Auto reply messages that do not read as auto replies

This page is about one kind of auto reply: the single fixed message, written once and sent to everyone who triggers it, which is what WhatsApp’s away message and greeting are. It is not the only kind. A keyword router, a menu of tapped questions and a model that reads what somebody actually wrote are auto replies too, and they fail in other ways. For the fixed kind the list of things you can get wrong is very short: what it promises, and whether the promise is true. The examples below are ours to copy, but the part worth copying is the method, because the sentence that works for a salon at ten at night is the wrong sentence for a plumber.

The three things it should contain

A time a human will answer. One thing the reader can usefully do before then. Nothing else.

WhatsApp describes its own away message in those same terms. The Business app features page, read by us on 19 September 2026, says it is there “to set expectations and let customers know when you’ll be back”. Setting an expectation, not answering a question. A fixed message cannot tell somebody what a job costs, whether Saturday morning is free, or whether you cover their postcode, because it was written before they asked.

The generic version fails on all three counts:

Thanks for your message! Your enquiry is very important to us and a member of our team will get back to you as soon as possible.

No time in it, nothing to do, two sentences of filler. Read as a customer, “as soon as possible” is not a commitment, it is a way of avoiding one.

What a hollow one costs, in Meta’s own words

Meta’s Service messages page publishes guidelines for what it calls sending high-quality messages. That page documents the Cloud API rather than the Business app, so treat what follows as Meta telling you what it considers a bad message rather than as a rule binding an away message you set on your phone. Two of the guidelines are directly about this:

Avoid sending open-ended welcome or introductory messages.

Make the messages highly personalized and useful to users.

A third is “Optimize your messages for content and length.” The same page explains why that guidance has teeth:

WhatsApp determines message quality from how WhatsApp users have received your messages over the past seven days, weighted by recency. It bases this score on user feedback signals such as blocks, reports, mutes, archives, and reasons users provide when they block you.

So a thank-you-for-contacting-us message is not just weak writing. It is the shape Meta’s own guidance tells an API sender to avoid, on a platform that scores that sender on whether people mute and block you. If you are in the app rather than on the API, read it as advice you have a reason to take rather than as a rule you are breaking.

The failure mode is the promise, not the tone

This is the one that costs real customers:

Thanks for your message! We reply to every enquiry within five minutes, 24 hours a day.

Nothing in that sentence reads robotic. It fails because it is a deadline you set for yourself in writing, and the first time you miss it the message has taught the reader that your promises are decorative. Pick the time you would still meet on your worst day, then say that.

We have not found a primary source for how many minutes a customer will wait before giving up, so no figure for it appears here. The test is not whether you beat somebody’s benchmark. It is whether the time you wrote down is the time you keep.

Four worked examples, and what each one is doing

A salon, after closing:

We are closed now. I read messages when we open at nine in the morning and I will answer yours then. If you want a slot this week, the booking link in our profile works overnight and you will not need me for it.

The time is an opening time, which is the one time of day you cannot get wrong. The second half hands over something that works while you sleep.

A plumber overnight:

Nobody is watching this phone overnight and I will reply from seven. If water is running and you cannot stop it, turn the stopcock off and call the number in our profile, because that one wakes somebody and this one does not.

The example assumes one genuine emergency case, so the message triages it and says plainly which channel is monitored. Naming what this inbox cannot do is what stops the real emergency sitting in it until morning.

A dental practice, during opening hours:

Thanks for messaging the practice. Reception reads WhatsApp between nine and five and answers the same day, usually sooner. If you are in pain today, say so in your next message and we will call you rather than type.

This is a greeting rather than an away message, and the two are documented differently. Meta’s features page describes the greeting as sending “a custom welcome message to new customers when they start a conversation with your business”, and describes the away message separately as replying automatically “to messages received outside your business hours”. The trigger it states for the greeting is a new conversation. It states no schedule for it, so nothing on that page tells you the greeting fires only while you are open, and an example written for use during hours has to stand on its own wording rather than on the feature doing the timing. Which is why this one names the hours it is claiming and asks a question the answer to which routes the person.

A restaurant, mid-service:

We are in service and will reply after nine tonight. For a table in the next two hours, please call instead, because the phone gets answered and this inbox does not until we close.

Same move as the plumber, for the case where the enquiry has a deadline of its own.

Why the honest version is also the cheap version

Promising an answer at opening time is not only the most truthful thing you can write, it is the thing the platform is built to let you do. Which platform, though, matters: the away message is a WhatsApp Business app feature and the 24-hour window is a Cloud API rule, and mixing the two produces advice that is wrong on both. So here they are separately.

In the WhatsApp Business app

The away message and the greeting are app features, and messages sent from the app are free. Meta’s onboarding documentation, describing a business running the app alongside its API, states it directly: “messages sent by the business via the WhatsApp Business app will continue to be free”. The away message carries a schedule tied to your opening hours, since the feature is described as replying automatically “to messages received outside your business hours”.

The window rules do not reach these messages at all, and Meta says so on the same onboarding page:

The 24-hour customer service window restriction applies to messages sent via Cloud API. Messages sent from the WhatsApp Business app are not subject to the customer service window and do not create, extend, or affect Cloud API conversation windows or Cloud API pricing.

So if your auto reply is the one in the app, nothing below applies to it. No window closes on it, no template is required, and the paragraph after this one is not about you.

On the Cloud API

If something else sends for you, you are on Meta’s API and the window governs. A customer messaging you opens a 24-hour one, and the Business Messaging Policy states what you may send inside it: “You may reply to a user message without use of a Message Template as long as it’s within 24 hours of the last user message”. Meta’s Service messages page adds that “service messages do not require pre-approval” and that “While the window is open, you can send any of the following service message types to the user.”

So a promise to reply by morning is one the API lets you keep, in your own words, with nobody approving it first. A promise to follow up later in the week is different machinery: “When the window closes, you can only send pre-approved template messages”.

mesej never messages anyone first, which is why every example on this page is a reply and not a reach-out. No channel is connected to it today, so nothing here has answered a customer.

Four checks before you switch it on

• Tie the schedule to your hours, not to a date. The away message that ages worst is the seasonal one: “We are back Monday!” still going out on Thursday.
• One promise per message. Two different times in one auto reply is a message that contradicts itself.
• Name only a channel that is answered. Pointing at a phone nobody picks up moves the problem rather than solving it.
• Put the real answers somewhere reusable. The features page describes quick replies as a way to “Create shortcuts for messages you frequently send to respond to common questions in just a few taps.” The away message buys the time; those carry the content.

The difference between an auto reply and a message that reads like one is not warmth or an exclamation mark. It is a specific time and a specific next step, short enough that both survive being read at eleven at night. Sounding like a person rather than a process is the same discipline as replying to a bad review in public.