Skip to content
Chatbotscape
Editorial flat-vector illustration for How to Set Up a Chatbot for After-Hours Customer Messages
Published 6 min read

How to Set Up a Chatbot for After-Hours Customer Messages

Quick answer: Tell the bot what time it is and what that changes. Set a real schedule with holidays, let the bot answer only what your content covers, collect the details a person needs, write a closed message that names a reply time you can keep, give urgent cases a separate path, and test the bot at the edges of the schedule. Most of the work is deciding, not configuring.

Step 1: Set a real schedule, including holidays and time zone

Write down the hours a person will actually read messages, not the hours on your website. Then enter them in your platform. Intercom's help article on default office hours says that if no hours are set the workspace is treated as available around the clock, so a missing schedule means the bot will act as if someone is always there. The same article describes 30-minute increments, automatic daylight saving adjustment and holiday hours that can be one-time or recurring; check what your own platform offers.

Decide three things in writing:

  1. Time zone. One clock for the whole team, or one per team?
  2. Holidays. Who updates them, and by when each year?
  3. Weekends. Closed, or a short window?

Step 2: Decide what the bot may answer without a person

After hours the bot should still answer what your content covers: hours, prices, order status, how-to steps. List the question types it handles alone and the ones that need a person. If a question needs a human decision (a refund, an exception, a complaint), the bot should not improvise; see when not to use a chatbot and the chatbot guardrails guide.

Step 3: Collect what the morning agent needs

For anything left over, ask a short set of questions, then stop. A useful default:

  • A contact detail (email or phone)
  • An order or account number, if relevant
  • One sentence on the problem
  • Whether it is urgent (see Step 5)

Keep it to three or four questions; each one loses some people. Pass the answers to your ticketing or inbox tool, ideally with a short recap; summarizing chatbot conversations for handoff shows how.

Step 4: Write an honest closed message

The message does the work an away message does, plus one thing: it names a reply time you can keep. Zendesk's legacy business-hours workflow article shows the pattern: check the schedule, transfer to an agent when open, and otherwise tell the user the business is closed and when to return. Its example types the hours into the message by hand, so a changed schedule needs a changed message too.

A template to adapt:

Our team is offline right now. I can still help with [what the bot covers]. If you leave your email and a short note, someone will reply by [time, in the customer's time zone if you can].

Rules of thumb: no "instantly", no promise of a person tonight, and no pretending to be a person. Wording ideas for failed answers are in chatbot fallback copywriting.

Step 5: Give urgent cases a separate path

Not everything can wait until morning. Define two or three urgent cases (a safety issue, an outage, a payment taken twice) and tell the bot to recognize them. Then give each a route that works at night: a phone number that rings, an on-call alert, or an emergency instruction. If you have no such route, say so in the message rather than implying one. Trigger design is covered in the chatbot escalation playbook.

Step 6: Test the clock

Bugs hide at the edges. Test with a saved list:

TestExpected result
Message at 1 minute before closingHandoff offered
Message at 1 minute after closingClosed message, details collected
Message on a holidayClosed message with holiday wording
Message in a different time zoneReply time stated in a way that makes sense
"My payment was taken twice" at midnightUrgent path
Question the knowledge base answersAnswered, no ticket

Run it after every schedule or wording change. The chatbot QA testing protocol has a fuller routine. Afterward watch two numbers: how many after-hours chats become tickets, and how long those tickets wait for a first reply (see first response time).

Check the arithmetic

An invented example, not a measurement. A small shop gets 200 chats a week, 60 of them outside hours. The bot answers 36 of the 60 alone: 36 ÷ 60 = 60%. The remaining 24 become tickets with contact details. Before the setup, all 60 got a single away message and a share of people left without writing; after it, 24 tickets arrive with a reason attached.

If the shop opens at 9:00 and the closed message promises a reply "by 11:00", 24 tickets at about 5 minutes each is 120 minutes, which is two hours of work for one person, so the 11:00 promise is only safe if someone starts at 9:00 sharp. A tighter promise would be "by noon". These numbers are made up to show the method.

Common mistakes

  • No schedule entered, so the bot behaves as always-open.
  • Hours typed into the message and never updated.
  • Promising a person tonight when nobody will read it.
  • Asking ten questions before taking a message.
  • No urgent path.
  • Testing only at noon.

FAQ

How do I set up a chatbot for after-hours customer messages? Enter a real schedule, decide what the bot may answer alone, collect a few details for the rest, write a closed message with a reply time you can keep, give urgent cases a separate path, and test the edges of the schedule.

What should an after-hours chatbot say? That the team is offline, what the bot can still help with, and when a person will reply. Do not promise what you cannot keep.

Can a chatbot hand off to a human only during working hours? Yes, in platforms that support a schedule condition. Zendesk's legacy workflow article, for example, branches on a business-hours schedule; check what your own platform supports.

What information should the bot collect after hours? A contact detail, an order number if relevant, and a sentence about the problem. Keep it to three or four questions.

Does the bot need to say it is automated? Plain wording that the team is away avoids false expectations. Disclosure rules vary by place, so check your own.

Is an away message enough? For a very small business with few after-hours chats, perhaps. Once chats pile up, a bot that answers and collects details saves the morning shift real time.

Sources