Why Trademark Risk Starts Before Filing

Where pre-filing risk actually originates

Three early choices carry disproportionate weight: the candidate name itself, the scope of goods and services, and the order of jurisdictions targeted for first filing. Each compounds. A name that sounds distinctive in English can collide with a registered mark in Spanish-speaking markets. A class list that quietly includes adjacent goods can pull the application into examiner territory the brand team never intended to enter.

These decisions are usually made before legal is brought in, which is the first structural problem. By the time clearance starts, the team is no longer evaluating risk — they are defending a name the marketing director has already shown to the CEO. The cheapest fix is also the least glamorous: pull legal into the shortlist conversation, not the final-name conversation.

The second source of avoidable risk is overscoping the goods-and-services list to give the brand 'room to grow.' Padding a Class 9 application with software, hardware, and downloadable content the company will not ship for three years invites refusals and oppositions that a tighter filing would never see.

What a working pre-filing workflow looks like

A mature pre-filing process has four discrete stages, and each has its own exit criteria. Stage one is a knockout screen of the candidate list against identical and near-identical marks in the priority jurisdictions. The goal is not certainty — it is to kill the obviously dead names cheaply, before anyone burns time on creative review.

Stage two is a phonetic, visual, and conceptual similarity screen across the surviving candidates. This is where most teams overspend: running a full search on every shortlisted name when a structured similarity pass would eliminate two-thirds of them in a fraction of the time. Stage three is a proper full-availability search on the one or two finalists, including common-law use, domain footprint, and social handles.

Stage four is the decision memo. It should not be a pile of search results; it should be a one-page risk readout that the business can sign off on, with named conflicts, likelihood of refusal, and the watch program that will follow filing. If your current process cannot produce that page, the rest of it is decoration.

Where AI-assisted screening helps and where it does not

Machine-assisted similarity scoring genuinely changes the economics of stages one and two. A modern screening engine can compare a candidate against tens of millions of marks across multiple registers in seconds, surface phonetic neighbours that a human reviewer would miss, and rank them by a similarity score the team can calibrate. For knockouts and shortlist triage, this is now the right default.

What it does not do — and where teams get into trouble — is replace the legal judgment in stage three. A high similarity score is a flag, not a conclusion. Whether a senior examiner or an opposing counsel would view two marks as confusingly similar still depends on goods overlap, market context, and the specific case law of the jurisdiction. Treating the score as the answer is how teams end up surprised in opposition proceedings.

The pragmatic split is to let the engine run wide and let people decide narrow. Use AI scoring to expand the candidate list you can realistically review, not to shrink the human review on the names that survive.

Operationalising it on a small team

Most in-house trademark teams are three to seven people supporting dozens of business units. The temptation is to build a heavyweight gating process that everyone routes around within a quarter. The version that survives is lighter than that and lives where the business already works.

Three things make it stick. First, a single intake form that captures candidate name, target markets, and intended goods in plain language — no class numbers required from the requester. Second, published service-level expectations for each stage, so a brand manager knows whether to expect a knockout in two days or two weeks. Third, a default-deny posture on candidates that fail stage one, with a documented escalation path if the business wants to override it.

The escalation path matters more than people think. Risk that is consciously accepted by a named decision-maker is a different beast from risk the legal team absorbs silently because they could not get the meeting.

What filing-stage risk looks like once pre-filing is solid

Teams with a disciplined pre-filing workflow tend to see two things shift at the filing stage. Refusal rates drop, often by a meaningful margin, because the marks that would have drawn objections were eliminated before they were filed. And the refusals that do come in are mostly procedural — class language, specimen quality, formalities — rather than substantive likelihood-of-confusion fights.

The other shift is in opposition exposure. When the cleared list is genuinely cleared, oppositions drop into the single digits as a share of filings, and the ones that remain are usually edge cases that no reasonable process would have caught. That is the right state to be in. You cannot drive opposition risk to zero, but you can drive it to a level where each one is a real surprise rather than a recurring tax on the program.