The brand name is the single most permanent decision in your brand identity. You can change a logo. You can refresh colors. You can rewrite your voice. Changing the brand name is essentially starting over, and most founders only get one shot at picking the right one.
Yet most founders pick their brand name in one of two ways: they spend six weeks in a naming workshop with an agency, producing a 400-page deck and then choosing the third option they'd thought of in the first hour; or they panic-pick at midnight the day before they need to register a domain. There's a middle path. This is a 4-step process that produces a name in 2 to 4 days, with confidence.
Step 1: Define the constraints (Day 1, 30 minutes)
Before you generate names, write down the constraints they need to satisfy. This is the single most skipped step and the one that saves the most time. Without it, you'll generate 200 names and reject 195 of them on instinct without knowing why.
The constraints to write down:
- Domain rule. Do you need an exact-match .com? Can you live with .co, .io, .ai, .so? Can you live with an extension or modifier ("getVellem.com", "tryVellem.com")? Each constraint here cuts your candidate pool by roughly 80%, so loosening it expands options significantly.
- Length. Are you OK with three-syllable names? Two? Hard cap on letters? Names of 5-8 letters tend to be most memorable.
- Spelling difficulty. Can a stranger spell it after hearing it once over the phone? This matters more than most founders think. "Tumblr" was a marketing problem for years.
- Avoid words. What words shouldn't appear in your brand name? (e.g., if you're in health, avoid "med" prefixes because they're saturated; if you're not actually using AI, avoid "AI" suffix.)
- Tone. Three adjectives the name should feel: e.g., "warm, simple, smart" or "technical, precise, fast" or "playful, light, distinct."
Spend exactly 30 minutes on this step. Write it on one page. You'll refer back to it constantly in the next steps.
Step 2: Generate 50 candidates across 5 categories (Day 1, 2 hours)
The mistake here is generating 50 names in one style. You want diversity to surface what actually fits, which means generating across categories. Aim for 10 names in each of these five buckets:
Category A: Real words used unexpectedly. An existing English word that's not literally describing what you do. Apple. Mint. Slack. Notion. Linear. These work because they're memorable and the meaning attaches to your brand, not the original word.
Category B: Made-up words that sound like English. Words that don't exist but feel like they could: Verbio, Klera, Tendo, Vellem. The benefit: usually trademark-available, usually domain-available, semantically empty so you can fill it with brand meaning.
Category C: Compound words. Two English words mashed together: Lyft, Snapchat, Squarespace, Stripe (technically not a compound, but in that genre). Easy to generate, easy to remember, often pronounceable on first try.
Category D: Latin or foreign-language roots. A short word from another language whose meaning is relevant. Vellem (Latin for unfulfilled wish). Lyft. Veritas. Anthropic. The advantage: usually unique, often available, gives the brand depth that you can reveal as a story.
Category E: Acronyms or initialisms. Usually a bad choice but occasionally a great one. IBM. NASA. Don't pick this unless your acronym actually sounds like a word, otherwise it becomes a string of letters nobody remembers.
Generate 10 in each category without filtering. The point is not to find your name in this step. It's to populate the search space.
Step 3: Vet the top 10 against four tests (Day 2, 2 hours)
Pick your top 10 by gut. Don't think too hard yet. You're just narrowing the search. Then run each of those 10 through these four tests:
Test 1: The pronunciation test. Tell three people the name verbally and ask them to spell it. If two out of three can't, eliminate. (Note: this is harsh and you'll lose some names you love. That's correct. If customers can't spell it, they can't find it later.)
Test 2: The domain check. Look up the exact match on standard TLDs. If the .com is taken by a large company, eliminate. If it's taken by a parked domain or expired site, it's negotiable but usually expensive ($500 to $50,000). Cheap modifier domains (try-, get-, use-) are fine but not ideal.
Test 3: The trademark check. Search the USPTO database (free at uspto.gov) for your candidate names. Look in your category specifically. Same name in a totally different industry is usually fine; same name in a related industry is a red flag. If you find a direct match, eliminate. We have a separate guide on the trademark process if you want to go deeper here.
Test 4: The social handle check. Quick search for the exact handle on the two platforms you'll definitely use. Acceptable variants: adding "official", "co", "hq". Unacceptable: numbers, underscores, or anything that makes the handle hard to remember.
After this round, you typically have 3 to 5 names that survived. Sit with them overnight.
Step 4: Pick by gut, justify by constraint (Day 3, 1 hour)
This is the step founders most often get wrong. They turn the final decision into a committee process, a customer poll, or an analysis paralysis spreadsheet. The decision is a gut call and the gut should drive it.
Look at your 3 to 5 finalists. Read each one out loud. Imagine someone telling a friend "I just signed up for [name]." Imagine your support email coming from "[name]." Imagine the URL printed on a business card.
Pick the one that feels right. Then. And only then. Write down why it fits your constraints from Step 1. If you can't justify it against the constraints you wrote, your gut might be wrong; reconsider. But if your gut pick also satisfies the constraints, you have your answer.
What to do if nothing fits
If after Step 3 you have zero finalists, you have one of two problems: your constraints are too tight, or your generation in Step 2 was too narrow. Either loosen the domain rule (accept .ai, .co, or "try-" prefix) or generate another 50 candidates across the same five categories. Don't compromise on Steps 3's tests. The tests catch real problems.
And don't keep generating forever. Two cycles through this process is usually enough. If you're still nameless after that, you're not actually trying to find a name; you're avoiding the decision.
The brand name is the most permanent decision you'll make. It's also a decision you can over-research forever. Two to four days, a clear process, a gut pick at the end. That's the sweet spot. Spend a week if you must. Don't spend a quarter.
Your brand kit, ready in 10 minutes.
Five quick taps. Free preview before you pay.
Start building free →