Unauthorised is not illegal

In the United States the first-sale doctrine means a party who lawfully acquires a genuine trademarked good may generally resell it; the Supreme Court confirmed in Quality King (1998) and Kirtsaeng (2013) that this holds regardless of where the goods were made. In the EU, Article 15(1) of the Trade Mark Regulation exhausts the proprietor's rights once goods are on the EEA market with its consent.

A reseller who buys your product legitimately from your distributor and lists it on Amazon without your permission is, in most cases, doing nothing unlawful. They have breached no contract with you, because they have none. Amazon will not remove them, and is correct not to. Brands that do not internalise this spend years filing complaints that go nowhere — and filing intellectual-property complaints against sellers of genuine goods purely because they are unauthorised is itself a marketplace policy violation that can cost you Brand Registry access.

Five seller types, five different remedies

Colour follows the data palette: green lawful, amber conditional, red unlawful. The remedy column has almost no overlap — a brand running one undifferentiated programme is applying the wrong tool most of the time.

Where the inventory comes from

There are only a handful of routes, and each leaves a different signature. Distributor diversion shows as order quantities outrunning known sell-through, concentrated at quarter-end or volume-tier boundaries. Cross-market arbitrage shows as sellers appearing where your price is highest, with lot codes traceable to the cheapest market. Returns and liquidation show as mixed condition and no consistent lot pattern. Retailer clearance is seasonal. And FBA commingling pools identical GTINs from multiple sellers into shared bins, obscuring the origin of any unit and defeating lot-code tracing unless your authorised sellers are contractually required to opt out.

Amazon does not create unauthorised sellers. It creates a low-friction venue for them. The inventory comes from the brand's own supply chain.

The double-counting error

Suppose an unauthorised seller moves 400,000 euro of your product on Amazon in a year. The instinctive framing is 400,000 euro of lost revenue. It almost never is: those units were manufactured by you, sold to a distributor at your wholesale price, and paid for. The seller is not taking sales from you. They are taking sales from your other channel partners, or from your own direct or first-party offer.

The genuine costs are different in kind, and each has its own mechanism and its own measurement.

• Margin mix shift — units that would have sold at direct or first-party margin sold at wholesale margin instead.
• Price erosion — real, compounding and mostly forward-looking: it damages next year's pricing, not this year's revenue.
• Channel attrition — authorised partners reducing orders or demanding margin support, which appears in the ledger as ordinary account softness.
• Advertising waste — paying to acquire demand a third party converts. The one cost you can quantify to the euro from data you already have.
• Warranty and service absorption — cost calibrated to a full-margin sale, on units that earned wholesale margin.
• Review and rating damage — negative reviews attach to your listing permanently, not to the seller who caused them.
• Compliance and recall exposure — you cannot execute a recall on units you cannot trace.
• Brand equity — real, hard to quantify, and easy to overstate.

Repricing turns one seller into a market-wide floor

Marketplace sellers overwhelmingly run automated repricing configured to match or undercut the lowest competing offer. When one seller enters a listing with cheap diverted stock, the lower price propagates across every seller on it within minutes; the floor is set by whoever has the lowest cost basis. The European Commission's 2018 decisions against Asus, Philips, Pioneer and Denon & Marantz rest explicitly on that amplification mechanic. Track minimum street price per SKU per market against your reference over time, overlay seller entry and exit dates, and the correlation is usually visible without statistical work.

The most-cited statistic in the category has no source

The figure that 80 to 83 per cent of Amazon sales flow through the buy box appears in essentially every vendor guide. Traced back, it is repricing and tooling vendors citing each other; Amazon does not publish the number, and there is no methodology or sample definition. The directional point survives — buy-box ownership dominates conversion, and dominates more on mobile where alternative offers are not visible without deliberate navigation. But a brand should not put a vendor-circulated percentage into a board paper. Amazon reports your buy-box percentage in Seller Central; the relationship to your units is observable in your own data within weeks, and that is the only number that belongs in your business case.

Compliance moved from background risk to binding law

The EU General Product Safety Regulation (EU) 2023/988 has applied since 13 December 2024, identically in every member state. Every product must be associated with a responsible economic operator established in the EU whose details appear in the online offer; manufacturers must maintain traceability; and where a recall is necessary, an effective remedy must reach the consumer. The Commission reported a record 4,671 Safety Gate alerts in 2025, up 13 per cent, with enforcement actions up 35 per cent. If product reached EU consumers through a channel you have no visibility into, your recall is incomplete by construction — a regulatory failure caused by a distribution failure you may have tolerated for years as a pricing annoyance. Note that GPSR applies in the EU and Northern Ireland, not in Great Britain.

An illustrative model — every figure an assumption

A 50m euro brand, three Amazon marketplaces, fourteen unauthorised sellers across its top forty SKUs, marketplace minimum averaging 18 per cent below reference, holding the buy box 45 per cent of the time, 600,000 euro of annual marketplace ad spend, 8m euro of revenue through accounts that also compete online.

Two things tend to hold across real cases. The largest single line is advertising waste or channel attrition, never the seller's own turnover. And the forward price-erosion figure often exceeds the entire current-year cost, which is why brands that tolerate the problem for three years find it much harder to fix in year four.

The data you need, and the ranking that goes wrong

Enforcement-grade monitoring captures, per listing, over time: seller identity, buy-box holder and share, total offer count, price against your reference, fulfilment method, timestamp and market, listing content changes, and review velocity. A single snapshot is nearly useless; the value is in the time series — entry and exit dates, price movement, and the correlation between them. Rank findings by cash impact, not percentage gap: a 40 per cent discount on a 12 euro accessory is a smaller problem than an 8 per cent discount on a 400 euro unit holding the buy box on your best-selling listing. And capture the absence too — the retailers where your product should be listed and is not.

Seller anonymity has largely ended

The US INFORM Consumers Act, in force since 27 June 2023, requires marketplaces to verify high-volume sellers and, above 20,000 dollars of annual revenue, to disclose the seller's full entity name, physical address and working contact details on the listing or order confirmation — and, on a purchaser's request, whether a different seller supplied the product. That last provision is a legislated test-buy tool most brands miss. In the EU, Article 30 of the Digital Services Act, fully in force since 17 February 2024, requires platforms to collect a trader's name, address, contact details, identity document, payment account and trade-register number before granting access, and to display the identifying details in a clear form. Combined with company registries and VAT lookups, a determined brand can now usually put a real legal entity behind a storefront.

Test buys that survive contact with a lawyer

1. Order from the target seller specifically, not merely from the listing, and verify the seller of record on the confirmation and invoice.
2. Preserve the shipment unopened where possible, or open it under photographic record.
3. Photograph outer packaging, shipping label, invoice, product packaging and every lot, batch or serial code with the order documentation in frame.
4. Record order date, seller name, price paid, buy-box status at purchase, and a timestamped listing screenshot.
5. Establish chain of custody: who received it, where it is stored, who has handled it.
6. Repeat over time. A single buy proves one unit; a pattern across months proves a supply route.

Lot-code tracing finds the leak

If your product carries a batch, lot or serial code and you know which ranges went to which distributor, a test-buy unit's code identifies the account that put it into the grey market. You then have a contract claim against a party you actually have a contract with — enormously more tractable than a trademark claim against an anonymous storefront. Two constraints: code removal is itself actionable in the US (Zino Davidoff v. CVS, 2d Cir. 2009, held removal of codes from genuine goods to be infringement because it defeated the brand's quality-control programme), and FBA commingling defeats serialisation unless authorised sellers are required to opt out.

Leak-source attribution rate — the share of unauthorised inventory traced to a specific distributor account — is the metric that distinguishes a real programme from a listing-removal treadmill. Almost nobody tracks it.

Data quality decides whether any of this is usable

You take a report into a negotiation, the distributor finds one wrong row, and every other figure is now disputed. One bad price point costs more credibility than fifty good ones earn. That demands automated anomaly flagging before anything reaches a human, manual verification weighted toward newly added retailers, low-confidence rows marked in the deliverable rather than silently dropped or included, and errors corrected the day they are found. Be wary of choosing a provider on a self-reported accuracy figure; ask instead what happens when a number turns out to be wrong.

Ordered by return on effort, not by drama.

Fix the source first

Every unit an unauthorised seller lists was sold by someone with an account. Remove the listing and leave the supply route open, and another listing appears. Brands run this treadmill for years, measure listings removed as success, and never move the underlying numbers. The terms that matter in an authorised-dealer agreement: no resale to unauthorised parties defined by a maintained list; no marketplace sales without written consent; preservation of lot and serial codes; mandatory opt-out of FBA commingling; audit rights; liquidated damages above the diversion margin; immediate termination for diversion.

The leak is usually visible in your own order book before you find it on a marketplace: orders that outrun demonstrable sell-through, quantity-break purchasing at tier boundaries or quarter-end, sudden growth in SKUs with no matching retail activity. The order book tells you which accounts could be leaking; the marketplace data tells you which leaks are reaching consumers, where, and at what price. Neither is sufficient alone. Joining them — an order anomaly matched against a new seller and a price break in the same SKU and market in the same week — is what converts a suspicion into a conversation you can hold with a distributor.

The leak is frequently one of your largest customers. Most programmes fail not for lack of evidence but because nobody wants to confront a 4m euro account over a 300,000 euro problem. Decide in advance, at commercial-leadership level, what you will do when the evidence points at a strategic account.

Channel and pricing architecture

Diversion is an arbitrage business; it exists because of a price gap. Rationalise the distributor list, accepting the real trade-off in concentration risk and coverage. Narrow cross-market wholesale differentials — you may change your prices, but in the EU you may not restrict cross-border passive sales, which is a hardcore restriction under Article 4 of the Vertical Block Exemption Regulation. Design promotions to resist arbitrage. And read your liquidation contracts: much unauthorised inventory originates in a returns or salvage agreement nobody in the commercial team has seen.

The US legal route is strong, if you built the programme in advance

Trademark law offers two well-litigated exceptions to first sale. A reseller infringes when the goods are materially different from those the brand authorises — a low threshold, and absence of the manufacturer's warranty has been held material (Beltronics, 10th Cir. 2009). Goods sold outside a legitimate quality-control programme are not genuine for trademark purposes. Otter Products v. Triplenet Pricing (D. Colo. 2021) is the template: Otter established non-pretextual quality controls that all authorised sellers followed, its warranty covered only authorised purchases, and Triplenet's listings falsely claimed that warranty. What Otter had to prove is what you must build months ahead: a documented programme, applied consistently, that the brand demonstrably follows itself. One invented after the fact is pretextual and fails.

A properly evidenced cease-and-desist backed by test-buy documentation and a real quality-control programme resolves a meaningful proportion of cases within weeks, because most marketplace operators run a thin-margin volume business and will move to another brand rather than fight. Litigate selectively, against persistent high-volume operators or where a precedent is needed.

The EU route is weaker on price and stronger on structure

Exhaustion is the rule, and there is no general material-difference doctrine. The narrow exception in Article 15(2) — legitimate reasons, particularly damage to a luxury mark's reputation (Copad v. Dior, 2009) — is strongest for genuine prestige brands operating a coherent selective system. The real structural tool is selective distribution on objective qualitative criteria applied uniformly, under which a supplier may prohibit sales via discernible third-party platforms (Coty, 2017). The limit: a blanket ban on all online sales is unlawful, as Rolex France's 91.6m euro fine in December 2023 made plain. Enforcing qualitative criteria against non-compliant members of the network is lawful and effective. Enforcing price is not — see the companion MAP paper.

What does not work

• Mass cease-and-desists with no supply cut-off. Moves no metrics and trains sellers that your threats do not escalate.
• Buying up unauthorised inventory. Funds the diverter and reliably produces more diversion.
• IP takedowns against genuine goods. A policy violation that risks your Brand Registry and invites counterclaims — Triplenet counterclaimed, and that is the standard response.
• Marketplace support tickets as an enforcement channel.
• Informal understandings with diverters about price. Unenforceable, and in the EU exactly the conduct that constitutes indirect resale price maintenance.
• Measuring listings removed. An activity metric masquerading as an outcome metric.

Exiting the marketplace is not the variable

Birkenstock withdrew from Amazon US from January 2017 and threatened to cut off any partner who supplied it; that supply-side threat is what made the exit work. Nike went the other way, selling first-party from 2017 to displace third parties, and found removed listings simply reappeared under new names; it ended the pilot in November 2019 and returned to Amazon in 2025. Nike had a first-party relationship and still could not control the listing environment, because the inventory was already in the grey market and the supply route was open. Owning the buy box treats the symptom. It does not close the leak.

Channel integrity sits across commercial, e-commerce and legal, and consequently often belongs to none of them. It needs one owner with authority across all three, and an escalation path for the strategic-account case defined in writing before the evidence arrives.

The first ninety days

1. Days 1 to 30 — see the problem. Instrument monitoring across every marketplace and market. Baseline seller count, offer count, buy-box hold rate, minimum price spread, recurrence and time-to-removal. Pull 24 months of order data. Read your distributor and liquidation contracts and inventory what they actually permit.
2. Days 31 to 60 — find the source. Test-buy the top five sellers by volume. Trace lot codes. Retrieve INFORM and DSA disclosures. Map accounts to operators. Cross-reference the order-book anomalies. Name the top two or three leak candidates.
3. Days 61 to 90 — decide and act. Present the cost model with your own numbers. Get the strategic-account decision in advance. Begin the agreement rewrite. Open conversations with the identified leak sources. Issue evidenced letters only where you have a genuine claim.

What to demand of a monitoring vendor

• Seller-level identity resolution, not just listing-level price capture.
• Buy-box holder history and full offer-count and listing-change history, not current state.
• Coverage of every marketplace and country you actually sell in, including the ones you did not authorise — and a written turnaround for adding one that is missing.
• Export in a form your legal team can use as evidence: timestamped, complete, reproducible.
• A commitment you can leave after month one. If a provider will not let you go before you have seen whether the data is any good, they are asking you to bet on it blind.

Build the internal case on the components you can measure with your own data — advertising waste, margin-support concessions, account attrition, buy-box share — not on vendor statistics or the seller's own revenue. A business case built on the 400,000 euro the seller turned over will not survive a competent CFO, who will ask whether you were paid for those units. One built on 198,000 euro of demonstrably wasted ad spend and 120,000 euro of margin support will.

Traceability is becoming infrastructure. GS1 Sunrise 2027 targets retail point-of-sale scanning of 2D barcodes by the end of 2027, and it converges with binding regulation: the EU Digital Product Passport under the Ecodesign for Sustainable Products Regulation (EU) 2024/1781 requires a unique product identifier accessible via a data carrier, with product-specific acts rolling out through 2026 and 2027. Brands are about to be required, for sustainability reasons, to build exactly the unit-level traceability that anti-diversion programmes have always had to justify on their own. Scope the two projects together.

Product-safety enforcement is tightening, seller anonymity keeps eroding as DSA and INFORM enforcement matures, and seller-network clustering — matching accounts to operators across marketplaces — is where tooling is improving fastest. None of it closes a leak. Regulation is making detection cheaper and attribution easier. It is not making the conversation with your largest distributor any easier, and that conversation remains the actual bottleneck.

Everything above describes what a brand needs to see. This is what we built to see it, and it is deliberately last: the analysis stands without it.

Seeing the market

For every monitored product on every monitored retailer: who is listing it, who holds the buy box, at what price against your reference for that market, and how that has changed — with seller entry and exit dates retained, because those dates are where the causation lives. Fulfilment method, so a distributor selling directly is distinguished from an operator running platform-fulfilled diverted stock. Listing content changes, because content drift is often the first symptom of a seller you have not yet identified. Review count and rating trend per listing. Coverage gaps. Out-of-stock across the network, which matters because an authorised partner running dry is frequently the moment an unauthorised seller takes the box and keeps it.

Identifying who you are dealing with

We resolve a storefront to a legal entity — registry records, VAT numbers, addresses — drawing on the disclosures INFORM and DSA Article 30 now mandate. We map multiple storefronts to a single operator using shared addresses, contact details, listing copy, synchronised repricing and common officers, which is what makes account-level action worth doing at all. And we compare the same SKU across countries, surfacing the price differentials that cause diversion rather than the symptoms that follow it.

Closing the loop

We take in your sales and order data and join it against the marketplace signal. The order book alone tells you which accounts could be leaking; marketplace data alone tells you product is reaching consumers through channels you did not authorise. Joining them identifies the source. That join is the difference between a monitoring dashboard and a programme that closes leaks, and it is why the metric we report is leak-source attribution rather than listings removed.

Evidence-grade export — timestamped, complete, reproducible — for test-buy documentation, cease-and-desist correspondence and the record a US quality-control claim requires. Alerts as events happen. Scheduled export into your own systems. And a written monthly distribution report: the artefact a commercial director takes into a distributor meeting.

How we work

Built with you, not configured for you. A new retailer, a new alert type, a new export format gets built, implemented as a general capability and then configured per client, so the product improves rather than fragmenting. A deliberately small client base, because the value is in understanding how a specific brand distributes. We flag low-confidence rows in the deliverable, state our matching accuracy plainly, and fix anything wrong the same day. Those are the commitments that survive contact with a distributor negotiation.

None of this is possible without seeing the problem first, in your own markets, on your own retailers, over time. That is the part we do.