Brand Protection

How to choose a brand protection company: 9 criteria

Branddi · Published on · Updated on

How to choose a brand protection company: 9 criteria

Choosing a brand protection company comes down to demanding numbers where the market usually offers adjectives. The nine criteria below — coverage with authorisation, confirmed removal rate, SLA per channel, behaviour when a notice is denied, recurrence, evidence, false positives, who operates it and portability on exit — separate a verifiable proposal from a nice presentation.

The difficulty in buying in this category is that almost every proposal looks the same: everyone monitors everything, everyone removes fast, everyone uses artificial intelligence. The difference appears when you ask for the number behind the sentence.

Channel coverage — and authorisation to act on each one

Criterion 1. Coverage without authorisation is a shop window. Monitoring a marketplace isn't the same as being able to file a removal there: each channel requires proof of ownership, and some require formal power of attorney.

Ask separately: which channels do you monitor and on which are you authorised to initiate removal. The two lists are rarely identical, and the gap between them is exactly the work that will fall to your team.

How do you calculate the real success rate?

Criterion 2. Notice volume isn't a result. The correct calculation is confirmed removals divided by notices issued, in the same period.

A vendor that sends 500 requests and achieves 200 removals has a 40% success rate — even if their report highlights the 500. Ask for the rate, not the volume.

And ask for it per channel. A consolidated 90% rate can hide 99% on an easy channel and 40% on precisely the one where your problem lives. Ask for the last six months too: any vendor can show you one good month.

Is the SLA per channel or a single number?

Criterion 3. A single average hides the slow channels. And here's a fact that changes the conversation: none of the major platforms publishes an official removal deadline.

What each one publishes is the channel the complaint goes through, and they differ significantly:

Note the asymmetry: Amazon offers near-immediate removal for brands in Project Zero, but requires a high accuracy record to grant access. YouTube separates copyright claims from trademark claims, with distinct flows and timelines. There is no single "takedown time" — there's a time per channel, per claim type and per evidence quality.

Since platforms make no public time commitment, the commitment has to come from the vendor, in writing. Demand median and 90th percentile per channel, not the average. The average is dragged down by easy cases; P90 shows how long it takes when the case is hard — which is exactly when you'll need it.

If a vendor presents a table of deadlines per platform, ask where the number came from. If it's from their own operation, good — that's exactly what you want to buy. If it's from an industry blog, it's a guess dressed as data.

And separate the stages: acknowledgement, action and escalation are three distinct moments. A contract measuring only the last gives you no visibility along the way.

What happens when the platform denies the first notice?

Criterion 4. This is the question that most exposes real differences between vendors, and it's almost never asked.

Marketplaces frequently reject trademark claims. What separates a mature operation from an amateur one is having a plan B: a copyright claim over the listing's photos or text often succeeds where the trademark claim failed, because it changes the legal basis and the team reviewing it on the other side.

If the answer is "we insist with the platform", the vendor has no plan B.

Is recurrence measured by URL or by infringer?

Criterion 5. The most neglected point in the sector. A removed listing isn't a solved problem — the same actor typically returns within hours, under another account.

A vendor reporting only by URL will show you a thousand removals while you still have the same problem. Ask for percentage of reappearance at 30 and 90 days, per infringer, and a contractual obligation to re-notify at no extra cost.

Evidence quality

Criterion 6. Evidence degrades fast: domains go down, content changes, WHOIS gets redacted. That's why capture has to happen at the moment of detection, not days later when someone got around to looking.

The minimum acceptable package: dated screenshot, WHOIS record, DNS records, HTTP headers and page content, all exportable and with chain of custody.

Refuse evidence that arrives as a screenshot pasted into an email. If the case escalates to litigation, that won't hold.

What's the false positive rate — and who is liable for it?

Criterion 7. Speed without precision is risk, not virtue. There was a public case where a vendor's enforcement action temporarily took down a legitimate indie games platform, which opened a serious discussion about takedown accuracy.

Two things in the contract: a cap on false positive rate and liability for wrongful removal. A vendor confident in its precision accepts both clauses.

And there's a silent cost: every false positive consumes analyst hours on your side to review and reverse. A high false positive rate is expensive even when it causes no external damage.

Who operates it: you or the vendor?

Criterion 8. There are two products sold under the same name.

A tool flags the problem and hands you the dashboard. It's cheaper and shifts the work of filing, following up and escalating to your team.

A managed service does the work. It costs more and your team becomes a reviewer, not an operator.

Neither is better. The mistake is buying a tool thinking you bought a service — the low price becomes hidden cost in expensive people's hours, and the project dies when whoever ran it goes on holiday.

What do you take with you if you leave?

Criterion 9. The question nobody asks at signature and everybody regrets at exit. Case history, evidence packages and notice records are yours, and you'll need them if you switch vendors or if an old case becomes litigation.

Demand portability in the contract: exportable format, delivery deadline and minimum retention after termination.

How to use the nine criteria in practice

Don't request proposals from five vendors and compare PDFs. Run a pilot measured on your own brands, with 2 or 3 vendors, on the same set of cases and over the same period.

Score on four things, in this order:

A demo runs on the vendor's best case. A pilot runs on yours.

Red flags

Three behaviours that show up early and work as a filter:

Frequently asked questions

Do I need a registered trademark to contract?

To monitor, no. To remove, in most channels yes — marketplaces and ad platforms require proven ownership to accept an intellectual property complaint. If registration is still in progress, that limits what's possible in the short term and needs to be on the table from the start.

How long should a pilot last?

Thirty days works for most cases. It's enough time to measure detection, first removal and — most importantly — the first recurrence. A two-week pilot shows detection and hides what happens next.

Can I start with just one channel?

You can, and it's often the right call when budget is tight. Choose the channel where damage is greatest today, not the one with highest volume. High volume on an irrelevant channel produces a report, not a result.

How do I know the problem justifies contracting?

Before any proposal, gather three numbers you already have: diverted clicks in paid search on your brand name, volume of irregular listings on the marketplaces where you sell, and support tickets from customers who fell for a scam. If all three are low, you don't need this yet.

If you want to see how these criteria translate into a real operation, talk to us to size up your case.

Back to blog