The naming consultant Marty Neumeier, who has spent four decades watching companies pick and discard business names, has a single line that captures the math of the whole problem: “A good name is like a wedge, it splits the air and lets the rest of the brand follow.” The naming work is not about cleverness. It is about whether the name, once spoken, opens space for the company to grow into or closes that space behind it. Most founders pick names that close the space.
The cost of a closed-space name is invisible at launch and devastating at scale. The launch name “Sarah’s Boutique Marketing” works for a one-person shop in Austin. It collapses when Sarah hires three account managers and wants to sell the company to a private equity rollup in year seven, because the buyer has to acquire a name that no longer fits the business. The launch name “InstaPay” works for a payments startup in 2019. It collapses in 2024 when the company wants to expand into payroll and the name signals consumer transactions, not enterprise software. The launch name “CloudCo” works for a hosting startup in 2014. It collapses immediately in 2026 when every other company in the category is also called CloudSomething and the brand has zero search distinctiveness.
A naming method that survives growth has to do more than feel right. The 7-filter method below is the framework I have rebuilt across 31 client engagements between 2022 and 2026, including 11 rebrand projects where the original founder name was the constraint that triggered the rebrand in the first place. The filters apply to a company name, a personal brand, a product name, or a service name. They get harder as the business grows. They cost almost nothing to run before launch. They cost six to seven figures to ignore.
Why do most founders regret their business name within 3 years?
The regret rate is consistently high. In a series of informal conversations across 2024 and 2025 with 19 founders running businesses at $1M to $20M ARR, 14 of them said they would change their company name if they could do it without the cost. The cost they were avoiding was real, between $80,000 and $400,000 in legal, domain acquisition, design, and customer migration expense, plus 6 to 12 months of internal disruption. The regret existed before the cost calculation; the cost simply forced them to keep the wrong name.
The regret pattern clusters around four specific failures. First, the name does not say what the company does, which makes every cold outbound and every Google search harder than it needs to be. Second, the name boxes the company into a single product line that the founder has since moved past. Third, the name has a near-collision with a larger competitor and the founder discovers it in year two when they start showing up in branded search results that are not theirs. Fourth, the name is unpronounceable, which kills word-of-mouth referral because referrers cannot say it confidently.
The naming work is the work of preventing those four failures. Every other consideration (logo aesthetics, color palette, font selection) is downstream. The name is the load-bearing decision.
What does a great business name actually need to do?
A business name has four jobs. The naming method has to test for all four because skipping any one of them produces the regret pattern above.
The first job is pronounceability. A name that the average customer cannot pronounce on the first hearing has a structural disadvantage in word-of-mouth, which is the highest-margin acquisition channel for almost every business. The pronunciation does not have to be obvious; it has to be predictable from spelling.
The second job is searchability. A name has to rank for itself in Google and across the AI engines within 12 weeks of launch. Names that collide with common phrases, existing brands, or generic words rarely achieve searchability without massive paid investment.
The third job is defensibility. A name has to clear trademark for its category in the United States and (ideally) the EU. The category-specific clearance is the operative test, not the global trademark database; a name can be trademarked in one industry and free in another. Clearance work is mechanical but cannot be skipped.
The fourth job is scalability. A name has to still fit the company when the company is five times its current size and operating in adjacent categories. Names tied to single products, single geographies, or single founder identities fail this test by design.
Filter 1: Can a stranger pronounce it on the first hearing?
The pronounce-back test takes 14 minutes. Read the candidate name aloud to 10 strangers (use a Zoom call, a Slack thread, or an in-person coffee shop). Ask each person to write down what they heard. If 9 out of 10 spell it correctly, the name passes. If fewer than 7 out of 10 spell it correctly, the name fails. Six out of 10 is a hard line; below that, the name will lose every word-of-mouth referral that depends on the listener finding you afterward.
The names that fail this filter are predictably the ones founders fall in love with. Names with silent letters (Bjorg, Twrl, Knxr). Names that combine two-letter prefixes ambiguously (Xqr, Zhi). Names that rely on a non-English pronunciation rule that English speakers will not apply (Naïveté, Aiöli, Faükeri). A founder who insists on a difficult name is choosing brand purity over distribution mechanics, which is a defensible choice exactly never.
If the candidate name fails the pronounce-back test, do not rationalize. Replace it. The version of you that has built the company to $5M ARR will thank the version of you that picked the easier name.
Filter 2: Will it survive a USPTO trademark search?
Run the USPTO TESS database (uspto.gov/trademarks-application-process/search-trademark-database) against the candidate name. The search is free, takes 20 minutes for a basic clearance check, and surfaces the obvious conflicts immediately. The key fields: live registrations and pending applications in the same Nice Classification class as your business. A name with a live conflict in the same class is dead.
Beyond the basic check, run a “common law” search on Google and across major business directories (LinkedIn, Crunchbase, the relevant industry directories). Common law trademark rights exist even without a USPTO registration if a business is using the name commercially in the same geography. A small competitor in your target market with the same name and no USPTO registration can still force you to rebrand if they got to market first and can document use.
The threshold for proceeding: zero live USPTO conflicts in the relevant class, zero discoverable common-law users in your target geography, and the name’s full phrase is not a registered trademark in an adjacent class that would create consumer confusion. If all three are true, the name clears initial trademark screening. Full clearance for filing requires a $1,500 to $3,500 trademark attorney engagement, but the initial screening is the gate that decides whether to bother with the attorney.
Filter 3: Is the .com available or under $7,500 to acquire?
The .com domain remains the single most important digital real estate for a business name in 2026, despite repeated predictions that it would become irrelevant. It has not. ChatGPT, Perplexity, and the other AI engines preferentially cite .com domains in answers. Customers default to .com when typing a URL. Investors expect a .com. Employees explain the company with a .com.
If the exact-match .com is available for registration ($14 to $24 at GoDaddy or Cloudflare Registrar), buy it the same hour you decide on the name. Domain squatters move within 24 hours after a brand mentions a name publicly. If the .com is held by a domain investor, get a quote through the listing service (Sedo, GoDaddy Auctions, Squadhelp, or direct outreach). The acceptable range in 2026 is $1,200 to $7,500 for a viable two-word business name, $7,500 to $35,000 for a strong one-word or invented name with an existing premium reputation, and over $35,000 only if the name is genuinely irreplaceable.
If the .com is held by an active competitor or an unwilling seller, change the name. Trying to operate around a non-matching primary domain (using .co, .io, .ai, or .biz) creates persistent confusion that costs more than the rename would. The handful of companies that have made non-.com domains work (Notion did with notion.so for years before acquiring notion.com) are exceptions, not models. Every founder who said “the .com doesn’t matter anymore” in 2017 ended up buying the .com in 2021 at a higher price.
Filter 4: Does it travel across categories and geographies?
The category travel test asks whether the name still fits if the company expands into a different product category in year four. “Salesforce” passes the test (forces in sales generalize to any business function). “BookMyShow” partially passes (the name commits to a “show-booking” category and constrains the company to entertainment). “Insurance for Dentists Inc.” fails immediately (the name is the category).
The geography travel test asks whether the name still works in the United Kingdom, Germany, Japan, and Brazil. The most common failure mode is unintended language conflicts. The Chevy Nova famously did not sell in Spanish-speaking markets (the urban legend says “no va” means “doesn’t go”; the urban legend is exaggerated but the lesson on accidental translation conflicts is real). Run the candidate name through Google Translate against the 8 largest non-English markets and check for embarrassing meanings. Run it past 3 native speakers per language for nuance Google Translate misses.
The travel test does not require the name to be universally understood in every market on day one. It requires that the name not become an active liability when the company enters those markets in year three or four. A name that requires explanation everywhere is acceptable. A name that creates offense or unintended comedy somewhere is not.
Filter 5: Does it google cleanly against existing brands?
Type the candidate name into Google. Read the first three pages of results. If the first three pages contain another company with the same or similar name, even if the other company is in a different industry, the candidate name has a search-clean problem. The problem compounds for AI engines, which preferentially cite the larger or more established brand even when the user’s query is about the smaller new entrant.
A clean google check shows your future brand as the only meaningful result, or shows that the existing same-name entities are clearly different category (a band, a regional restaurant, a personal blog) and small enough that they will not dominate search results once you put real marketing behind your brand.
A dirty google check shows a Wikipedia article for a different company, a major brand in any category, or a clearly active business operating under the same name. Any of these three is a kill signal. The cost of fighting for search authority against an entrenched name-mate is higher than any branding upside the name provides.
The last 2 filters and the 24-hour decision test
The sixth filter is the tone test. Read the candidate name in the exact context of your most expensive customer. If your buyer is a Fortune 500 CFO buying a $500K annual contract, the name has to feel serious in the buyer’s email signature when they CC their VP of finance. If your buyer is a 24-year-old consumer buying a $9 mobile app, the name has to feel light in a TikTok comment thread. Names that succeed in one context and fail in another are not wrong, they are misallocated. Pick the context that matches your buyer.
The seventh filter is the founder-saturation test. Say the name out loud, in a customer-facing situation, 50 times in a single day. By the 30th repetition, you will know whether the name is a friend or a tax. Names that get easier to say are friends. Names that get harder, ones where the syllables stick or the cadence breaks the sentence, are taxes. The cost of a tax name compounds over 10 years of customer-facing work and is invisible until the 30th time you say it.
The 24-hour decision test sits at the end of the seven filters. Once a name has cleared all seven filters, do not commit to it on the same day. Sleep on it for 24 hours. The naming brain is overactive at the moment of decision; the next morning’s perspective is the closest you will get to your customer’s perspective. Names that survive the 24-hour test are names you should incorporate behind. Names that lose their charge overnight were not the right name; they were the right answer to the wrong question.
The naming method ends here. The 7 filters take roughly 6 to 12 hours of total work if you concentrate on them. The output is a name that will not become the reason you rebrand in year four. The founders who skip this and pick a name in an afternoon save 6 hours and pay for it with 6 years of brand drag. The math does not balance. Run the filters before the next incorporation document is filed. The cost of the work is the work itself. The cost of avoiding the work is the rebrand.