Checkout Is Not The Sale

In cash-on-delivery commerce, checkout is only an expression of intent. The merchant still has to confirm the customer, prepare inventory, assign a courier, pay delivery cost, and wait for cash collection.

That means a COD platform should not treat every new order as equally ready to ship.

The platform should ask a sharper question:

What do we know about this order before the merchant pays to deliver it?

Failed Deliveries Are Data

Most COD teams treat failed deliveries as operational noise:

• customer unreachable;
• bad address;
• refused at door;
• asked to reschedule;
• courier could not locate customer;
• repeated refusal from the same phone or address.

Those outcomes are not just historical reports. They are signals for the next decision.

If the same phone number repeatedly refuses delivery, that is a different risk than a first-time customer who asked for a callback. If one courier produces vague failure reasons, that is a different problem from buyer fraud.

The system needs a failure taxonomy before it needs an AI model.

Scoring Should Explain Itself

A risk score without explanation is not useful for a merchant. It creates another black box.

A better score says:

Risk: high
Reasons:
- 3 failed confirmation attempts
- prior refusal linked to same phone
- address missing neighborhood detail
Recommended action:
- call again before dispatch

This gives the merchant a decision surface, not just a number.

Architecture Implication

Risk scoring depends on workflow history. That is why the order lifecycle matters.

The platform needs to preserve:

• confirmation attempts;
• callback outcomes;
• courier assignments;
• failed delivery reasons;
• final delivery state;
• merchant-specific failure patterns.

Without that event history, scoring becomes guesswork.

The Product Principle

COD operations should close the loop:

order arrives
  -> confirmation workflow
  -> dispatch decision
  -> delivery outcome
  -> risk signal for future orders

The goal is not to block customers aggressively. The goal is to reduce preventable shipping loss while keeping legitimate buyers moving through the system.

That is why Wasilio is not just an order-management project. The deeper product direction is decision support for COD merchants.