at nine the next morning was not slow; you were closed. Count response time within your operating hours, and send an away reply outside them so the visitor knows when to expect an answer.

Does DChat measure first response time for free?

Yes. Reports show average and median first-response time by agent, department, inbox and label, and a first-response SLA target with its report and dashboard panel is free on every edition. Next-response and resolution targets with breach alerts are part of Enterprise.

Will an AI chatbot fix slow response times?

It can answer common questions at once, at any hour, and hand the rest to a person. It does not fix a queue with too few agents. Measure first, fix staffing and routing, then add the bot for the questions your help articles already answer.

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.