Fake and Prank COD Orders: Catching Them Before They Ship

Why COD collects bad orders that prepaid never sees

A prepaid checkout has a natural filter: the card. Anyone unwilling or unable to pay stops there, and the seller never spends a cent on them. Cash on delivery removes that filter entirely. Placing an order costs nothing, requires nothing verifiable, and can be done by anyone with a phone number, real or not.

The result is a category of order that prepaid sellers rarely think about: the address that does not exist, the phone that never answers, the same name ordering the same item four times in an hour, the order placed by a child on a parent's phone, and the deliberately malicious order aimed at a competitor. None of these are fraud in the payment sense. They are simply parcels that will never be paid for, and each one costs you outbound and return freight.

The signals that show up before dispatch

Most bad orders announce themselves, if anyone is looking. The signals are unglamorous and almost always visible in the conversation or the order record before a courier is ever booked.

• The phone number does not match the country format, is unreachable, or was typed with a digit missing.
• The address is a single word, a city name with no street, or a description that no courier could route.
• The same number or address placed several identical orders in a short window.
• The buyer never answered a single confirmation message, yet the order sits in the pipeline as if it were agreed.
• The order was placed at an unusual hour with no conversation at all around it, straight from a form.
• The buyer refuses to confirm the total, or reacts to the price as if they had never seen it.

Blocklists help, but they are the wrong primary tool

Every experienced COD seller keeps a list of numbers that burned them. It is worth maintaining, and it stops repeat offenders. But it is inherently backward-looking: it only knows the people who already cost you a delivery, and it does nothing about first-time bad orders, which are the majority.

The stronger filter is forward-looking and costs nothing per order: require a real, two-way confirmation before anything is dispatched. Not a tick box, not an automated "your order is confirmed" message, but an exchange where the buyer restates what they want, agrees to the total, and gives an address a courier can actually find.

Filtering without insulting real customers

The risk of any filter is that it treats good buyers like suspects. In a trust-driven market, that is expensive. The way through is to make verification feel like service rather than interrogation: confirming the delivery window is helpful, repeating the total is transparent, and asking for a landmark near the address sounds like competence, not suspicion.

Framing matters more than the questions themselves. "So the driver finds you quickly, which neighbourhood are you in?" collects exactly the same data as an accusatory check, and a genuine customer answers it happily. Someone who never intended to receive the parcel usually just stops replying, which is the outcome you wanted.

Where an Agent does this better than a person

Verification is repetitive work that has to happen on every order, at any hour, without ever getting lazy at the end of a shift. That is exactly what a human confirmation team does worst and an Agent does best. A Replyk Agent asks for the missing pieces one at a time, checks the phone number against the expected format for the country, and only creates a complete order record when the details actually hold together.

Just as usefully, it never lets a silent order sit. A buyer who stopped replying gets a scheduled follow-up rather than a parcel. Orders that clear the conversation are the ones that reach a courier, and the rest quietly never become shipping costs.

Frequently Asked Questions

Are fake COD orders actually fraud? Usually not in the legal sense, since no payment is taken. They are still a direct cost: you pay to ship out and to ship back on an order that earns nothing.

Should I ask for a deposit? In some markets a small deposit is normal and filters non-serious orders effectively. In others it kills conversion outright, so test it rather than assuming.

Can an AI Agent tell a real buyer from a fake one? It cannot know intent, but it can enforce verification on every order without exception, and that is what removes most bad orders before dispatch.