The Rising Tide of E-commerce Fraud
We are deeply aligned with our aim to democratize our platform—with new entrepreneurs joining and logistics partners working across pin codes in India. This inclusivity is our biggest strength—and also the reason why we need to be extra vigilant. Every month, fraud entities were putting a dent in acquisition fees, cataloging costs, and customer refunds. Worse, genuine buyers and sellers—our core community—faced slow deliveries, empty cartons, and disappointing experiences. A small fraction of malicious accounts erodes trust—amplified by fraud rings. These coordinated networks of sellers, buyers, or delivery agents undertake damages which leads to both monetary and reputational losses for meesho. Worse, once an entity blocked, they re-enter the problem, leading to a huge problem of fraud-rentry,
These were the two main problems:
- Fraud Entities Re-entering the platform
- Non-Incremental entities placing transactions on one another and abusing the platform
By understanding both the monetary drain and the damage to our community’s trust, we knew a proactive, top-of-funnel solution was critical to safeguard Meesho’s inclusive marketplace. Clearly, a reactive approach for fraud detection was not enough. We needed a structural solution that could identify the true person (or network) behind every sign-up before any damage occurred.
Enter the Digital DNA
The Deduplication Model achieves this by building a unified “digital DNA” for each new account. Rather than looking at every new sign up in isolation, we capture multiple layers of intelligence using which we identify the person behind the scene.
For every entity on meesho, the environment through which they operate on meesho’s platform can be divided into 4 key components:
- Their specific metadata which they provide us ( eg: Phone number )
- The device and network signals
- Their behaviour with other entities on meesho
Building the Digital DNA Engine
At Meesho, trust is at the heart of every interaction. To safeguard the integrity of our platform, our Trust & Safety team continuously builds intelligent systems that can detect unusual patterns and uncover hidden connections. We use data across various platform touchpoints to understand how different entities relate to one another. By identifying subtle similarities and connections, we were able to group these entities into meaningful clusters. These clusters help us flag potential anomalies and surface behaviors that may not be obvious at first glance. What makes this approach powerful is its ability to go beyond isolated signals. Instead, it focuses on relationships — looking at the broader network of interactions to detect patterns that traditional methods might miss. By carefully filtering and refining these connections, we’re able to identify communities of activity that warrant closer inspection.
This capability plays a key role in strengthening platform safety and ensuring a trusted experience for our entire ecosystem.
This helps reveal several associations and links between different users, sellers and Delivery boys and other stakeholders of meesho. Over time, it started catching several linked identities and transactions proactively—before damage was done.
Impact and Next Steps :
The Digital DNA model by incorporating successfully identified fraud rings pro-actively which helped safeguard our platform from malicious actions and led to huge NMV uptick on Meesho in 1Q 2025.
Detecting and preventing fraud at scale not only protects Meesho’s ecosystem but also brings tangible benefits across the board.
- Smarter Onboarding, Lower Costs: With fewer fraudulent accounts slipping through, resources spent on acquiring and incentivizing fake users — whether buyers or sellers — have significantly reduced.
- Fewer Losses, Leaner Operations: By catching suspicious activity early, we’ve seen a meaningful drop in unnecessary reimbursements, fake refund claims, and the manual effort spent resolving them.
- Stronger Trust, Better Retention: Most importantly, proactive detection of organized fraud helps protect genuine users/sellers from bad experiences. This means more buyers and sellers continue to transact confidently on Meesho, deepening long-term engagement and trust.
Next, we are working on enhancing the digital DNA as it becomes a core-part of Meesho’s fraud detection infrastructure, by enhancing the model’s intelligence via more proactive signals. The model with multiple use-cases across fraud detection and identifying non-incremental entities has paved the way for proactive detection
