Taking orders on WhatsApp without losing half of them

In a lot of markets, the honest answer to “how do your delivery orders arrive?” is: as chat messages, on a phone behind the till, handled by whoever is nearest.

This is not a failure of discipline. It works because it is where customers already are — no app to install, no account to create, no forgotten password. That is a genuine advantage and it is worth keeping.

The problem is that a chat app is an excellent conversation tool and a terrible order system, and most restaurants are using it as both.

What actually goes wrong

The orders live in a thread

A chat conversation has no queue, no status and no shared view. The kitchen cannot see it. The driver cannot see it. A second person cannot pick it up mid-way without scrolling back through a conversation that includes three revisions and a voice note.

When it is busy — the only time it matters — the thread becomes the bottleneck.

One phone, one person

Orders arrive on a single handset held by a single person. When that person is serving a table, on a break, or off that day, orders sit unread. Customers who are ignored for eleven minutes order elsewhere and do not tell you why.

The number belongs to an employee

This is the one that costs the most and gets noticed the latest.

If your ordering number is a staff member’s personal phone, then every regular customer, every conversation history and every repeat order is on a device you do not own. When that person leaves — and eventually they leave — your customer base walks out with them, along with the ability to message it.

Get a number that belongs to the business. Do this before anything else in this article.

Nothing is recorded

Orders taken in chat and never entered into the system do not appear in sales, do not deduct stock, do not appear in the channel comparison and do not count toward the customer’s history.

The consequence is not just accounting. It means every conclusion you draw about your business is drawn from partial data: your best-selling item is your best-selling item among orders that happened to be recorded. If a quarter of your delivery volume never enters the system, you do not know your delivery margin, your item mix or your peak hours.

The address is a sentence

“The white building next to the mosque, third floor, ask for Ahmed.”

Every delivery person in the world has read that message. It is not a structured address: there is no zone, so there is no delivery fee calculation, no distance, no way to assign the nearest driver, and no way to find it again next time.

The menu you sent is out of date

A photo of the menu is a snapshot. The moment a price changes or an item is unavailable, every customer holding that image has the wrong information, and your staff spend the evening correcting it one conversation at a time.

The shape that works: chat is the entrance, not the system

The fix is not to abandon WhatsApp. Customers are there, and telling them to install an app instead is how you lose orders to whoever did not.

The fix is to let chat do what it is good at — being reachable, answering questions, handling exceptions — and let the order itself land somewhere structured.

In practice, one of two flows:

1. Send the ordering link. The customer messages, and your first reply includes a link to your own ordering page. They choose items at current prices, enter an address that becomes real data, see the delivery fee, and confirm. The conversation continues in chat; the order exists in your system.
2. Transcribe immediately. If a customer prefers to order conversationally — and many will — whoever takes the message enters it into the POS while they are replying, not at the end of the shift. It becomes a real order with a kitchen ticket, a driver, a record.

Either way the principle is the same: no food gets cooked from a chat thread. The message is where the conversation lives; the order lives in the system.

That is also why your QR menu and your own ordering page earn their keep here — not as a replacement for chat, but as the thing chat points at. One link, always current prices, structured address, and an order that arrives where the kitchen already looks.

Setting up the business side properly

Most restaurants use maybe a quarter of what the business version of the app offers. The pieces that actually save time:

• A business profile. Address, hours, and a link. This is what people see before they type.
• A greeting message. Automatic, on first contact. Put the ordering link in it. Half your customers will never need to ask a question.
• An away message. Outside hours, tell people when you open. An unanswered message at 2am costs you the next lunch too.
• Quick replies. You answer the same six questions every day: opening hours, delivery zones, delivery fee, payment methods, where you are, is the item available. Save them once.
• The catalogue. Keep it current or do not use it. An out-of-date catalogue is worse than no catalogue.
• Labels. As a stopgap, label conversations by state — new, confirmed, out for delivery, paid. This is a weak substitute for an order queue, but it is far better than scrolling.
• A click-to-chat link. Put it on your Google Maps profile, your social bios, your receipts and your packaging. It removes the step where someone has to save your number first.

Two details that pay for themselves

Ask for the location pin, not the address. One sentence — “please send your location” — converts an unreliable description into a coordinate. The driver navigates instead of phoning. Save it against the customer and the second order is easier than the first.

Confirm in a fixed format. Every confirmation should say the same four things in the same order: items, total, delivery fee, expected time. A confirmation that varies by whoever typed it produces disputes at the door.

About marketing messages

The pull is obvious: you have hundreds of customer numbers and a way to reach them instantly.

Be careful, and be specific:

• Broadcasting to people who did not ask for it gets your number blocked and reported, and blocked numbers are not easily recovered. Losing your ordering number to a spam block is a bad week.
• Business-initiated messages have rules. On the business platform, messaging a customer outside a recent conversation window generally requires pre-approved templates and consent. Check the current rules before building anything on top of this — they change.
• Ask permission and make it easy to stop. A customer who chose to hear from you is worth more than fifty who did not.
• A useful message beats a promotional one. “Your order is on the way” and “we are open late tonight” get read. Weekly discount blasts get muted.

Measuring a channel you previously could not see

Once orders from chat exist in your system, you can finally answer questions you have been guessing at:

• What share of orders arrives through chat? Usually more than owners think.
• What is the average check by channel? Chat orders are often larger than app orders and smaller than dine-in. Knowing which is true for you changes where you push.
• What does each channel actually earn after costs? Chat has no commission but has labour cost — somebody is typing. Compare properly rather than assuming.
• Who orders again? A customer whose number and history you own can be brought back. A customer who exists only in an aggregator’s database cannot.

That last one is the real prize. The value of taking orders in chat is not saving commission on one order — it is that the customer relationship is yours afterwards.

The one-week plan

1. Move to a business number the restaurant owns. Today.
2. Set up the greeting message with your ordering link, and an away message.
3. Write six quick replies for the questions you answer daily.
4. Put your click-to-chat link on your Maps profile and your packaging.
5. Make the rule explicit: every chat order is entered into the system before the kitchen starts it.
6. Start asking for the location pin on every delivery.
7. At the end of the week, look at what share of orders came through chat — and start treating it like the channel it evidently is.