Which WhatsApp Messages Do Customers Reply To?

Customers reply when a message asks one easy question

On WhatsApp, reading is not the problem; replying is. WhatsApp messages to customers get a reply when they:

• Are about the customer's own thing: their order, delivery, appointment, quote or question.
• Say who is writing and why in the first line: "[store] about your washing machine order", not "Dear customer".
• Ask one question, or one thing the customer can do in the chat: a tap, a few words or a photo.
• Arrive when the customer can act, and when your team can answer back.

When the customer replies, the conversation continues only if someone answers back quickly.

Not every message needs a reply. When one does, measure replies and the outcome you need, not only reads. The test in this guide is an example, not DRING data or industry data.

Why do customers read a message and not reply?

Usually because the message gives them nothing easy to reply to. Reading takes a second; replying takes a decision.

A message is read when it shows as read in WhatsApp. That counts only customers with read receipts turned on, so the real number of readers can be higher. A reply is when the customer writes back or taps a button.

The usual reasons:

• No question. The message is a notice: it informs and asks no question, even when a decision is needed.
• Too many questions. It asks for the day, the address and the floor at once.
• It sounds like a mass message. "Dear customer" tells the customer that nobody is waiting for an answer.
• It asks for work outside the chat. "Visit this link" or "call us" turns a reply into a task.
• It arrives at the wrong time: late at night, on a day off, or days after the event.
• Nobody answered last time, so the customer expects no answer now.

What should the first message say?

Who is writing, why now, and one question the customer can answer in one tap. The first message opens the conversation from the business side.

• Who is writing: your business name in the first line, as in "Hi [name], this is [store]".
• Why now: the customer's own thing, as in "about your washing machine order".
• One question: a single question with a clear set of answers, such as "We can deliver it on Thursday between 10:00 and 13:00. Does that work for you?"
• Reply in one tap: the answers sit under the message as buttons the customer taps, here "Yes, Thursday" and "Another day".

A customer's message or call opens a customer service window: for 24 hours, you can send that customer any normal message, and each new message or call starts it again. When the window closes, you can only send a template, a message approved by Meta in advance (see Meta's guide to sending WhatsApp messages). So a first message to a customer who has not written or called in the last 24 hours is a template.

How many questions should one message ask?

One. Every question asks for a decision, and a message with three decisions is easy to leave for later.

Ask for a choice, not an essay:

• Two or three answers as buttons. In a WhatsApp reply buttons message, Meta allows up to three buttons, each label at most 20 characters (see Meta's reply buttons guide).
• Numbered options where buttons are not available: "Reply 1 to confirm, 2 to change the time."
• The next question in the next message, once the customer has replied.

Do not end with "Let us know if you have any questions" in place of a question: it leaves the customer to invent one. Ask the question you need answered.

When is a customer most likely to reply?

Soon after the event they are waiting for, in their daytime hours, and when your team can answer back. These points are advice, not rules.

• Right after the event: the order is ready, the quote is sent, the call was missed. A message about something days old is easier to leave unanswered.
• In the customer's daytime, in their time zone, not late at night. Working days differ by market too: in much of the Gulf, the weekend starts on Friday.
• When someone can answer back within minutes. A customer who replies at 21:00 and hears back at 09:00 may have moved on by then.

The store in the example test sends at 10:00 on a working day, while its team is there to answer.

How do you test which message works?

Split one batch of customers at random and send each half one version, at the same time. Then compare reads, replies and the outcome you need.

Example, not real data. An online home appliance store has 600 orders to deliver later this week, and each customer must choose a delivery day and time window. On Tuesday at 10:00, it splits the customers at random into two groups of 300 and sends each group one first message, a template, since none of them has written or called in the last 24 hours:

• Message A, a notice with a link: "Dear customer, your order [number] is ready for delivery. To schedule your delivery, please visit [link] or call us at [phone]."
• Message B, one question with two buttons: "Hi [name], this is [store] about your washing machine order. We can deliver it on Thursday between 10:00 and 13:00. Does that work for you?" Buttons: "Yes, Thursday" and "Another day". For other orders, message B names the customer's own product.

The reply rate is customers who replied divided by customers the message was sent to. Delivery booked means the customer has chosen a delivery day and time window. The results table counts everything up to 18:00 the same day, and each share is out of the 300 sent.

Where the numbers come from:

• Message A: 39 customers replied in the chat. 21 of them booked a delivery there; the other 18 first asked about something else, such as a delivery fee, a Saturday delivery or a different address, and had not booked by 18:00. 45 more booked through the link without replying. 21 + 45 = 66 booked.
• Message B: 171 customers tapped "Yes, Thursday" and 33 tapped "Another day": 171 + 33 = 204 replied. By 18:00, 24 of the 33 had chosen another day in the chat and 9 had not finished. 171 + 24 = 195 booked.

What the results table shows:

• Reads were almost the same: 261 of 300 (87%) for message A, 267 of 300 (89%) for message B. The reply rate was not: 39 of 300 (13%) against 204 of 300 (68%).
• Message A asked no question. It asked for work outside the chat: open a link or call. Most customers read it and did nothing.
• The outcome is what counts: delivery booked for 66 of 300 (22%) with message A, and 195 of 300 (65%) with message B.
• A reply is only the start. 18 of the 39 customers who replied to message A had not booked by 18:00. With message B, the 9 of 33 who tapped "Another day" and had not finished need a follow-up, from a person or a reminder.

Your own numbers will differ. This example changes the whole message to show the gap; in your own tests, change one thing at a time, such as the first line or the buttons.

What do messages that get a reply look like?

They say who is writing, name the customer's own thing, and end with one question or one thing to do in the chat. The table in this section shows five common WhatsApp messages to customers, before and after; each situation needs a decision or an action from the customer, so a notice is not enough.

The quote row asks which point is open, because "We're still looking at it", the easy reply to a check-in, tells you nothing.

What should happen after the customer replies?

Someone answers back within minutes, handles the customer's question before asking the next one, and sums up everything agreed at the end. That speed is an example target for your opening hours; pick your own and measure it.

The customer's reply opens a new customer service window, so for the next 24 hours you can write normal messages. If the window closes, the next message must be a template again.

A complaint, an angry customer or a request your rules do not cover goes to a person on the team with the whole thread, so the customer does not repeat themselves.

A thread with a customer who tapped "Another day" on message B:

• Store: "Hi [name], this is [store] about your washing machine order. We can deliver it on Thursday between 10:00 and 13:00. Does that work for you?" Buttons: "Yes, Thursday" and "Another day".
• Customer: taps "Another day".
• Store: "No problem. Which is better for you: Friday or Saturday, between 14:00 and 17:00?"
• Customer: "Saturday. Can the courier call me before coming?"
• Store: "Yes, the courier will call you about 30 minutes before arriving. Is the address still [address]?"
• Customer: "Yes, same address."
• Store: "Done: Saturday between 14:00 and 17:00 at [address], and the courier calls about 30 minutes before. Thank you, [name]."

Each store message asks one question at most, the store answers the courier question before asking about the address, and the last message sums up what was agreed.

Which messages need no reply at all?

Notices. A message that only informs, such as an order shipped, a receipt or a short note before an appointment, is complete when it is read.

• Do not add a question just to get a reply. If every message asks something, customers learn to ignore the questions.
• Send fewer messages, and ask only when you need a decision.
• Follow up a notice when the customer still needs to act, such as a delivery window nobody confirmed: send one question, or call. Message A is that case.
• Give customers an easy way to stop messages they do not want.

Our guide on when to send an appointment reminder uses both kinds: its main reminder asks for an answer, and its short note the day before ends with "No need to reply."

Where does DRING come in?

Co-chat is the shared inbox where your team and DRING AI agents answer WhatsApp and your other written channels together.

• Sending: Chat engine sends approved templates for WhatsApp and SMS by rule or by the agent, with variables filled from the record. Depending on the package, Co-chat's campaign manager sends an approved message to a segment, and every reply comes back to the same inbox.
• Answering back: the AI agent answers directly or drafts an answer for a teammate to review, set per channel and per conversation type. Reminders flag a conversation that has waited too long for an answer, and a handover passes the whole thread to a person.
• Inside the chat: a booking, an order status or a payment link can happen in the chat itself, and the outcome is written to your CRM through Connect.
• When a message is not enough: Notif sends an event message on WhatsApp, SMS or email the moment the event fires. If it is not read, or read but not answered where an answer was expected, an AI agent calls after the wait window you set.

62 languages are available and 10 are live today; your language is validated before launch.

Get a call in two minutes about the WhatsApp message your team sends most. DRING calls you in two minutes.