Fraud on Both Sides of the Transaction: Marketplaces Threats Don’t End at Checkout

Author: Daniel Basilio, Head of Non-Financial GTM at Sumsub

Every marketplace makes the same quiet promise to its users: whoever is on the other side of this transaction can be trusted – that promise is getting harder to keep.

Losses to online payment fraud are projected to exceed $107 billion by 2029, yet most of the tools built to contain them were designed for merchants who only ever had to doubt the buyer. Marketplaces must doubt both sides at once, and the people attacking them know it.

Protecting a platform therefore depends less on a strong checkout than on whether trust, once granted, keeps being tested for as long as a buyer or seller stays.

Fraud has become an industry

The scale alone is daunting. INTERPOL's 2026 Global Financial Fraud Threat Assessment puts global financial fraud losses at $442 billion in 2025, much of it generated by cross-border networks with every reason to keep investing, since the same report found AI-enhanced fraud to be 4.5 times more profitable than traditional methods.

Adding to this, Sumsub's Identity Fraud Report 2025-2026 points the same way. It ranks e-commerce among the ten industries most affected by fraud. Its more telling finding, however, is about how fraud has changed rather than how much of it there is – ‘sophisticated,’ multi-step attacks grew 180% last year as stronger controls pushed fraudsters beyond simple tactics. As such, marketplaces are no longer fending off amateurs testing a stolen card. Instead, they are up against operations that plan, adapt and continually come back.

On the buyer side, a valid payment can still be fraud

For these operations, the buyer account is usually the easiest way in. Card-not-present (CNP) payments offer no physical card to inspect, which helps explain why around 70% of UK card fraud losses occur there, and the disguise improves further once fraudsters move from stolen cards to stolen accounts. Account takeover cases rose 155% in 2023 and 250% in 2024, because a hijacked account comes with a history the platform already trusts.

Nor does the damage end with the first order. Even Mastercard expects global chargeback losses to climb from $33.79 billion in 2025 to $41.69 billion by 2028. Against this backdrop, disputes at commerce marketplaces continue to rise. Armingly, this still understates the true cost, since refund abuse, promo abuse and triangulation fraud can erode margins without a single chargeback.

Worryingly, on a marketplace, a cleared payment says far less about the buyer than it seems.

On the seller side, the merchant can be the threat

Traditional retailers never had to ask whether they themselves were the fraud, but marketplaces must ask it of every seller they approve. A fraudulent merchant can list goods that will never ship, or buy from its own storefront with illicit funds so that dirty money returns as apparent revenue. The Financial Action Task Force has flagged online marketplaces as vulnerable to exactly these schemes, as high volumes and cross-border flows give fictitious transactions ample cover.

Transaction laundering, which routes payments for prohibited goods through a legitimate-looking merchant account, thrives wherever onboarding and monitoring are weak. And when the merchant is the one committing fraud, the checkout is the last place it will show, so sellers deserve at least the same scrutiny as buyers.

The threat that arrives after approval

It would be reassuring to think that thorough verification at sign-up solves both problems. Checking ID and biometrics for buyers, and running business verification checks on sellers and the people behind them, does keep many bad actors out. The difficulty is that fraud has learned to wait. 

Sumsub research shows that most fraud occurs after onboarding, when a buyer account is quietly taken over or a seller that passed every check begins to behave differently. Against a patient opponent, a single check at the door records only who someone was on the day they arrived.

Yet the obvious response, more checks for everyone, would punish the honest majority and drive them toward competitors. The better answer is to let risk decide when friction appears. Real-time transaction monitoring, device fingerprinting and behavioral analytics measure each action against an account's usual pattern.

A regular customer reordering a familiar item therefore passes untouched, while a high-value order from a new device or a sudden change of payout details prompts a closer look. AI-powered detection then surfaces the coordinated networks that rules alone miss, keeping protection tight where it matters and almost invisible where it does not.

Trust that keeps being tested

In reality, marketplaces have always asked their users to trust strangers, which is precisely what makes them so attractive to fraud that is now organized, automated and willing to wait. A defense concentrated at the checkout, or even at onboarding, will catch some of it, but it will miss the account hijacked weeks later and the seller laundering money through its own orders.

The platforms that keep their promise will be those that verify buyers and sellers when they arrive and keep testing that trust for as long as they stay.

For a closer look at each fraud type and the controls that stop it, read Sumsub's complete guide to e-commerce fraud prevention, or download our Identity Fraud Report 2025-2026.