The Relisting Problem: Why Takedowns Don't Stick

Joel celebrated on the Thursday.

The notice had gone in on the Monday, the listing came down on the Wednesday night, and by Thursday lunchtime he'd screenshotted the 404 and put it in the team channel with a flag emoji. First takedown. 6 months of being quietly furious, resolved in 3 days.

On the following Tuesday the same product was back. Different listing ID, different title word order, same photographs with a 4-pixel border added, 2 dollars cheaper than before. Same shop.

He filed again. It came down. It came back on the 19th.

"I'm not protecting the brand," he said. "I'm playing whack-a-mole with a man in another timezone who's faster than me."

He was right about that, and wrong about what to do with it.

The economics are against you, listing by listing

A listing costs a seller almost nothing to create. The photographs are already on their hard drive. The description is already written. Posting takes 4 minutes and the marketplace charges pennies, or nothing.

Your notice costs you 15 to 40 minutes if you're organised and have the evidence to hand, and more if you don't. Then it costs a platform review cycle, which runs anywhere from 2 days to 3 weeks.

So the exchange rate is terrible. 4 minutes of their time against 30 of yours, repeated. Anyone who measures success in listings removed will burn out inside 2 months, and that's not a motivation problem, it's arithmetic.

The move is to stop counting listings.

The seller is the unit

Here's the shift, and it's the whole article.

A listing is a symptom. A seller is the cause. One shop can carry 30 copies of your products, and when you take down 1 listing you've reduced that shop's inventory by 3%. When a shop gets terminated you've removed 30 listings and the capacity to make more.

Everything about how you work should follow from that. File your evidence by seller, not by listing. Name the folder after the shop ID. Keep a running count per seller of notices filed, listings removed, counter-notices received, and listings that came back. Those four numbers per seller are the only record that matters, and almost nobody keeps them.

Keeping them is what turns 9 separate complaints into 1 pattern, and a pattern is a different kind of request.

The repeat infringer lever, which brands almost never pull

This is the most useful piece of law in the whole subject and it's in a subsection nobody reads.

A platform's safe harbour from copyright liability in the United States isn't free. Under 17 U.S.C. 512(i)(1)(A), a service provider only keeps it if it has adopted and reasonably implemented a policy providing for the termination of subscribers who are repeat infringers, and has told its users about it.

Read that as an obligation the platform carries and you'll see what it gives you. Every marketplace of any size has a repeat infringer policy, because it must. The policy is only triggered by evidence of repetition, and the only party in a position to document repetition against your brand is you.

So the escalation isn't "please remove this listing again". It's a short message with a table: this seller ID, these 7 notices, these dates, these 5 removals, these 4 relistings of substantially the same material. Under your repeat infringer policy, please review the account.

That request lands in a different queue, read by a different person, with a different set of options. I've seen it work where 11 individual notices had not. It works because you've stopped asking for a favour and started handing over a completed case.

The EU route is structurally similar. The Digital Services Act, Regulation (EU) 2022/2065, obliges online platforms to suspend users who frequently submit manifestly illegal content under Article 23, and requires marketplaces to collect and verify trader details under Article 30. The platform knows who the seller is. Your job is to make the frequency undeniable.

What to actually watch after the first notice

Four things, in rough order of how often they pay off.

The same seller ID. Obvious, cheap, and the one people forget because they bookmarked the listing rather than the shop. Capture the seller name and the seller ID at the moment you find the first listing, before anything comes down. After the removal, the listing URL is dead and the ID is how you find your way back.

Your own photographs, altered slightly. Relisters crop, mirror, add a border, bump the saturation, or paste the product onto a different background. Visual matching still catches most of that, which is why image search is better than text search at spotting a relist. A seller who has reshot the product from scratch is a different and more serious problem, and much rarer, because it costs money.

Your invented words. Colourway names, model numbers, made-up materials. A relister working at speed copies the text and misses the signature. Search your own invented words monthly and you'll find the copies that changed everything except the one thing that identifies them.

Shipping origin and handling time. Two listings from two unrelated shops with the same despatch city and the same 12 to 19 day handling window are usually the same operation. This is how you find the second shop before it has a history.

The counter-notice week

Sometimes a removal produces a counter-notice rather than a relisting, and the clock here is set by statute rather than by the platform's mood.

Under 17 U.S.C. 512(g), once a valid counter-notice is received the platform restores the material in not less than 10 and not more than 14 business days, unless the complainant tells it that a court action has been filed. That's it. There's no appeal to the platform, no further evidence round, and no extension because you were on holiday.

So treat the fortnight after a successful takedown as a live window rather than a finish line. If a counter-notice arrives, you have a decision to make about whether this specific listing is worth a court filing, and for a 40-dollar planter the honest answer is usually no. Which is fine. Log it, note that the seller contests, and let it inform the escalation case you're building on the account.

What you must not do is file a notice you can't stand behind in order to buy 2 more weeks of removal. Section 512(f) makes knowing misrepresentation actionable, and you'd be handing the other side the only real weapon they have.

Making the watch automatic, because you won't do it by hand

Everything above is doable with a spreadsheet and discipline. Almost nobody sustains it past 6 weeks, and the gap between the first month and the fourth is where the copies win.

The mechanical parts are: re-check removed listings on a schedule, flag when the same seller posts substantially the same material again, and keep the per-seller history in one place so the escalation case writes itself.

Brand Protection Tools, which Verity AI runs, tracks each notice through Filed, Removed, Counter-notice received, and Came back, with dates, and on its Growth and Pro plans the relisting watch re-checks removed listings and flags when the same seller puts them back. Evidence is captured when the listing is found rather than when you need it, which is the difference between a case and a memory. The searching covers Etsy and Alibaba.com today, with AliExpress, Temu, eBay and Amazon on the list to come.

It won't file for you, and it shouldn't. The notice is a sworn statement and that stays with the brand owner.

The number that tells you whether it's working

Not listings removed. Not notices filed.

Sellers terminated, and the share of your known copycat sellers who have stopped relisting.

Joel's seller came back 4 times. On the fifth notice he sent the table instead of the form: 1 seller ID, 5 notices, 4 removals, 4 relistings of the same photographs, dates on all of it. The shop was gone inside a week and the 23 other listings in it went with it.

He now has 3 sellers under watch and 1 he can't make stick, and his monthly review takes 20 minutes. He stopped putting flag emojis in the channel, which is probably the healthiest outcome of the lot.

For the governance version of the same point, including the four numbers a board should be asking for, read counterfeits as a governance issue. If you'd rather have a sequence than a set of principles, the 30-day plan puts this in order. And the wider argument sits in When the AI shopping assistant recommends the fake.