How do bots fake the sales that win a marketplace bestseller badge?

The badge measures a window, not a business

A bestseller badge on a marketplace listing reads as a verdict from shoppers. It is not. It is a ranking over a short recent window, usually hours or days, inside one category. That window is exactly what bot operators buy, because a short window is cheap to fill and the badge then sells the product for them.

Best-seller status on Amazon is relative and hourly. The listing that sold the most in its category in the last day or so wears the badge, and the algorithm weighs concentrated sales heavily. A book can sell enough copies in a single day to rank in the top 100 and wear the badge while never selling another copy. The overall number one on a given day might have sold around 5,000 copies, but in a small category the winning number can be far smaller. That is why the attack is timed: the operator does not need sustained demand, only a spike that lands inside the window.

The badge then compounds: shoppers treat it as a recommendation, the listing climbs in search, and real orders start arriving. The fake sales were the down payment; the real sales are the return.

Four ways to manufacture a sales spike

Self-purchasing is the simplest version: the seller or an associate buys their own listing through a handful of accounts, or dozens. The orders are real transactions with real money moving, which is why an automated system struggles to reject them. The cost is the marketplace fee on each order plus the product cost, and the operator treats that as advertising spend.

Paid buyers are the scaled version. A service recruits people to buy the product and refunds them off-platform, or pays them a fee on top. The buyer keeps the item or returns it. The review that often follows is part of the package, which is where the FTC's Consumer Reviews Rule comes in: 16 CFR Part 465, effective October 21, 2024, bans fake reviews, buying positive or negative reviews, and fake social indicators, with penalties up to $53,088 per violation.

Claim codes are the quiet version. A seller distributes codes that make the product nearly free to a selected group, and the redemptions register as sales. Amazon prohibits using claim codes to inflate rank, and sellers must not solicit or knowingly accept fraudulent orders, including orders for their own products.

Paid rank services wrap all of it. They sell the outcome, a badge in a named category within a named number of days, and they choose the mix. The buyer never sees which method ran. When the account is suspended, the service is gone and the seller holds the Plan of Action.

Brushing is the version sellers never see

Brushing skips the buyer entirely. The operator creates fake orders that ship to an accomplice or an unsuspecting stranger, with a tracking number attached. The recipient gets a cheap item, sometimes a packet of seeds, while the platform records a sale of the expensive product. The tracking number is the whole trick: a shipped order with a delivery scan looks like a completed sale in every report the platform runs.

The seeds episode in 2020 showed the shape of it: consumers around the world received packages labeled as jewelry that contained mustard, cabbage, and hibiscus seeds. The recipients were never customers. A seller whose listing suddenly accumulates orders to addresses they never marketed to, with no matching reviews or messages, is looking at brushing or something close to it.

The badge pays for itself, which is the problem

University of Illinois research on platform manipulation found marketplaces can profit from the distortion: a top-ranked Sponsored, Best Seller, or Amazon's Choice position is inventory the platform can charge for, and sellers bid for top ranking. The incentive to police rank manipulation competes with revenue from the ranking itself.

That does not make every badge fake. Most are earned. But a badge is not evidence on its own, and the honest listing competing against a manufactured spike is not losing on product. It is losing on a number it cannot see being filled.

How to read a badge before you trust it

Check the review dates first. A real bestseller accumulates reviews over weeks with varied wording. A manufactured one shows clusters: many reviews in forty-eight hours, similar phrasing, accounts with one review each. Then check the seller's other listings. A store with one badge and eleven reviews and nothing else is a narrower story than a store with a year of steady sales.

If you are the seller being undercut, document the spike: screenshots with timestamps, the review cluster, the category, the price changes. Amazon won $75,000 in 2018 against a book marketer found to have artificially inflated reader numbers, which shows the platform will pursue manipulation when the evidence is packaged. Report through the platform's violation channel with dates and order patterns, not motive.

And if you are tempted to buy the spike yourself, count the real cost: the fees, the refunds, the reviews the FTC rule now prices at up to $53,088 each, and the account that holds your entire catalog. A bought badge rents the top of a category for a week and risks the store that took years to build.

Keep reading: Review velocity: how a flood of five-star reviews gets your listing flagged, What is a brushing scam, and why do marketplace sellers get blamed for it?, and How do fake reviews get manufactured and planted on marketplace listings? cover the nearby cases in this series.

Frequently asked questions