How to deal with unauthorized sellers on Amazon at scale
Two sellers turn up on the same ASIN in the same week. One is shipping a copy of your product in packaging that reproduces your logo. The other is shipping your genuine product, bought from a distributor who wasn’t supposed to wholesale it. On your dashboard, they’re one problem with two rows.
Amazon sees two problems, and only one of them is Amazon’s. Its reporting tools act on counterfeits, on trademark and copyright misuse including your own photography and listing copy, and on offers that misdescribe what’s in the box. A reseller moving your real product outside your distribution agreement falls outside all of it, whatever it costs you in margin.
So the queue you’re staring at isn’t one queue. It’s three lanes: what Amazon will remove for you, what your contracts have to handle, and what the same seller does next. Two of them lead somewhere Amazon can’t take you.
What Amazon acts on, and what it hands back to you
Start with what Amazon handles without you, because it changes what’s left over. Amazon says it identified, seized, and disposed of more than 15 million counterfeit products worldwide in 2025, in its 2025 trustworthy shopping experience report. That’s Amazon’s own count inside Amazon’s own store, and counterfeits are only the first of your three lanes.
The second lane is the one that surprises people. A seller who bought your genuine product and resold it unchanged isn’t infringing anything in Amazon’s view. The American Intellectual Property Law Association, writing up the wording on Amazon’s infringement form, records the position plainly: violations of a distribution agreement aren’t intellectual property infringement, and enforcement is a matter between the manufacturer and the retailers.
That isn’t Amazon being difficult. Buy a branded item legitimately, and you can generally resell it as is. Altered packaging, removed codes, expired stock, or missing inserts can change that, though whether a difference counts is a question for your counsel.
Which leaves the third lane, the one nobody staffs: what the same operator does once the listing disappears. That third lane is where Bolster AI does its work.
How the same operator survives a takedown
Pressure on one marketplace relocates inventory. It rarely ends it.
The listing goes, the operation doesn’t
An operator who loses an ASIN still holds the stock, the supplier relationship, and the photography. Opening a new storefront under a new name takes an afternoon. Your case closes; their week barely changes.
The catalogue reappears next door
The same images turn up on a second marketplace within days, because the listing content was the expensive part and it’s already made, and a program scoped to Amazon sees none of it.
The storefront that doesn’t need a marketplace
The version that costs you most leaves marketplaces entirely. A domain that reads as an outlet or a clearance site, your imagery, a checkout, and enough of your brand to convert. There’s no Report a Violation form for that, which is why marketplace monitoring and takedowns are a different discipline from working an Amazon queue.
Three surfaces, one operator, and only the first of them is visible from Seller Central.
Your file records what you filed, not what happened
Here’s the structural problem underneath all of it. Every lever in this work belongs to Amazon. Report a Violation, Project Zero, and Transparency, the unit-level serialization program, all belong to Amazon, and they run inside Amazon’s store on Amazon’s schedule. None of them is a control you own.
What comes back to you is a status on your own submission. Not an account of what the seller did next, because that concerns an account that isn’t yours. So your record grows a row every time your team acts, and never grows a row for what actually happened to the operator.
Your submission history goes up and to the right whether the operator is shrinking or expanding, because it was never measuring the operator. You are counting your own effort and calling it enforcement.
The gap between what you filed and what changed is where unauthorized sellers on Amazon actually live. It’s the gap Bolster AI is built to close.
What enforcement at scale actually covers
A program that holds up at volume covers five things. Each is a standard rather than an aspiration, and each is one Bolster AI is built to meet.
Discovery that catches the relist, not just the listing you reported
A seller removed once will change the title and the storefront name. What they won’t change is the photography, because reshooting a catalogue costs real money. So the standard isn’t detection that finds listings. It’s detection that matches the image and the seller’s behavior, then files the hit into the case you already have open. Bolster AI’s marketplace monitoring does that last part, which is the part a keyword alert can’t.
Campaign-level linkage, not just a record that survives a rename
One file per operator is the floor, not the standard. The listings, storefronts, domains, and social accounts have to arrive already connected, because a rename only costs you history if the storefront was the identity to begin with.
Marketplace disclosure helps without finishing the job. The Federal Trade Commission’s guidance on the INFORM Consumers Act makes marketplaces publish a high-volume seller’s name and contact details once they clear $20,000 a year on the platform, with a full address unless the seller works from a residence, where you get country and state. That’s half an identity, and Bolster AI Signals matches the other half against imagery, product copy, and shared infrastructure.
A routing judgment made before anything is filed
Counterfeit, misdescribed offer, and genuine goods outside your channel have to be separated before a request goes out under your name. It’s a judgment rather than a match, which is why it’s worth asking about first. Bolster AI’s brand protection workflow puts that call before submission rather than after a rejection. Get it right and the filing sticks. Get it wrong and you’ve spent accuracy you needed, and nobody sends you a note about it.
One case covering the ASIN, the domain, and the social account
Two subscriptions and two dashboards isn’t coverage. It’s reconciliation work you’ve hired someone to do. The standard is that the lookalike site and the fraudulent social storefront open inside the Amazon case, carrying the same evidence and seller identity. That’s how Bolster AI runs domain monitoring and social media monitoring, off the seller record, so the second surface arrives as an update to an open case rather than a new investigation.
An escalation path for a report that’s ignored
When a platform sits on a report, you need somewhere else to send it, and a public abuse form isn’t somewhere else. Take the seller’s own site, the one they move inventory to when the ASIN goes. The registrar and the host sit above that site, so a request landing there removes the whole destination, not one page of it.
Automated takedowns submit directly into those providers, a route a brand working through public forms can’t build for itself. Ask a vendor what happens on day 10 of an ignored report, and listen for an answer other than resubmitting.
Where most programs break down
Four things brands are doing now, and what each one leaves on the table.
Report a Violation, worked case by case. The right instrument, the wrong unit. A copyright report on your own photography will take the listing down. It won’t tell you who relisted it that afternoon.
Project Zero self-service removal. Genuine leverage, and gated. Amazon asks that you’ve used Report a Violation for suspected trademark infringement in the last six months with at least a 90% acceptance rate on those reports. Filing against resellers of genuine goods spends the accuracy that qualifies you, so the instinct to file against everything disqualifies you from the one program that lets you pull a listing yourself.
Tightening the distribution agreement. Correct for the second lane, and the slowest thing you own. By the time you’ve found which distributor leaked the inventory has shipped, and the contract reaches none of the domains or social accounts the same operator runs.
A spreadsheet per case. It records your filings, and nothing else. Ask it which sellers came back last quarter and it has nothing, because nobody wrote that down.
None of these is wrong. They’re all scoped to Amazon, and the operator isn’t. Bolster AI is built for that mismatch specifically: seller-keyed detection across marketplaces, domains, and social accounts, and an enforcement path into the hosts and registries behind them, in one workflow rather than two products you integrate yourself.
Bolster AI detects external threats including fake listings, lookalike domains, fraudulent social accounts, fraudulent ads, and fake mobile apps, connects the related infrastructure into a single campaign, and removes it. Detection and takedown run as one workflow rather than as two promises you have to reconcile yourself, with automation carrying the volume and analysts taking the cases that need judgment. For a marketplace program, that resolves into four things.
Discovery that catches the relist. Marketplace monitoring reads product imagery, brand keyword abuse, and seller behavior, so a listing you’ve never seen surfaces the same way a known one does.
Listings, domains, and accounts that arrive already linked. When the same seller shows up as a second storefront, a lookalike domain, and a social account posting your product photography, Bolster AI treats that as one campaign rather than three unrelated reports. The file survives a rename, which is the whole point of keying it to the operator.
Submission into the places that have no Amazon form. Bolster AI reports direct partnerships with over 1,500 registries and hosting providers and a 98% takedown success rate. Those numbers describe how reliably a request reaches somebody who can act on it. They aren’t a clock on the listing coming down, because removal still happens on the registrar’s or the platform’s schedule rather than yours.
Somebody still watching after the listing goes. Reappearance under a new account is flagged against the existing file rather than opening a fresh one, so the second and third rounds cost your team less than the first did. That’s the compounding a spreadsheet never gets: the evidence, the imagery match, and the identity details are already written down, so the next filing starts half finished. Bolster AI reports that SoFi cut analyst workload by 20% running its external threat program this way. Workload is the number that matters when the queue is the constraint rather than the policy.
Now the limitation because it decides how you plan. Bolster AI can’t turn a reseller of your genuine product into a case Amazon will act on. Nothing can. That lane is a contract problem ,and it stays one. Any vendor who says otherwise is describing a product Amazon’s policy doesn’t allow to exist.
None of this replaces what you already run. Brand Registry, Project Zero, and Transparency stay yours, and the difference Bolster AI makes isn’t in the Amazon lane at all. It’s in the surfaces those programs were never built to reach, which is where the same operator tends to be by the end of the week.
Treat the seller as the case, not the listing
One number tells you whether this is working, and it isn’t how many cases you filed. It’s how often a seller you removed came back under a name nobody recognized. The uncomfortable part is that a program scoped to Amazon can’t produce that number at all, because you can’t count what you were never watching.
Amazon’s forms stop at Amazon’s border, and the storefront that reopens on another marketplace or on a domain of its own is where most brand programs run out of route. See the escalation path Bolster AI takes when a report stalls and the seller keeps selling.
TL;DR: Amazon’s reporting tools act on counterfeits, trademark misuse, and offers that misdescribe the product, so unauthorized sellers on Amazon split into cases Amazon will remove and cases it treats as your contract problem. Filing carelessly against the second group is expensive, because Project Zero is gated on a 90% acceptance rate on your recent trademark reports. The harder problem is that every lever Amazon gives you stops at the edge of Amazon’s store, while the operator keeps the stock, the supplier, and the photography. Running this at scale means keeping one record per seller rather than one per complaint, and watching the marketplaces, domains, and social accounts that record points at. Bolster AI connects those surfaces into a single case and files the takedowns that Amazon has no form for.
Frequently asked questions
What counts as an unauthorized seller on Amazon?
In practice, it covers three different sellers that get filed as one. Someone selling counterfeits, someone selling a genuine item that doesn’t match the detail page, and someone selling your real product outside your distribution agreement. Amazon acts on the first two through trademark, copyright, or patent reports depending on what’s wrong. The third is a matter between you and your dealers, which is why routing matters before filing does.
Will Amazon remove a seller who’s offering my genuine product?
No, not on intellectual property grounds. Amazon’s position is that a distribution agreement is a contract between you and your retailers, and that enforcing it isn’t Amazon’s role. The exception worth knowing is materially different goods, meaning altered packaging, removed codes, expired stock, or missing inserts, where the item arguably isn’t the same product anymore. Ask your counsel before filing on that basis.
Do I need a registered trademark to report an unauthorized seller?
For Brand Registry, Amazon asks for a logo that includes your brand name, permanently affixed to your products or packaging, plus a pending or registered trademark issued by the trademark office of a country with a corresponding Amazon store. A pending application is often enough to enroll. Project Zero is the one that needs a registered mark, so if your trademark is still in flight, self-service removal is the lever you’re waiting on.
Can Bolster AI get an unauthorized reseller removed from Amazon?
No, and be suspicious of anyone who says they can. If the seller is moving your genuine product, no vendor can make Amazon treat that as infringement. What Bolster AI does is separate that case from the ones Amazon will act on before a filing goes out under your name, and then work the same operator’s listings, domains, and social accounts on the surfaces where enforcement is actually available.
How do I protect my Report a Violation acceptance rate at volume?
Route before you file, and make the routing rule written rather than remembered. Every filing aimed at a reseller of genuine goods spends the acceptance rate you need for Project Zero eligibility. A test buy is what turns a suspicion into evidence somebody else can check, so budget for the purchases rather than filing on a hunch. A written rule only covers the cases somebody opened, though, which is why Bolster AI puts the routing call inside detection rather than leaving it to whoever works the queue that day.
The same seller keeps coming back under new names. What actually stops that?
Nothing stops it. You make it unprofitable. That means keeping one record per operator rather than one per complaint, capturing the seller identity details the marketplace has to disclose, and watching the storefronts, domains, and social accounts the same operator runs. Bolster AI keys detection to the seller for that reason, so round three costs your team less than round one did.
