The trademark reality check your naming process needs

Every naming story has the same deleted chapter. The team brainstorms, falls in love with a name — weeks of it, the name on mockups, founders saying it to each other in the kitchen — and then the lawyer calls. "There's a SaaS logistics company in Ohio with a federal registration." Silence. Grief. A desperate search for a spelling with three consonants removed. Then the worst version: the second-choice name, chosen in a foul mood in one afternoon, that you ship because the launch date said so.

Naming isn't hard because creativity is scarce. It's hard because availability is scarce, and teams spend their emotional budget before checking their legal one. We've written the generative side before — the field guide from brainstorm to shortlist, the international linguistic checks. This piece is the reality check: the screening funnel that takes a thousand candidates down to the three that survive, in the order that wastes the least love.

The funnel shape: cheap screens first, love last

The core discipline is sequencing: run the cheap, brutal screens before the expensive, emotional ones. In practice, that means a funnel where a thousand names become three, and nobody mocks up a logo until stage four.

Stage 1 — generate wide (1000 → 200). Brief first: what the name must do, what registers it may live in (descriptive, evocative, invented), what it must not collide with. Then volume, because wide generation is what makes the later deaths painless. Rules that keep volume honest: no new name may be a variant of one already on the wall ("but with -ly!" is banned; those giveaways are on the domain market for a reason), and every name gets written once, not defended. Volume is coverage, not attachment.

Stage 2 — the ten-second screens (200 → 40). Google it in four flavours: the word alone, the word plus your industry, the word plus "app," the word plus "scandal." Check .com status with a registrar lookup — not by typing it into a browser, which leaks interest to domain watchers who will own it by Friday. Kill fast and without ceremony. The criterion isn't "is it free," it's "is its neighbourhood habitable" — a name whose search results are a lawsuit, a hate group, or a category giant fails even if the mark is technically available.

Stage 3 — trademark screening proper (40 → 8). Now the registers: your market's trademark database (IP Australia, USPTO, EUIPO, UKIPO — whichever jurisdictions you'll actually trade in), searched in the classes you operate in, not just "does the word appear somewhere." Trademarks are class-scoped: a name registered for agricultural machinery and one registered for project-management software can coexist, which is why "the name exists already" is not automatically fatal and "it's not in the database" is not automatically safe. What you're screening for at this stage is same-class or adjacent-class collisions, live applications, and — this is the part that kills darlings — phonetic similarity, because trademark conflict is judged on confusion, and "Konstructa" and "Constructr" confuse beautifully.

Stage 4 — the lawyer (8 → 3). Send your survivors to counsel with context: what classes you need, which markets, your risk tolerance. Lawyers need two to four weeks for a proper clearance opinion, and the timeline is your project's friend if you start it while the shortlist still has eight names on it. Teams who send one beloved name to the lawyer get a binary answer on a hostage situation; teams who send eight get a ranking.

Stage 5 — the living-in-it test (3 → 1). Availability is necessary, not sufficient. The final three get lived with for a week: said aloud on a fake sales call, typed into an email address, written on a mock invoice, tested in the voice the brand will actually speak in. Names that survive lawyers can still die here — one finalist of ours died the day someone asked "and how do you spell your email?" and the answer required nineteen seconds.

The domain compromise positions

Let go of the fantasy of the clean .com, unless the budget for fantasy is real — five figures minimum, often six, for a dictionary-adjacent word. The mature positions, ranked by how well they've aged for brands we've watched:

A distinctive .com with a verb or modifier: get-, use-, join-, try-, with-. Mildly compromised, universally legible, and customers find you via search anyway.
A prime country or sector TLD that means something to your audience: .com.au carries trust in Australia; .dev and .io aged well in software. Know the registry's rules — some ccTLDs have presence requirements.
The clean invented word on .com: the original reason invented names exist. If your shortlist is all real words fighting for taken domains, that's a brief problem, not a domain problem — go back to stage 1 and generate invented candidates with intention.
Buy later, deliberately: launch on the modifier domain, and treat acquiring the pure .com as a line item in a future raise, negotiated through a broker once the brand has value. Several huge brands did exactly this; nobody remembers.

What never works: awkward hyphenation, the pun spelling that must be dictated over the phone, or picking the name to fit a parked domain you happen to own. That's letting the registry write your brand.

App-store collisions: the wild west screen

App stores are trademark-free-for-alls with name squatting, and plenty of names that clear trademark screening die here. Search both major stores for your shortlist and every plausible misspelling. An existing app with your name in a different category is survivable; an existing app with your name in your category with 40,000 reviews is a rebrand-in-waiting. Also check the social handles now, across every platform including the ones you'll never post on, because handle squatting is free and regret is not. Add both to the stage 2 screens on future projects; the screening order that saves the most pain changes as the surfaces multiply.

Timelines: what lawyers and registrars actually need

Build the calendar backwards from launch and watch the naming panic evaporate:

Generation and ten-second screens: two weeks.
Trademark screening to shortlist of eight: one week.
Legal clearance opinion: two to four weeks (this is the long pole; start it while design hasn't committed to anything).
Living-in-it week + domain/handle acquisition: one to two weeks. Buy domains the day the legal opinion lands — waiting a fortnight to "sleep on it" is how founders end up negotiating with a squatter who registered it while they slept.

Total: six to ten weeks from brief to owned name. Compare with the cost of the alternative sequencing — falling in love in week one, lawyering in week eight, restarting in week ten — and the funnel pays for itself before it starts.

Letting a name die gracefully

The final skill in naming is grief management, because names will die late, and how the team handles a death determines the quality of its replacement. Practices that work:

No private favourites. Names live on the shared wall from day one; nobody gets to nurse a secret beloved. Attachment scales with secrecy.
Kill criteria, pre-agreed. The team decides in week one what kills a name (a same-class registration; unanswerable pronunciation; a search neighbourhood that's a swamp). When the lawyer calls, it's the criteria saying no, not anyone in the room. That distinction saves marriages between co-founders.
Always have two live alternates. We never let a shortlist drop below three until contracts are signed. The teams that mourn longest are the ones who treated the shortlist as a coronation.
Name the new favourite fast. The replacement in a foul mood is risk; the replacement as an open return to the surviving eight is process. The names that survive stages 2–4 are usually stronger than the dead darling anyway. You just hadn't said them in the kitchen yet.

A name is a piece of infrastructure — load-bearing, expensive to change, checked against reality before anyone admires the view. The creative work and the legal work aren't enemies; run right, the constraint of the funnel is what forces the thousand-name wide enough to find the one worth keeping.

Key takeaways

Sequence screens cheap-to-expensive: ten-second searches before trademark registers, registers before lawyers, lawyers before love.
Generate a thousand; ban variant-spelling candidates; let volume make deaths painless.
Trademark screening is class-scoped and phonetic — "not in the database" isn't safe and "the word exists" isn't automatically fatal.
Send eight names to counsel, not one beloved: lawyers give better answers to shortlists than to hostages.
Let go of the clean-.com fantasy; modifier domains and meaningful TLDs age better than hyphenated pun-spellings.
Check app stores and social handles even in different categories — squatting is free, regret isn't.
Pre-agree kill criteria and keep the shortlist at three until contracts sign. Grief management is name management.

FAQ

Can't we just launch and trademark later? You can file later; you can't check later. Launching unscreened means the cease-and-desist arrives after the signage, the domain authority, and the merch. File early in your core markets, absolutely screen before anything public, and remember that unregistered rights are a much weaker umbrella than a registration in your operating classes.

Is a descriptive name safer legally than an invented one? Safer from collisions, weaker as protection — descriptive marks are hard to register and harder to defend, and anyone in your category can say the same words. Invented and evocative names are the strongest marks if they clear, which is why the funnel runs wide enough to find clearable ones. Descriptive names also age into SEO fights you're competing in instead of owning.

The .com of our dead favourite name is parked and for sale. Buy it and revive the name? The parked domain was never the problem; the trademark conflict was. Buying the domain revives the name's only soluble problem while leaving the fatal one. Resist the resurrection. If the legal objection was actually about the domain rather than the mark, then yes — broker it quietly, through a third party, before anyone publicly commits to the name.

Our name is perfect but a tiny company in another country has the mark in our class. Options? Talk to counsel about coexistence agreements, scope carve-outs, or acquisition of the mark — small marks do change hands — and price that against renaming now, which is cheaper than it will ever be again. Do not take the "they're small, they'll never enforce" position: marks get bought by bigger companies, and the dispute inherits you.

When in the brand project should naming start? First, before any visual identity — the funnel needs its six to ten weeks and everything downstream hangs off the result. A rebrand whose name isn't legally clear by concept stage is a rollout sequence building to a cliff edge, and the launch-date pressure will write the rest of the story for you.