How to Come Up With a Business Name (a Process That Actually Works)
A practical process for naming a startup or business: seven naming patterns with real examples, how to generate a real shortlist, and the screening step most founders skip.
The way to come up with a business name is to generate a wide list using proven naming patterns, then screen every candidate for trademark conflicts, domain availability, and existing businesses — because a great name you can't own is a bad name.
Most naming advice stops at the first half. Brainstorm, be creative, sleep on it. Then the founder falls in love with one name, buys a logo, and discovers at incorporation — or worse, in a cease-and-desist — that the name was never available. So this guide covers the whole process: how to generate names that are genuinely good, and how to find out fast which ones you can actually have.
Part 1: Generate wide
Naming is a numbers game. Professional naming agencies generate hundreds of candidates to find one; you should aim for at least twenty before judging any of them. The fastest way to get to twenty is to work the patterns deliberately instead of waiting for inspiration.
The seven naming patterns
1. Real words, borrowed. An existing word lifted into a new context: Stripe, Slack, Amazon, Apple. These are memorable and pronounceable by default — and the hardest to own, because dictionary words come with existing trademarks, taken domains, and crowded search results. The trick is distance: the word should have nothing to do with your category (a payments company called Stripe) so the association becomes uniquely yours.
2. Compounds. Two words fused: Facebook, Salesforce, DoorDash, BrandScreen. The workhorse pattern for a reason — compounds can describe what you do while still being distinctive enough to trademark and rank for. Generate these mechanically: list ten words about what you do, ten about the outcome you deliver, and combine.
3. Blends. Two words melted together rather than stacked: Pinterest (pin + interest), Instacart, Microsoft. More distinctive than compounds, slightly riskier on pronunciation. Test every blend out loud on someone who hasn't seen it written.
4. Truncations and bent spellings. A real word clipped or deliberately misspelled: Tumblr, Flickr, Lyft, or number-substitutions like Plan0. You gain ownability — the modified spelling is usually clear where the real word isn't — at the cost of a lifetime of spelling it out loud. Worth it when the base word is strong and everything else about it is taken.
5. Coined words. Invented from scratch: Kodak, Xerox, Häagen-Dazs, Notchkin. Maximum ownability — you'll rarely find conflicts on a word that didn't exist — and maximum marketing burden, since the name arrives meaning nothing and you pay to fill it. Best when you're playing a long game and want a name that can outgrow any single product.
6. Foreign and classical words. Borrowed from another language or mythology: Nike, Volvo ("I roll" in Latin), Samsara. They carry built-in meaning and sound distinctive in English — but check what they mean everywhere you'll operate, and expect surprising trademark crowding around popular mythological names.
7. Names and places. Founders, places, or invented personas: Tesla, Patagonia, Warby Parker. Rich in story, and often more ownable than they look — but a real person's name comes with a real person's search results.
Generating the list
Work each pattern for ten minutes. For your core concept, list the obvious words, the outcome words, the feeling words, and the adjacent-metaphor words — then run them through compounds, blends, truncations. If you want the mechanical part done for you, an AI generator like BrandScreen's name generator will produce thematically coherent candidates that are pre-screened before you ever see them — which quietly solves the problem the rest of this article is about.
Don't judge while generating. A "bad" name on the list often unlocks a good one two lines later.
Part 2: The filter most founders skip
Here's the uncomfortable math: on a typical shortlist of twenty plausible names, most will have a problem you can't see from a brainstorm. A confusingly similar live trademark. A same-name company registered in Delaware. A .com that's parked at a four-figure asking price. A search results page owned by someone else.
Finding this out per-name, manually, means a USPTO search (including phonetic variants, not just exact matches), Secretary of State entity searches, domain lookups, handle checks, and a hard look at what ranks. That's an afternoon per name — which is exactly why nobody does it, and why so many companies rename in year one.
So invert the order: screen before you fall in love. Run every serious candidate through a check before the team debate, so you're only debating names you can have. A BrandScreen report does the full pass in about a minute per name and returns a single 0–100 Brand Score — trademarks, entities, domains, handles, and search competition in one number, with the evidence itemized.
Part 3: Decide on evidence
With scores in hand, the decision gets rational:
- Cut everything with serious conflicts. A live similar trademark in your category is not a coin flip; it's a rename waiting to happen.
- Shortlist the top three ownable names. Not the top three favorites — the top three you can own.
- Test out loud. Say each over a bad phone connection. Ask someone to spell it after hearing it once. Imagine answering "what's the company called?" at a dinner party without wincing.
- Check the mundane stuff. Initials, unfortunate substrings, meanings in your customers' languages.
- Then commit fast. Register the entity, buy the domain, claim the handles, and consider a trademark filing early — a name that screened clean today is only guaranteed clean today.
The whole process, in one line
Generate twenty names across the seven patterns, screen all of them before debating any of them, and pick from the names the evidence says you can own.
The founders who skip the middle step aren't lazier — they just don't find out until the cease-and-desist arrives. Checking first is cheaper.
For the full step-by-step naming process — from setting criteria to pressure-testing finalists — see our how to name a startup playbook.
Generate pre-screened names → or check a name you already love →