Brand Naming · Mistakes
The seven brand naming mistakes founders make.
I built Etymolt after one of these mistakes cost me $40,217 and a quarter of momentum. The full story is in the founder narrative — week 16 of a brand launch, a USPTO Office Action under §2(d), and the slow realization that the same naming process I had used was the naming process almost every founder uses, and that the failure modes were not random.
The seven mistakes below are the ones I see in our inbox, in our corpus, and in the week-16 envelopes that keep arriving. None of them are exotic. All of them are preventable in three seconds with the right tool. The point of writing them down is to shorten the path from I had no idea this was a failure mode to I caught it before I shipped.
Mistake 1 — The descriptive trap
The founder picks a name that describes the product. Coldbrew Coffee. Quick Pay. Cloud Storage. The USPTO refuses registration under §2(e)(1) — descriptive marks. Search returns category competitors instead of the founder's brand. The name does none of the three jobs a name should do.
In our corpus, 23% of first-draft names fail §2(e)(1) at >60% confidence. The descriptive trap is the most common failure mode and the most preventable. See the Coldbrew case study for a worked example — verdict ABANDON at 28/100, driven entirely by descriptiveness and a crowded register.
Mistake 2 — Period-marking
The founder picks a name that reads as 2010 (or 2014, or 2018) instead of 2026. Foo-ly. Bar.io. Baz.ai. The name period-marks the company as belonging to a previous SaaS register, and the buyer cohort filters it out as off-cohort before reading the tagline.
The fix is to read the current canon, identified in SaaS Brand Naming Patterns, and to know what the 2026 register sounds like. Latinate two-syllable (Linear, Vercel, Modal). Monosyllable found-object (Stripe, Bolt, Ramp). Compound with concrete imagery (Snowflake, Tinybird). Period-marking happens when the founder names against the canon they learned five years ago. The Lexicon and Catchword frameworks, unpacked in The Best Brand Naming Frameworks, give the vocabulary for diagnosing the mismatch.
Mistake 3 — Cohort-mismatch
The founder picks a name that sounds like a different category. A B2B SaaS named like a D2C consumer brand. A fintech named like a children's toy. A developer tool named like a wellness app. The cohort calibrates the phoneme palette, and a mis-cohort name forces the buyer to do translation work at every first encounter.
I have watched a developer-tool founder ship under a name that sounded like a cosmetics brand. The product was good; the cohort signal was wrong. Every Hacker News thread about the company spent half its energy on the name and half on the product. The founder eventually renamed in year three, after the cost of the mis-cohort had compounded to a six-figure cost in lost first-touches.
Mistake 4 — Sound-symbolism backfire
The founder picks a name whose phonemes encode the wrong feeling. A luxury-goods brand built on hard plosives reads as cheap. A fintech built on soft labials reads as untrustworthy. A children's product built on hard sibilants reads as menacing. The phonosemantic axis is real — the Sapir / Köhler / Maurer / Ćwiek research line, summarized in Sound Symbolism for Brand Naming, has demonstrated this cross-culturally.
The fix is to score the candidate on the phonosemantic axis before shipping. Our scoring engine is calibrated against the Kawahara & Shinohara corpus and returns a per-axis feeling vector (sharp/round, fast/slow, light/heavy, masculine/feminine, luxury/cheap). A mismatch between the vector and the brief is a fixable problem if caught early; it's an expensive problem if caught after launch.
Mistake 5 — The.com obsession
The founder makes.com availability the dominant constraint. The shortlist gets filtered to candidates whose.com is free at registrar prices, and the best candidate is dropped because someone holds the.com on aftermarket for $15K. The founder ships under a worse name to save $15K on year-one balance sheet.
The math is wrong. A worse name costs five-to-twenty times the aftermarket.com over the company's life — in lost first-touches, in cohort-signal drag, in second-rebrand costs if the worse name eventually has to be replaced. Stripe paid six figures for stripe.com because the name fit the brief that exactly. The cost was a line item. The value of the right name is unbounded.
The corollary mistake: insisting on.com when a.ai,.so, or.dev would do the cohort signaling better in your category. Developer-tool founders in 2025-2026 face a buyer who treats.dev and.ai as native. The.com obsession is sometimes a holdover from 2014 thinking.
Mistake 6 — Handle blindness
The founder picks a name without checking that @brand is available on the canonical 12 social platforms. They ship under @brandhq on X, @brand_official on Instagram, @use-brand on GitHub. The dilution is permanent. Every social touchpoint costs a fraction of a brand impression because the handle does not match the brand.
I made this mistake. In the rebrand after the Office Action, I ran the trademark check carefully and skipped the handle check. The.com cleared. The handles didn't. We shipped under suffixed handles on every platform and accepted the dilution forever. The cost is not infinite — but it is permanent, and the lift to prevent it pre-launch is twenty seconds.
Mistake 7 — Late verification ("I'll just check later")
The founder picks a name, ships the domain, sets up the social handles, writes the landing page, prints stickers, and then runs the trademark check. The trademark check returns ABANDON. The sunk-cost fallacy kicks in. The founder spends three weeks trying to make the failing name work — petitioning the senior mark holder, filing a coexistence agreement attempt, paying an attorney to write a non-confusion brief. The work fails. The founder rebrands anyway, but now with three weeks of additional emotional and financial drag.
This is the mistake I made. The full reconstruction is in the founder story. The week-16 envelope arrived on a Tuesday in November 2025. The $40,217 legal bill that followed was almost entirely the cost of having verified late. Three seconds of verifier time would have caught the conflict in week one.
The fix is the order of operations: verify first, commit second. Run the candidate through the verifier before you register the domain, before you buy stickers, before you set up the social handles, before the name leaves the brainstorm room. The cost of verification at week one is zero. The cost of verification at week sixteen is everything.
The system fix
All seven mistakes resolve to one system fix: verify every candidate across the five axes before commit. Trademark, domain, cultural, sound symbolism, pronunciation. Three seconds. Five free per IP. Every flag traces to a record number. The verdict gets a permalink that stays with you forever. If the verdict comes back PROCEED, the founder has evidence to show an attorney and an investor. If the verdict is STRATEGIC, the founder has a punch list. If the verdict is ABANDON, the founder finds out before paying for the domain.
That's the product. I built it because I needed it in November 2025 and didn't have it. The methodology is at /methodology and public. The legal posture is the Bureau Model: we surface signals; we are not a law firm.
The second-order mistakes
Beyond the seven primary mistakes, three second-order errors recur in our inbox.
Naming by committee. The founder collects feedback from eight people, each with a different cohort and different aesthetic priors. The shortlist gets diluted toward the median — a name nobody hates and nobody loves. The median name is the worst possible outcome; the right answer is a name a specific cohort loves and an out-of-cohort observer might find odd.
Verifying the wrong jurisdiction. The founder runs the U.S. trademark check, the candidate clears, and the founder ships into a European market without running the EU register. The U.S. and EU registers are independent — a U.S. clearance has no weight against an EUIPO senior mark. The fix is to specify launch jurisdictions upfront and verify against every register the candidate will operate in within the first eighteen months.
Trusting the "available" signal alone. The founder runs a tool that returns "available" and ships. The tool checked exact-match only. The senior mark that kills the brand four months later is at 88% phonetic similarity — close enough for the USPTO examiner to issue a §2(d) refusal, far enough that exact-match search missed it. The fix is to verify against phonetic-similarity scoring and §2(d) collision probability, which is what the trademark axis of our methodology operationalizes.
Related reading
- How to Come Up With a Brand Name — the 7-step process that prevents most of these mistakes.
- What Makes a Good Brand Name — the rubric every candidate has to pass.
- Coldbrew (ABANDON, 28/100) — the descriptive-trap mistake at full scale.
Disclaimer
Etymolt operates under the Bureau Model. We surface clearance signals across the five axes; we do not provide legal advice. A trademark opinion belongs to a licensed attorney in your filing jurisdictions.
Take the next step
Verify a candidate before you commit.
Five free per IP. No signup. Every flag traces to a record number. Three seconds. The week-16 envelope is preventable.
Verify a name