How to Stop Fake Cash on Delivery Orders

Cash on delivery remains the dominant payment method for online shopping in Nepal. Digital wallets and bank transfers are growing, but a large share of customers still prefer to pay only when the product arrives at their door. This preference is reasonable because many buyers have had bad experiences with receiving wrong items or no items at all, and COD gives them a sense of security. But it creates a serious vulnerability for sellers.

When a customer places a COD order, they make no financial commitment. The seller bears all the risk. They pick the item from inventory, pack it, label it, hand it to a courier, and pay the shipping fee. They do all of this before receiving a single rupee. If the customer refuses the delivery, gives a wrong address, or simply does not answer their phone when the courier calls, the seller absorbs every cost associated with that order.

Those costs add up faster than most sellers realise. There is the packaging material: the box, the bubble wrap, the tape, the printed invoice. There is the courier fee for the outbound shipment. If the package is returned, there is often a return shipping fee as well. There is the staff time spent picking, packing, and processing the order. There is the inventory that was held up in transit for days and could not be sold to anyone else during that time. And there is the morale cost, because few things are more demoralising for a small team than watching a stack of returned packages pile up by the door every week.

The scenarios are painfully familiar to any e-commerce seller:

• The delivery rider calls the customer's number and nobody picks up. They try again the next day. Still no answer. The package comes back.
• The address on the order does not exist. The street name is wrong, or the house number is made up. The courier cannot find the location.
• The customer answers the phone and says they never placed an order. Someone else used their number, or they forgot they ordered something while scrolling through Facebook at midnight.
• A competitor places dozens of fake orders to tie up your inventory and overwhelm your operations during a sale period.

For a seller processing 30 to 100 orders per day, even a 15% fake order rate means several wasted shipments every single day. Over a month, the financial loss can easily reach tens of thousands of rupees, which is money that comes directly out of profit margins that are already thin in Nepal's competitive e-commerce market.

Before you can fix the problem, it helps to understand why it happens. Not every fake order is malicious. In fact, most of them are not. Understanding the reasons behind fake COD orders helps you design solutions that target the actual causes rather than punishing all customers for the behaviour of a few.

No commitment required. The fundamental issue with COD is that placing an order costs the customer nothing. There is no payment to process, no wallet to open, no OTP to enter. They tap "Order Now" and it is done. This zero-friction checkout is great for conversion rates but terrible for order quality. A customer who had to pull out their phone, open eSewa, enter an amount, and confirm with a PIN has demonstrated real purchase intent. A customer who tapped one button while lying in bed at midnight has not.

Impulse purchases. Social media advertising on Facebook and Instagram is designed to trigger impulse decisions. A customer sees an attractive product in their feed, taps through to the store, and places an order within two minutes. By the time the delivery arrives three days later, the impulse has faded and they no longer want the item, or they forgot they ordered it entirely.

Shopping around. Some customers order the same product from multiple sellers and plan to accept whichever delivery arrives first. The other orders get refused at the door. The customer gets their item quickly; the losing sellers eat the shipping costs.

Ordering to try with no intention to pay. A subset of customers treat COD as a "try before you buy" service. They want to see the product in person before deciding. If it does not match their expectations, even if the product matches the listing exactly, they refuse delivery.

Prank or competitor sabotage. In some cases, fake orders are placed deliberately to harm a business. This could be a disgruntled customer, a bored teenager, or a competitor trying to drain your resources during a promotional period.

Children ordering without parents knowing. With smartphones in the hands of children and teenagers, it is not uncommon for a young person to order something they saw online without understanding the implications. When the delivery arrives, the parents refuse it.

Genuine forgetfulness. This is more common than you might think. A customer places an order, forgets about it after a few days, and then is confused when a courier shows up. They say they never ordered it, not out of dishonesty but out of genuine memory lapse.

Recognising these different motivations matters because the solutions are different. A confirmation call solves forgetfulness. A small deposit solves lack of commitment. A blacklist solves repeat offenders. No single approach handles all of them.

The most direct way to reduce fake COD orders is to verify each order before it leaves your warehouse. This means adding a step between "order received" and "order shipped" where someone or something confirms that the customer is real, the address is valid, and the purchase intent is genuine.

Here are the manual verification methods that sellers commonly use:

Call the customer to confirm. This is the most straightforward approach. A staff member calls the customer, confirms the items ordered, the delivery address, and the total amount. If the customer confirms, the order proceeds. If they do not answer or the number is invalid, the order is held. Calling works, and it catches the majority of fake orders, but it is time-consuming. For a seller processing 50 or more orders per day, calling each customer can take one or two full-time staff members doing nothing else.

Send a WhatsApp confirmation message. Instead of calling, send the customer a WhatsApp message with the order details and ask them to reply with a confirmation. This is faster than calling and less intrusive. Many customers respond quickly because they are already active on WhatsApp. However, some customers ignore messages from unknown numbers, and you cannot be sure a message was read if the blue ticks are disabled.

Check if the phone number is valid. Before calling or messaging, check whether the phone number has an active WhatsApp profile. A number without WhatsApp in Nepal is not necessarily fake, but the combination of no WhatsApp profile, no profile picture, and an unresponsive number is a strong signal worth flagging.

Verify the address on Google Maps. Copy the delivery address into Google Maps and see if it resolves to a real location. This will not catch every fake address, because some Nepali addresses are too informal for Google Maps to resolve, but it catches obvious fakes like nonexistent street names or addresses in the wrong city.

Check order history. If you keep records of past orders (and you should), look up the customer's phone number and address. Have they ordered before? Were those orders delivered successfully? A customer with a history of refused deliveries should receive extra scrutiny. A customer with three successful past orders probably does not need a confirmation call.

Flag orders with unusual patterns. Certain characteristics correlate with higher fake order rates: orders placed between midnight and 5 AM, orders with very high values from first-time customers, orders to areas where you have had many returns before. These are not proof of a fake order, but they warrant a closer look.

The problem with all these manual methods is obvious: they are slow, repetitive, and staff generally dislike doing them. Calling 50 customers per day is tedious work. Whoever is assigned to the task will rush through it, skip some calls, or eventually stop doing it altogether. This is where automation becomes valuable.

Automated AI confirmation calls take the most effective manual verification method, which is calling the customer, and remove the human bottleneck. Here is how the process typically works:

1. A new COD order is placed on your website or through a social media channel.
2. An automated system places a phone call to the customer within minutes of the order being received.
3. The AI voice confirms the order details with the customer: the items, the delivery address, the total amount, and the payment method.
4. If the customer confirms, the order is marked as verified and moves to fulfilment.
5. If the customer does not answer, the system tries again after a set interval, typically one to two hours later.
6. If there is no answer after two attempts, the order is flagged for manual review rather than being shipped automatically.

The benefits of this approach are significant. Consistency is the biggest one, because the AI calls every single order, every time, without getting tired or skipping anyone. Speed matters too, because the call happens within minutes, not hours, so the customer still remembers their order and can confirm quickly. And because the calls are automated, they work outside of business hours. An order placed at 10 PM can be confirmed by 10:15 PM, ready to ship first thing in the morning.

Socify App is one tool that can make these AI confirmation calls automatically after a COD order is placed. It connects to your ordering system, triggers a call to the customer's phone number, and updates the order status based on the customer's response. The setup does not require technical expertise, and the AI handles both Nepali and English conversations.

To be honest about the limitations: AI confirmation calls work well for straightforward order verifications, such as confirming items, address, and payment method. Some customers, however, find automated calls impersonal, and a small percentage hang up when they realise they are talking to an AI. Older customers in particular may be confused by the experience. For high-value or complex orders, a human follow-up call after the AI confirmation can add an extra layer of assurance.

The key point is that AI calls do not need to replace human interaction entirely. They handle the routine confirmations, which make up the 80% of orders that are straightforward, and free up your staff to spend their time on the orders that actually need human attention.

Requiring a small deposit before shipping is one of the most effective ways to filter out fake orders. The logic is simple: a customer who is willing to pay even NPR 100 to 500 upfront has demonstrated genuine purchase intent. The deposit does not need to cover the full product cost. It just needs to be enough to require a deliberate action from the customer.

Here is how sellers typically implement this:

1. The customer places a COD order as usual.
2. The seller sends a message (via WhatsApp or SMS) with a payment link or QR code for a small deposit, usually NPR 100 to 500, depending on the order value.
3. The customer pays the deposit through eSewa, Khalti, or bank transfer.
4. Once the deposit is received, the order moves to fulfilment. The remaining balance is collected on delivery.
5. If the deposit is not paid within a set window, typically 12 to 24 hours, the order is cancelled automatically.

The results from sellers who have implemented this are striking. Some report that their fake order rate drops from 20-25% to under 5% overnight. The reason is clear: customers who were never serious about the purchase are not going to open their digital wallet for even a small deposit.

There are trade-offs to consider, though. Some genuine customers will drop off at the deposit step. They may not have a digital wallet set up, they may find the extra step annoying, or they may interpret the deposit request as a sign that the seller is not trustworthy. Sellers who introduce deposits typically see their overall order volume dip by 10 to 20% initially, but their fulfilled order volume stays the same or improves because the orders they lost were the fake ones.

Finding the right balance takes experimentation. Some sellers apply deposits only to orders above a certain value, such as NPR 3,000 or more. Others apply them only to first-time customers. Some make the deposit optional but offer a small discount for customers who prepay. There is no single right approach; the right one depends on your product category, your customer base, and how severe your fake order problem is.

A practical tip: keep the deposit amount low enough that it feels trivial to the customer. NPR 100 is not much, but it is enough to filter out someone with zero intention to pay. Making the deposit too high, such as NPR 1,000 or more, starts to feel like a prepayment and may drive away legitimate COD customers.

As your business processes more orders over time, you accumulate valuable data about your customers. Using that data systematically can transform your approach to fake order prevention from reactive to proactive.

A customer trust score is a simple internal rating that reflects how reliable a customer has been in the past. It does not need to be complicated. At its most basic, it tracks three things:

• Completed deliveries: How many orders has this customer received and paid for successfully? Each completed delivery increases their trust score.
• Returns and cancellations: How many orders has this customer refused, returned, or cancelled? Each one decreases their trust score.
• Verification history: Has this customer confirmed orders via phone or WhatsApp in the past? Customers who consistently confirm are more trustworthy.

Based on the trust score, you can apply different levels of verification:

• New customers (no history): Full verification. This means a confirmation call plus a deposit for high-value orders. This is where the highest percentage of fake orders come from.
• Returning customers with good history: Minimal verification, perhaps just a WhatsApp confirmation or no verification at all. These customers have proven they are reliable, and making them jump through hoops every time risks annoying them.
• Customers with mixed history: Standard verification, with a confirmation call required but no deposit needed.
• Blacklisted customers: Orders from these phone numbers or addresses are held for manual review or declined outright. A customer who has refused three or more deliveries without explanation should be on this list.

Building this system does not require expensive software. A simple spreadsheet tracking phone numbers, order counts, and delivery outcomes can get you started. As your volume grows, you will want to move this into your order management system so that the trust score is checked automatically when a new order comes in.

One practice that some seller communities in Nepal have adopted is sharing blacklists among trusted peers. If a group of sellers agrees to pool their data on known fake orderers, everyone benefits. A customer who has refused deliveries from five different sellers is almost certainly a bad actor, and every seller in the network can flag orders from that number. This approach requires trust and clear data-sharing agreements between the participating sellers, but it can be highly effective.

A word of caution: be fair with your blacklist. A customer who refused one order once may have had a legitimate reason, such as a family emergency, a delivery that arrived damaged, or a genuine mistake. Reserve the blacklist for clear patterns of abuse, not one-time incidents.

The financial impact of fake orders is often larger than sellers realise because the costs are spread across several line items that do not always get tracked together. Let us work through a concrete example.

Suppose you are an e-commerce seller processing 50 COD orders per day. Your fake order rate is 20%, meaning 10 orders per day are undeliverable. Here is what those 10 fake orders cost you:

• Outbound courier fee: NPR 150 to 300 per shipment, depending on destination. Call it NPR 200 on average. That is NPR 2,000 per day.
• Return shipping fee: Not every package comes back because some are abandoned, but for those that do, another NPR 100 to 200. Estimate NPR 100 average across all fake orders. That is NPR 1,000 per day.
• Packaging materials: Boxes, wrapping, labels. NPR 30 to 50 per order. Call it NPR 40. That is NPR 400 per day.
• Staff time: Each order takes 5 to 10 minutes to pick, pack, and process. At 10 orders, that is roughly 1 to 1.5 hours of staff time wasted daily.
• Inventory holding cost: Products tied up in transit for 3 to 7 days cannot be sold to other customers.

Just the direct costs of courier fees and packaging come to approximately NPR 3,400 per day. Over a 30-day month, that is NPR 1,02,000 lost to fake orders. The indirect costs of wasted staff time and locked-up inventory add substantially to that figure.

Now suppose you implement a combination of AI confirmation calls and partial deposits, and your fake order rate drops from 20% to 5%. Instead of 10 fake orders per day, you have 2 or 3. Your monthly direct cost from fake orders drops from approximately NPR 1,02,000 to roughly NPR 25,000, which is a saving of about NPR 77,000 per month. For a seller processing more orders or dealing with higher-value products, the savings scale proportionally.

But the financial savings are only part of the story. When your team is not spending their mornings processing returns and dealing with courier complaints about undeliverable packages, they can focus on activities that actually grow the business: improving product listings, responding to customer enquiries, handling marketing, and packing orders that will actually be received and paid for.

Your relationship with courier partners improves too. Couriers track delivery success rates, and sellers with high return rates may face higher fees or slower service. Reducing your fake orders makes you a better partner for courier companies, which can translate to better rates and priority handling over time.

You do not need to implement everything at once. Here is a four-week plan that starts with the simplest changes and builds up to a comprehensive verification system.

Week 1: WhatsApp confirmation for all COD orders

Start with the lowest-effort, highest-impact change. After every COD order is placed, send the customer a WhatsApp message with the order summary, including items, delivery address, and total amount, and ask them to reply "Confirm" to proceed. Orders that do not receive a confirmation within 12 hours are held and followed up with a phone call. This single step will catch many fake orders because customers who never intended to pay will not bother responding.

Week 2: Add AI confirmation calls for orders above a threshold

For orders above a value where a failed delivery would hurt, such as NPR 2,000 or above, add an automated confirmation call in addition to the WhatsApp message. This can be done through Socify App or a similar tool that supports automated calling. The AI call serves as a second layer of verification. If the customer confirmed via WhatsApp and also answers the call, you can be confident the order is real. If neither gets a response, do not ship.

Week 3: Introduce partial prepayment option

For first-time customers and orders above NPR 3,000, add a small deposit step. Send a payment link for NPR 200 to 300 via eSewa or Khalti after the order is confirmed. Frame it positively: "To reserve your order and ensure priority dispatch, please confirm with a small deposit." Make it clear the deposit is deducted from the total, not an additional charge. Returning customers with good purchase history can skip this step.

Week 4: Build your customer trust database

Set up a simple tracking system. Even a Google Sheet will work to start. It should record each customer's phone number, number of orders, number of successful deliveries, and number of refused deliveries. Use this data to categorise customers into trust tiers. High-trust customers get expedited processing. New customers get full verification. Known bad actors get flagged or blocked.

Measure and adjust

Track your numbers week by week. The metrics that matter are: total COD orders placed, orders confirmed via WhatsApp or call, orders where deposit was paid, orders shipped, orders successfully delivered, and orders returned or refused. Calculate your fake order rate each week and compare it to your baseline. Adjust your thresholds and processes based on what the data tells you.

Every business is different, and the right combination of verification, deposits, and trust scoring will depend on your products, your customers, and your volume. The important thing is to start. Even implementing just the WhatsApp confirmation step in week one will likely reduce your fake order rate noticeably, and you can build from there.