Identifying High-Risk Customers Before You Ship: A Guide for Pakistani COD Stores

The problem with COD fraud in Pakistan

COD is the dominant payment method in Pakistani ecommerce — and it creates a specific fraud vector that doesn't exist with prepaid orders: the customer can simply refuse the parcel. No chargeback process, no bank involvement, no dispute mechanism. The courier brings it back, and you've paid for a round trip.

Some customers do this repeatedly. They place COD orders impulsively and refuse delivery consistently. Others give fake or incomplete addresses. Some are professional "serial RTOs" who place orders across multiple stores, never intending to receive them.

The challenge: each store sees only its own order history. A customer who has RTOed 50 orders across 10 different stores looks like a first-time customer to your store.

What makes a customer "high-risk"?

Risk indicators that Courierify's cross-merchant system tracks (on anonymised, privacy-preserving signals):

• High RTO rate across the network: The customer's phone number is associated with a high proportion of returned deliveries across multiple stores.
• Repeat non-answer pattern: Consistently not available during delivery attempts across different couriers and stores.
• Address-level signals: Delivery zone has unusually high RTO rates, or the specific address has a pattern of failed deliveries.
• Order pattern signals: Very high order frequency, multiple simultaneous orders, or other behavioural patterns associated with fraudulent activity.

Why cross-merchant data matters

A single store processing 300 orders per day generates meaningful data — but a network of stores processing tens of thousands of orders collectively generates powerful signal. Patterns that are invisible at the single-store level become clearly identifiable at network scale.

This is the core value of Courierify's cross-merchant risk database: you benefit from the collective delivery experience of the entire Courierify merchant network, even for customers who have never ordered from your store before.

How to use risk scores in practice

A risk score is a signal, not a verdict. The right response depends on the score level and your business context:

• Low risk: Proceed to booking normally.
• Medium risk: Send a WhatsApp confirmation before booking. A customer who confirms actively is much more likely to accept delivery, even if they have a historically mixed record.
• High risk: Hold for manual review. Your team looks at the order, the customer history, and the product — and decides whether to proceed, require prepayment, use a cheaper courier, or cancel.
• Very high risk: Auto-cancel or require prepayment only.

The thresholds and responses are configurable in Courierify — set them to match your risk tolerance and team capacity.

Integrating risk into your booking workflow

On the Courierify Pro plan, high-risk customer identification integrates directly with the Booking Intelligence Engine — risk score is available as a routing condition.

Example rules:

• High risk + order value > PKR 3,000 → Manual review queue
• Medium risk + no WhatsApp confirmation after 2 hours → Hold dispatch
• Any risk level + first-time customer + COD > PKR 8,000 → Send WhatsApp confirmation before booking

These rules run automatically — your team only sees the exceptions, not every order.

Measuring impact

Track these KPIs before and after enabling risk-based routing:

• Overall RTO rate — should decrease as risky orders are confirmed or cancelled before shipping
• RTO rate by risk segment — confirms the risk scoring is predictive for your order mix
• False positive rate — % of orders flagged as high-risk that were actually delivered successfully. Tune thresholds if too high.
• Revenue impact — some high-risk orders that get cancelled would have been delivered. Balance fraud prevention against lost genuine sales.

Courierify's analytics dashboard (Pro) tracks all of these metrics over time, allowing you to continuously optimise your risk configuration. Start your free trial to see the full analytics suite.