What COD Fraud Is and How It Works in Indian E-Commerce
COD fraud is any behaviour where a buyer places a cash-on-delivery order without a genuine intention to accept and pay for the shipment. In the Indian context, it manifests in several distinct patterns that sellers must learn to distinguish.
The most common pattern is deliberate non-acceptance: the buyer is present but refuses delivery, often citing a fabricated defect or claiming they never placed the order. A second pattern is the ghost address order, where an incomplete or entirely fictitious address is provided, making delivery structurally impossible and triggering an automatic return. A third, more sophisticated variant is wardrobing or return abuse, where the buyer accepts the order, uses the product, and then initiates a return under a false pretext — though this is more prevalent on prepaid transactions.
What makes COD fraud particularly damaging is the asymmetric cost structure it creates. The seller pays for forward shipping, packaging, and the opportunity cost of reserved inventory regardless of outcome. When the parcel returns, a reverse logistics fee is layered on top. Multiply that across dozens of fraudulent orders per month and the cumulative drag on gross margin becomes material.
Fraud also clusters geographically and demographically: certain pin codes, first-time buyers with unverifiable contact details, and orders placed during high-discount sale events carry statistically higher risk. Recognising these patterns is the first step toward a systematic defence.
Key Risk Signals That Flag a COD Order as High-Risk
Effective COD fraud prevention begins at the order intake stage, not after dispatch. Sellers who build a checklist of pre-shipment risk signals can intercept the majority of fraudulent orders before a single courier movement occurs.
Phone number quality is one of the strongest individual signals. Numbers that are switched off, do not exist in telecom databases, or belong to virtual SIM services correlate strongly with fraudulent intent. A simple IVR confirmation call or SMS-with-OTP step at order placement eliminates a large proportion of ghost orders instantly.
Address completeness and verifiability is the second critical dimension. Orders missing a house or flat number, a recognisable locality name, or a serviceable pin code should trigger a hold, not immediate dispatch. Cross-referencing the delivery pin code against your logistics partner's serviceability map catches structurally undeliverable orders early.
Buyer history is the most powerful signal for repeat offenders. A buyer whose phone number or email is associated with previous RTO shipments on your platform — or on a shared industry blacklist — represents a known, quantifiable risk rather than a probabilistic one. Investing in a buyer risk score that aggregates order history, RTO rate, and payment mode preference pays dividends over time.
Finally, order profile anomalies — unusually high-value COD orders, multiple orders to the same address within a short window, or orders placed immediately after an account is created — are behavioural flags that warrant additional verification before acceptance.
Verification Workflows That Stop Fraud Before Dispatch
Knowing the risk signals is necessary but insufficient; sellers need operational workflows that act on those signals consistently at scale. Manual review does not scale beyond a certain order volume, so the goal is to build a tiered verification system where human attention is reserved for genuinely ambiguous cases.
The first tier is automated OTP or IVR confirmation. Any COD order above a defined value threshold, or from a first-time buyer, should trigger an automated call or SMS requiring the buyer to confirm the order. This single step eliminates orders placed accidentally or by minors, and deters low-effort fraudsters who rely on anonymity.
The second tier is address enrichment and pin-code validation. Integrate your order management system with a pin-code serviceability API so that incomplete or unserviceable addresses are flagged before the order enters the picking queue. Where addresses are ambiguous, a brief outbound call from your customer support team to confirm landmarks adds minimal cost but substantially reduces failed delivery attempts.
The third tier is selective COD acceptance based on risk score. Rather than accepting or rejecting orders in binary fashion, sellers can offer high-risk buyers a prepaid conversion incentive — a small discount or free shipping — that simultaneously reduces fraud exposure and improves cash flow. Buyers with genuine intent nearly always accept a reasonable prepaid offer; those with fraudulent intent typically abandon.
Documenting every verification interaction creates an audit trail that is useful both for internal learning and, where applicable, for filing complaints with platforms or authorities.
Building a Buyer Blacklist and Using Industry-Wide Tools
Individual blacklisting is the natural output of a mature verification workflow, but its value multiplies when sellers participate in shared fraud intelligence networks. The Indian e-commerce ecosystem has developed several mechanisms — at the logistics aggregator level, marketplace level, and through third-party data providers — that allow sellers to cross-reference buyers against a consortium of RTO and fraud histories.
A seller-maintained blacklist should capture the buyer's phone number, email address, and delivery address (including pin code), along with the nature of the fraud and its date. This structured record enables automated blocking at checkout rather than requiring manual review of every repeat order. Most modern order management systems and e-commerce platforms support custom blocklist rules that can trigger on phone number or email match.
Participating in logistics aggregator RTO databases is a complementary step. Shiprocket and similar platforms maintain aggregated delivery performance data across their seller networks. Sellers who share return data contribute to — and benefit from — a collective intelligence layer that individual blacklists cannot replicate.
It is important to apply blacklists with appropriate governance: a buyer incorrectly flagged loses access to legitimate purchases, creating customer service liability. Blacklist entries should be reviewed periodically and should require corroboration from more than one fraudulent event before permanent blocking is applied. A probationary state — where the buyer is required to prepay — is often a more proportionate and legally defensible response than outright blocking on a first incident.
Common Mistakes Sellers Make in COD Fraud Management
Even sellers who are aware of COD fraud risks often undermine their own defences through predictable operational mistakes. Understanding these failure modes is as important as knowing the best practices.
The most common mistake is acting only after the return arrives. Retroactive cancellation of fraudulent orders after the parcel has shipped recovers nothing — the forward logistics cost is already incurred and the inventory is in transit. Effective fraud prevention is entirely a pre-dispatch activity.
A second frequent error is applying verification universally but superficially. Sending an OTP to every buyer regardless of risk level increases friction for genuine customers, raising cart abandonment, while doing nothing to stop sophisticated fraudsters who simply intercept the OTP on a real number. Risk-tiered verification — lighter for established buyers, heavier for new or flagged accounts — balances security against conversion rate.
Sellers also routinely underestimate the cost of a single RTO. They calculate the reverse logistics fee but overlook the forward shipping cost already paid, the re-packaging labour, potential product damage, and the working-capital impact of funds not received. Building a true per-RTO cost model that captures all these components often reveals that even a low fraud rate is destroying a disproportionate share of net margin.
Finally, many sellers treat COD fraud as an external problem to be solved by their logistics partner or marketplace, rather than a first-party risk management responsibility. Platforms and couriers can assist, but the seller's own data — order patterns, customer history, address quality — is the most actionable intelligence available, and only the seller controls how it is used.
Building a Long-Term COD Risk Policy for Sustainable Growth
Tactical interventions reduce fraud in the short term, but sustainable protection requires a formalised COD risk policy that encodes rules, thresholds, and responsibilities into repeatable processes. Without this, fraud prevention reverts to ad-hoc decision-making whenever staff changes or order volumes spike.
A robust policy document should specify: which order attributes trigger automatic verification, what value threshold separates low-risk from high-risk COD orders, who has authority to override a hold or blacklist entry, how frequently blacklists are reviewed, and how fraud loss data feeds back into policy updates. This governance structure turns fraud prevention from an individual skill into an institutional capability.
The policy should also address the COD-to-prepaid conversion strategy explicitly. Defining which customer segments to target for prepaid nudges — new buyers, high-value orders, repeat RTO buyers — and what incentives are permissible ensures that the conversion effort is consistent and measurable rather than opportunistic.
As a seller's order volume grows, the economics of investing in machine-learning-based fraud scoring improve. At scale, automated models that ingest dozens of order attributes and return a real-time risk score outperform manual checklists. Many logistics platforms and fraud prevention SaaS providers offer APIs that integrate into existing checkout flows with relatively low engineering effort.
Finally, sellers should treat their RTO rate as a primary business metric, reviewed weekly alongside conversion rate and gross margin. A rising RTO rate is an early warning indicator of fraud exposure increasing; catching it early allows policy adjustments before losses compound into a structural problem.