Domain Yoga

The naming brief: what to fill in before you generate names

By Domain Yoga · Last updated July 31, 2026

A naming brief is a one-page document you fill in before a single name exists: eight fields that decide how names will be judged, plus two that set the terms of the engagement — clearance scope and who ends up owning the domain. All ten are below. If you’re naming your own project, you don’t need the document: compress it to one sentence and run the 1-hour naming sprint instead. The full brief is for the other job — naming something for a client — where the hard part isn’t writing the fields but filling them in, because the answers live in someone else’s head, half-formed. Clients often don’t fully know what they want, contradict themselves without noticing, and reject shortlists for reasons nobody ever stated. Every field below earns its place by naming the specific failure that arrives three weeks later if you leave it blank.

Why write a brief when the client will “know it when they see it”?

“We’ll know it when we see it” is the sentence that costs the most later, because it means the judging criteria will be invented after the candidates exist. And once candidates exist, everything changes: names attract emotion, and a conversation about criteria held in the presence of a name is really an argument about that name. The stakeholder who asks “shouldn’t it feel more premium?” while looking at your favorite candidate isn’t refining the brief — they’re killing a name and dressing it as strategy, and neither of you can tell the difference.

The brief’s real function is sequencing. It forces the argument about what good means to happen while it’s cheap — before anyone has generated hundreds of names, filtered them, grown attached, and built a deck. A founder naming their own thing can skip the document because briefer and briefed are the same person — not because they never disagree with themselves, but because nobody has to arbitrate it. You don’t have that luxury. For you, the brief is a small contract about how names will be judged — and later, the only thing standing between your shortlist and whoever is most confident in the room.

What breaks when a field is left blank?

  1. The job. One sentence about what the name must do: reassure enterprise buyers, stand out in an app store, survive being said on a podcast. Blank: the shortlist gets judged on cleverness, which nobody can argue with or against.

  2. The thing being named. Company, product, or one name stretched across both — name the product or the company is the decision to walk through together. Blank: scope drift; round two opens with “could it also cover the other product?”

  3. Audience and languages. Who says the name aloud, and in which markets — naming for a multilingual EU market covers what goes wrong across languages. Blank: you check the languages you happen to speak, and the objection arrives from the market nobody in the room represents.

  4. Style range. Where between descriptive and coined the client is willing to live — a range, not a point; brandable vs. descriptive maps the spectrum. Blank: half your pool is dead on arrival. “We’d never use a made-up word” is a kickoff sentence, not a reveal-meeting sentence.

  5. Tone anchors. Three existing names the client wishes were theirs and three they hate — actual names, not adjectives, because adjectives don’t discriminate: nearly every client wants “modern, trustworthy, a bit playful.” Names are falsifiable; adjectives agree with everything. Blank: you discover in week four that “modern” meant a different decade to each stakeholder.

  6. Exclusions. Banned words, competitor echoes, the previous name, the founder’s dead venture, sounds-like traps. This field only fills through provocation — see the next section. Blank: you proudly present a name that rhymes with the rival the client despises.

  7. Domain and budget constraints. Must it be the exact .com, or is the client honestly open to another TLD? Is there budget to buy a taken domain, or must every candidate be free to hand-register? Blank: you fall in love, jointly, with names the client cannot afford to own.

  8. The decision process. Who’s in the room, how many rounds, and how the final call is made — against the brief, ideally with a structured head-to-head and a real-world test for the finalists rather than a show of hands. Blank: consensus by attrition, where the blandest name wins because nobody hated it enough to fight.

  9. Clearance scope. Exactly which checks your fee includes — domain and handle availability, a knockout search for identical marks in the client’s category (here’s what an EU screen looks like) — and the explicit line that a screen is not legal advice: final clearance belongs with the client’s counsel. Blank: the client believes you cleared the trademark; you believed they’d hire a lawyer; nobody did. This is general information, not legal advice — what a screen can and can’t establish is a question for a trademark attorney. The first eight fields are judging criteria — a name can satisfy or violate each one. The last two are terms of engagement: they govern the work, not the names, and the shortlist test at the end of this article does not apply to them.

  10. Ownership and handover. Who registers the winning domain, in whose registrar account it lives, who pays renewals. The answer should be the client, with you holding delegated access — the ownership question is a contract clause, not an afterthought. Blank: the domain stays in your account by default and resurfaces as an offboarding dispute years later.

What does the finished brief look like?

Copy this into a doc and fill it in during the kickoff. It holds no names — every line is a requirement, and the names come afterwards, judged against what you wrote here.

NAMING BRIEF — [client] — [date]

JUDGING CRITERIA
1. The job          The name must ……………………………………
2. Naming           [ ] company  [ ] product  [ ] both  →  ………………
3. Audience         Says it aloud: ………………  Markets/languages: ………………
4. Style range      Descriptive |—————————————| Coined
                    Acceptable band: from ……………… to ………………
5. Tone anchors     Wishes it were theirs: ………, ………, ………
                    Actively dislikes:      ………, ………, ………
6. Exclusions       Banned words/echoes: ………………………………………
7. Constraints      Exact .com required?  [ ] yes  [ ] no
                    Budget to buy a taken domain?  [ ] yes  £/$………  [ ] no
                    Forced trades — .io only: [ ] alive [ ] dead
                                    misspellable: [ ] alive [ ] dead
8. Decision         In the room: ………………………………………
                    Also sees it (asked twice): ………………………………
                    Rounds: ……   Final call by: ………………

TERMS OF ENGAGEMENT — govern the work, not the names
9. Clearance        Included: [ ] domain  [ ] handles  [ ] knockout mark search
                    Not included: full legal clearance (client's counsel)
10. Ownership       Registrant: ………………  Account: ………………  Renewals: ………………

Signed ………………………………  Date ………………
Changes to this brief are amendments, and get a round and a date.

Two things about the form itself. The forced trades in field 7 are there because a client who has not chosen between them has not finished the brief, and you want that discovered now rather than in round three. And the “asked twice” prompt in field 8 is not a typo — it is the reminder to ask who else sees the shortlist a second time, later in the conversation, which is when the real answer tends to arrive.

How do you extract answers the client doesn’t know they have?

You can’t get this brief by emailing the questions. Most of the fields are preferences the client has never articulated, and direct questions produce polite noise. So extract instead.

Show names; don’t request adjectives. Open the kickoff with five or six sacrificial names spread deliberately across the style range — among them one plainly descriptive, one coined, one metaphor, one clearly too far. Say out loud that they’re probes, not candidates. Watching a client flinch fills in the style range and the exclusions field, because “I hate that” is data and “we want something bold” is not.

Ask for hates before likes. Dislike is easier to state precisely than liking: “not that” points at something specific, while “something bold” commits to nothing. The hate list converts almost directly into exclusions; the wish list becomes your tone anchors.

Ask the rejection question. “Describe a name you’d have loved and still turned down — what killed it?” Whatever answers that is a criterion that exists and was never going to be volunteered. Write it into the brief verbatim.

Ask who else sees the shortlist — then ask again later. The first answer names the room. The second, asked casually, names the veto: a co-founder, a chair, a spouse. The unnamed approver is the failure mode the brief exists to catch, because no naming technique works around a person who was never in the room. Every person mentioned goes into the decision-process field, along with the round they enter.

Read it back before anyone signs. Reading the brief aloud is the last cheap opportunity for “well, actually…” to surface. Every objection during the read-back is a rejection you didn’t have to build a shortlist to trigger.

What do you do when the answers contradict each other?

They will contradict each other. The standard bundle: short, a real word, obviously meaningful, unlike anything else in the category, exact-match .com, at registration price. Those requirements practically never hold together, and the brief is where they collide safely — as sentences on a page rather than as serial rejections of real names.

Don’t average the contradiction away; make the client choose which constraint bends. Hypothetical forced trades work better than abstract ranking: “A name you love exists, but only the .io is available — alive or dead?” “Perfect meaning, but it’s a coined word people misspell on first hearing — alive or dead?” Record each verdict in the constraints field, because “all of the above” is not a brief. And if the client refuses to choose, that’s a finding too: it usually means the real decision-maker hasn’t been in the room yet, and your decision-process field is wrong.

How does the brief defend the shortlist?

Present every shortlisted name annotated against the fields it satisfies — the job, the style range, the tone anchors, the constraints, its clearance status. Then, when a rejection arrives (one will), ask a single question: which field does this name break?

Three things can come back. It breaks a field — your filtering slipped, the kill is legitimate, say thank you. It breaks no field, so a criterion has surfaced that was never stated; that is normal, and it gets written down. Or, most often, you disagree about what a field means: the brief said the style range reaches “coined”, and this name is apparently too coined. That third answer is the one to expect, and the brief does not settle it — it relocates the argument from the name to the word, which is a much smaller argument and one you can have without anyone defending a favourite.

The second and third cases are where the brief earns its fee, and the reason is pricing rather than bureaucracy. Unstated criteria are free to hold and expensive to serve: every one that surfaces mid-project silently converts into rework nobody scoped. Written down, it becomes a visible amendment — a new round, a new date, a line on an invoice, or an explicit decision to absorb it. You are not stopping the client from changing their mind, which you cannot do and should not want to. You are making the change legible enough to price, so that round three is a conversation about scope instead of an argument about whether you understood the assignment.

A signed brief doesn’t guarantee the client will love your shortlist. It guarantees that “I just don’t like it” resolves into either a violated field or a new field — and you can work with both, where you can’t work with a mood.

Once the brief is signed, generation is genuinely the easy part: the document you just extracted is most of the prompt, and how to prompt AI for name ideas covers turning fields into instructions. On tooling, client work has a particular shape: costs want to attach to a project. Domain Yoga charges per search — $2–$5 each, no subscription, and credits never expire — so a search is a line item you can bill through to the engagement it belongs to, which a seat licence never is. Each search takes a description of the project and returns around 250 availability-checked ideas ranked by brandability — run one per direction in the brief, and you walk into the presentation with names that were available when you checked them, not names you’ll have to negotiate for. Re-check the finalists the morning of the meeting; a multi-round project gives the market weeks to take one.