Selling on WhatsApp without losing half your orders in the scroll

There is a shop we have all seen. Business is good. Orders come in on WhatsApp all day, on two or three phones. Payment arrives on MoMo. Somebody writes the important ones in a notebook. And every week, two or three orders simply evaporate — scrolled past during a busy hour, never written down, remembered only when the customer calls to ask where their delivery is.

The instinct is to blame whoever was on the phone. It is not their fault. You have asked a person to be a database.

Why WhatsApp is right and the setup is wrong

Every consultant's first suggestion is to move customers onto an app or a web shop. For most businesses here, that suggestion is wrong, and expensive, and it fails.

Your customers are on WhatsApp already. They will not install an application to buy a crate of drinks from a supplier they have used for six years. Asking them to is asking them to do you a favour, and some of them will simply buy elsewhere.

So the channel is not the problem. The problem is that nothing happens after the message. There is no order record, no stock movement, no payment match, and no way to answer "what did we sell on Tuesday" without asking three people.

The four things that have to exist

You do not need a big system. You need four small ones, and they can be built in the order below — each is useful on its own.

One price list. Every item, one price per customer type, one place, one person allowed to change it. Do this first, before anything technical. It costs nothing and it ends the most common argument in the business: two staff quoting two different prices to the same customer.

A twenty-second capture. The order still arrives on WhatsApp. But the person reading it taps it into a short form on their phone — customer, items, quantity — and it lands in a real record with the price filled in automatically from the list. Twenty seconds, at the moment the message is read, while it is still in front of them. The discipline is the thing; the technology is trivial.

Payments that attach themselves. MoMo payments arriving as screenshots that someone matches by hand is where an hour a day disappears. Connected properly, the payment attaches to the order, and what is left is a short list of genuine exceptions instead of a pile of matching.

A number you can look at. One screen: what was sold today, what is unpaid, what is running out. On a phone. The owner checks it in the morning and the whole business becomes visible for the first time.

What this is not

It is not an app for your customers. Nothing changes for them; they message the same number and get an answer from the same people.

It is not a full ERP. Businesses at this stage that buy an ERP end up with an expensive system in which one person enters data and nobody else looks. Buy the small thing that gets used.

It is not a reason to hire. If anything it removes the argument for hiring an extra person to keep up with the mess.

Where the AI part goes — later

Once those four exist, an agent becomes genuinely useful, and it takes a fortnight instead of six months. It can answer the price question without a person, confirm stock because there is now a stock number to check, take the order because there is somewhere to put it, and send the payment link because it knows what the order is worth.

We have written about that sequence at more length — the short version is that automating a business with no records is automating nothing. The records come first, and they are cheap.

What we would ask you first

• How many orders a week do you think you lose in the scroll? Guess; the guess is usually close.
• How many phones are taking orders, and can any of them see what the others took?
• Who knows the price of everything, and what happens the week they are away?
• How long does it take you to answer "what did we sell last Tuesday"?

If those questions are uncomfortable, the fix is smaller and cheaper than you are expecting. It is also the thing that has to happen before any of the clever software is worth buying.