Manav.id
Vertical · 4 min read

Marketplace seller payout redirection: the takeover that looks like an update

Marketplace seller payout redirection: the takeover that looks like an update

Seller account takeover persists across every marketplace for one structural reason: a single settings field redirects an entire revenue stream, and the notification announcing the change goes to an inbox the attacker already controls.

How does a marketplace seller takeover actually pay out?

Through a settings change. The attacker alters the payout destination on a compromised seller account and sells nothing; trading continues as normal and the real seller discovers the diversion at the next payout cycle, by which time several settlement periods may have gone.

Key takeaways
  • Payout details are a settings field. Platforms authenticate sessions, not payment instructions, and the exposure window is the payout cycle length.
  • Sellers are frequently small businesses for whom a missed payout cycle is an existential cash-flow event, not an inconvenience.
  • International seller populations make enrolment coverage the binding design constraint.

The exposure window is the payout cycle

Seller account compromisedphishing or reusePayout destination changeda settings fieldTrading continues normallyno anomaly to detectDiscovered at the payout cycledays or weeks later
Nothing about the account's behaviour changes. Only the destination does.

The mechanics are simple and the timing is what makes them costly.

day 0   attacker compromises seller account
day 0   payout destination changed; notification sent to compromised inbox
day 1-13 sales continue normally; seller sees orders, ships goods
day 14  payout issued to attacker's account
day 15+ seller notices missing payout, opens support case
day 20+ platform investigates; funds long gone

The seller shipped fourteen days of goods and received nothing. The platform issued a payout correctly, to the details on file, following a change made from an authenticated session.

Shorter payout cycles reduce the exposure proportionally, which is why daily-payout platforms see smaller per-incident losses. Weekly and biweekly cycles are where this hurts.

Why notification does not help

Every platform sends a notification when payout details change. It is the standard compensating control and it fails for a specific reason.

The account takeover that enables the payout change almost always begins with email compromise, because email is how the platform account is recovered. So the notification about the change routes to the attacker, who deletes it.

Platforms that send to a secondary contact, or to the old destination's associated address, do materially better. Most do not have a secondary contact because collecting one adds signup friction.

Who the sellers actually are

This is where the design constraint comes from. Marketplace seller populations are overwhelmingly small businesses and sole traders, distributed internationally, operating on thin margins.

Population characteristicDesign consequence
Small business cash flowA missed payout cycle can end the business, not just annoy it
International distributionDevice access, connectivity and platform availability vary widely
Shared or family devicesCredential enrolment must handle a device used by several people
High seller turnoverEnrolment must be part of onboarding, not a campaign
Language diversityRendered instructions must be localised, including the account details

A control designed for a US-based professional seller will exclude a meaningful share of the population, and the excluded share is not evenly distributed.

The payout field control

Treat the destination change as a payment instruction rather than a profile edit.

{
  "type": "manav-stmt/1",
  "action": "payout_destination_change",
  "render": [
    "Seller: [account name]",
    "From: [existing destination, full detail]",
    "To: [new destination, full detail, unmasked]",
    "Effective: [next payout cycle]"
  ],
  "seller": "[credential assertion]"
}

Unmasked detail matters here as much as anywhere else in this series. A seller confirming a destination shown as four digits has confirmed nothing about the rest.

Coverage before enforcement

The sequencing that works:

  1. Enrol at onboarding, where every new seller already completes verification steps.
  2. Offer enrolment to existing sellers at natural touchpoints — payout confirmations, dispute resolutions, app updates.
  3. Measure coverage by seller revenue decile, not by seller count. Coverage of the top decile protects most of the value.
  4. Enforce for enrolled sellers. Unenrolled sellers keep the current path plus a hold and dual notification.
  5. Never block a payout because a seller cannot enrol. That converts a security control into a cash-flow harm.

What the platform saves

Most platforms make sellers whole for diverted payouts, because the alternative is a small business publicising that the platform let its revenue be stolen. That reimbursement is a direct cost.

Add support handling, investigation, and the seller churn that follows even when reimbursed. The total is usually well above what the control costs, and the calculation is entirely internal — no external data required.

Why behavioural detection does not fire

What the platform observes
SignalValue
Listing activityUnchanged
Order volumeUnchanged
Customer messagesUnchanged
Login locationSometimes anomalous, often proxied
Payout destinationChanged — the only signal, and it is not treated as one

Objections and honest limits

“We email the seller when payout details change.” To an inbox the attacker frequently controls, since the account compromise usually started there. The notification needs a channel the session cannot reach.

“A hold would upset legitimate sellers.” A short hold on the first payout after a change, not on every payout. Sellers change bank details rarely; the cost falls on an event that happens once every few years.

Protecting the payout field

  1. Treat it as a payment instruction. Not a profile setting.
  2. Require a bound assertion to change it. Rendering old and new destination.
  3. Notify out of band, stating the new destination. Not merely that something changed.
  4. Hold the first payout after a change. The only realistic recovery window.

Terms used here

Seller account takeover
Gaining control of an established seller account, whose value is its history rather than its balance.
Payout destination
The bank account settlement funds are paid to — the single field that converts a takeover into cash.
Settlement cycle
The period between payouts, which is also the detection delay for a diverted destination.

Frequently asked questions

Does MFA on login solve this? It helps considerably and is worth deploying. It does not address a session compromised after login, and it applies the same strength to every settings field rather than to the one that moves money.

What about sellers without smartphones? They keep the current path with a hold and dual notification. Blocking their payouts would cause the harm the control exists to prevent.

Should we shorten payout cycles instead? Shorter cycles reduce exposure proportionally and are worth doing independently. They do not prevent the diversion; they limit how much is lost.

How do we measure coverage usefully? By revenue decile rather than seller count. Covering the top decile protects most of the value at a fraction of the enrolment effort.

Why doesn't behavioural detection catch this? Because nothing about the account's behaviour changes. Listings, orders and messages continue exactly as before.

Why is email notification insufficient? The compromise usually began with the mailbox. The notification needs a channel the attacker's session cannot reach.

Does a payout hold hurt sellers? Only on the first payout after a bank change, which most sellers do once in several years.

Where this fits in Manav

Manav proves a specific person authorised a specific action, without a vault, a token or surveillance. The biometric never leaves the device and the platform receives a signature rather than a profile.

What we do not do →

Sources and further reading