Fake seller accounts: farmed storefronts that inherit trust they never earned
Account age is doing work it cannot do
Marketplace risk models lean on age and history because they predict ordinary sellers well. A dormant account that suddenly lists high demand branded goods at volume is not an ordinary seller, but the model reads the history it has, not the change of hands it cannot see. Takeover and purchase leave no event in the record. The password changes, the payout bank changes, the catalogue changes completely, and each change alone has an innocent explanation.
Watch the combination instead: payout destination change, login geography change, and a catalogue pivot inside the same short window. That sequence is the sound of an account changing owners.
What the scheme needs from the platform
It needs a payout lag it can beat and a detection lag it can outrun. The operator lists goods that will not ship, or ships counterfeits, collects orders until complaints and chargebacks reach the threshold that freezes the account, and aims to withdraw before the freeze. Feedback manipulation props the window open: a run of small genuine orders, or purchased reviews, keeps the metrics above the line while the real volume does the damage.
Shortening either lag hurts the scheme more than another onboarding check. Holds on first payouts after a payout destination change, and listing limits that reset when an account pivots category, attack the economics directly.
Signals a marketplace can act on without accusing anyone
Behavioural drift is observable without claiming to know who anyone is. A seller who sold handmade candles for three years and now lists current season trainers at a discount is a review candidate on the facts alone. So is a storefront whose dispatch location, response language, and support hours all change in the same week.
Graduated responses keep this fair: verification step up, temporary listing cap, payout hold pending delivery confirmation. A legitimate seller who bought a business can clear each one. A farm working a two week window cannot clear any of them in time, and time is the whole scheme.
Related reading: How do bots manipulate marketplace reviews to tilt seller ratings?, How do inventory-hoarding bots manipulate marketplace carts?.
Frequently asked questions
Is buying an established seller account allowed?
On most marketplaces, no. Accounts are licensed to the vetted party and are not transferable, which is why account sales happen off platform and quietly. A buyer who acquires a business legitimately is usually expected to open a new account or go through a formal transfer process where one exists.
How do farms keep dormant accounts in good standing?
By doing almost nothing, which is the point. Occasional low value genuine sales keep the account active and build a thin layer of real feedback. The cost of maintaining hundreds of sleeping accounts is small next to what one burned account can extract in a single payout cycle.
What should a buyer check before trusting an aged storefront?
Read the recent feedback for product fit, not just the score. A five star history for phone cases says nothing about the same shop selling designer bags this month. A sudden catalogue change visible in the review history is the most reliable public signal there is.
Can a marketplace detect an account sale automatically?
Not the sale itself, but its footprint: simultaneous changes to payout details, login geography, device, and catalogue. Detection works when those events are scored together. Scored separately, each one routes to a routine support flow and nothing ever adds up.