The 1-hour naming sprint

Sixty minutes is enough to go from a blank page to two or three names that are available, say-able, and ready to test on real people. Not because naming is trivial, but because the two things that actually produce good names — volume and ruthless filtering — compress beautifully under a timer, while the thing that eats naming weeks, open-ended deliberation, adds almost nothing. Most founders don’t spend a month generating names; they spend a month circling the same six ideas. The sprint replaces circling with throughput: five timed phases, each with a hard exit condition, and no reopening a phase once it’s closed. This is the compressed cousin of the full process in how to name a startup — same passes, a fraction of the hours.

Before you start: set an actual timer, work alone or with one partner at most, and agree that the clock outranks your feelings. When a phase ends, it ends.

Minutes 0–5: write the one-sentence brief

Write a single sentence describing what the name must do — not what you’d like it to be. “Signal speed to developers.” “Feel calm enough for a finance app.” “Work in English, German, and French.” One job, maybe two. Resist writing a paragraph; a brief with five requirements filters out every name on Earth.

This sentence is the referee for the next fifty-five minutes. Every time you’re tempted to keep a name because it’s clever, you’ll check it against the sentence instead of your mood.

Minutes 5–25: generate wide, judge nothing

Your target is forty or more candidates, and the only rule is that evaluation is banned — write everything down, including the embarrassing ones, because bad names are stepping stones to adjacent good ones. Cover four styles in roughly five minutes each so you get a varied pool instead of forty flavors of one idea:

• Coined words. Invent from fragments and roots — smash syllables together until something sounds like a company.
• Real-word combinations. Collide two ordinary words and look for pairs with friction or surprise in them, not category descriptions.
• Metaphors. Name the feeling or image the product evokes rather than the feature it ships.
• Borrowed roots. Latin, Greek, mythology, loanwords — distinctive to your audience, with a story you can tell.

This is the phase where an AI generator earns its keep: a well-briefed one does the four-style fan-out in seconds and hands you a hundred candidates before your coffee cools, which converts your twenty minutes from producing names to curating them — a much better use of a human. The briefing matters, though; how to prompt AI for name ideas covers how to feed it your one-sentence brief instead of “name my startup.” Either way, exit this phase with a big, unjudged list.

Minutes 25–40: filter on ownability and say-ability

Now judgment switches on, and it should be merciless. Two filters kill most of the pool, and that’s the pool doing its job:

• Can you own it? Check the whole list at once — never one name at a time, which is where naming evenings go to die. Paste the pool into a bulk checker, or skip the step entirely by generating with a tool that only shows available names; checking domain availability in bulk covers both workflows. Add a thirty-second search for obvious trademark conflicts in your own category — the full screen comes later, but a name identical to a known player in your space dies now.
• Can you say it? Read each survivor aloud in the sentence “I work at ___,” then imagine spelling it over the phone. Hesitate on either, and the name pays a tax every day you own it. Cut it.

Expect brutal attrition — a forty-name pool shrinking to six is a healthy outcome, not a failed brainstorm. Exit with five to eight names that are both available and comfortable in your mouth.

Minutes 40–55: pressure-test the survivors

You can’t run a proper naming test in fifteen minutes, but you can run a triage that catches the worst mistakes:

• Type the domain on your phone. Autocorrect fights, awkward letter runs, and double-letter boundaries reveal themselves immediately.
• Mock the headline. Write “Name — one-line pitch” for each survivor as if it were your landing page hero. Some names collapse next to a real sentence.
• Send it to one person. Message your top three to a single honest friend and ask them to say each aloud and guess what it is. One reply is thin data, but it’s data from outside your head.

Rank what’s left and keep two or three. The deeper protocol — recall tests, smoke tests, weighing conflicting signals — lives in how to test a name before committing, and your finalists are now exactly what that process needs as input.

Minutes 55–60: decide or park

Two exits, both wins. Decide: if one finalist clearly beats the brief, register it now — availability is a snapshot, not a reservation. Park: if two or three are genuinely close, write them down, schedule the real testing week, and stop. What you may not do is reopen generation “for a few more ideas” — that’s the deliberation loop the sprint exists to kill. And if nothing survived to this phase, the pool wasn’t the problem; the brief was. Rewrite the sentence and run the sprint again tomorrow. Second runs are faster.

If you want the twenty-minute generation phase and the availability filter collapsed into one step, Domain Yoga takes a description of what you’re building and returns around 250 availability-checked, brandability-ranked ideas across many TLDs — so your hour starts at the curation phase, with names you can actually have.