What Makes a Brand Name Work

A brand name has one job before it has any others: survive the trip from one person’s mouth to another person’s memory. Everything else, the meaning, the story, the logo, gets built on top later. If the name cannot make that trip, no amount of design rescues it.

That trip breaks down into three mechanical tests, and then a set of quieter factors that decide how the name feels. Here is the whole thing, with the numbers behind it.

The scorecard: five tests, four must pass

The practitioner consensus across naming checklists is that a name should pass at least four of these five. And one of them is not optional.

Test	How to run it	Pass bar
Pronounceable on first read	Show the written name to five people, ask them to say it	Four of five say it the same way
Spellable when heard	Say it aloud, ask them to type the URL	If three of five misspell it, reconsider
Memorable	Say it, change the subject, ask again 20 minutes later	Recalled correctly
Trademarkable in your class	Search USPTO, EUIPO, or your national registry	Clear in the classes you actually sell in
Domain available	Registrar check	Exact match, .com if you can get it

The hard one is trademark. A name that is perfect on every dimension except trademarkability is useless, because you will either be sued or forced to rebrand at exactly the moment the name starts being worth something.

The radio test, in detail

The spellability test is the one most people skip, and it is the one that costs the most. Say the name out loud, once, at normal speed, with no context. Then ask the listener to write down the web address.

Watch for these failure modes:

• Doubled versus single consonants. Is it Mirah or Mirrah? Both look plausible, so both will be typed.
• C versus K versus CK. Micara, Mikara, Mickara.
• PH versus F. If your name contains an /f/ sound, pick one spelling and expect the other to be typed anyway.
• S versus Z. Verizon versus Verison. Verizon can afford to own both. You probably cannot.
• Endings in -y, -ie, or -i. Three spellings, one sound.
• Vowel digraphs at the seam. If your name is a blend and the splice point sits inside an ea, ie, ei, or ou, readers will guess wrong.

A practical rule: if you can think of more than two plausible alternate spellings, the name is hard to spell. Consistent phonetic swaps survive this (Lyft for lift, Flickr for flicker) because the substitution is systematic and the reader can reconstruct it. Arbitrary respellings do not.

Length: what the data actually says

Short is not a style preference. It shows up in every dataset anyone has counted.

Source	Finding
Atom.com domain data	Average domain length 13 to 14 characters; the top 10,000 sites average 8; the top 100 average 6
An analysis of 400+ best-selling brands by nuname	Mean 10 characters, median 9
Syllable counts across top brands	About 29 percent are exactly two syllables; on Amazon’s physical-product charts the two-syllable share passes 40 percent
Memory research summaries	Names of 6 to 8 characters perform best in recall tests
Practitioner rule of thumb	7 letters or fewer, 2 syllables or fewer

The last row explains itself in the field. Customers shorten names whether you permit it or not. Chevrolet became Chevy. Coca-Cola became Coke. Federal Express became FedEx, and eventually the company gave in and made the short form official.

A working target: 5 to 10 characters, 2 to 3 syllables. Treat 12 characters or 4 syllables as the point where you need a reason. There is a happy accident here worth knowing: real linguistic blends gravitate toward 2 to 3 syllables all by themselves, so the prosodic rules and the branding rules agree. A well-formed portmanteau is usually already the right length.

Sound symbolism: names carry meaning before they mean anything

This is the part that sounds like astrology and is not. The systematic review literature maps sound symbolism onto three dimensions from Osgood’s semantic differential work: evaluation (is it good?), potency (is it strong?), and activity (is it fast?). The organizing variable is acoustic frequency.

Higher-frequency sounds are front vowels (i, e, ee), fricatives (f, s, v, z, sh), and voiceless consonants (p, t, k, f, s). Lower-frequency sounds are back vowels (o, u, oo, aw), stops (b, d, g, p, t, k), and voiced consonants.

The review counted how many studies found each association:

Dimension	High-frequency sounds associate with	Studies agreeing
Evaluation	Higher evaluation: good, pleasant, sweet	39 of 51, 76.5 percent
Potency	Lower potency: small, light, weak, delicate	76 of 78, 97.4 percent
Activity	Higher activity: fast, sharp	27 of 43, 62.8 percent

The potency effect at 97.4 percent is close to categorical. That is the strongest single lever you have. If your product needs to feel light, quick, precise, or premium, load the name with front vowels and fricatives. If it needs to feel heavy, solid, durable, or powerful, use back vowels and voiced stops.

Real names line up with this almost too neatly. Nike has a front /i/ and reads light and agile. Volvo is all back vowels and reads heavy and stable. Google has a back /u/ and two voiced stops and reads round and friendly. Kodak is bookended by voiceless velar stops and reads sharp, like a shutter.

Bouba and kiki

The related finding is the bouba/kiki effect, first observed by Köhler in 1929 and popularized by Ramachandran and Hubbard in 2001. Given a rounded blob and a spiky star, people across languages call the blob bouba and the spike kiki. It is one of the most replicated results in the field. It shows up when people draw shapes freely rather than choosing between two, it is driven by sound properties rather than anything speech-specific, and brain imaging shows round and spiky sound-symbolic words producing distinct responses in auditory, parietal, and even early visual cortex. It has been reported in naive baby chicks.

Practically: decide whether your brand is round or sharp before you generate, then filter for it. A wellness brand and a cybersecurity brand should not be drawing from the same sound pool. The brand name generator is the fastest way to produce candidates on one side or the other, and the product name generator is the place to do it for individual SKUs inside an existing brand.

Examples by name type

Different name types earn their keep in different places. Here is what each buys you.

Type	Examples	What it does well
Real word, used arbitrarily	Apple, Amazon, Oracle, Slack	Instant familiarity, strong trademark, zero explanation of the product
Compound	Facebook, YouTube, PayPal, Snapchat	Self-explaining, easy to say, long
Clipped compound	Microsoft, Netflix, Amtrak, Velcro, Nabisco	Short, ownable, keeps a trace of meaning
Blend / portmanteau	Groupon, Pinterest, Instagram, Verizon, Frappuccino	Compresses two ideas into one word, usually registrable as suggestive
Affixed	Shopify, Spotify, Calendly, Grammarly	Cheap to generate, instantly reads as a product, now very crowded
Invented and poetic	Google, Kodak, Oreo, Xerox	Maximum trademark strength, maximum memorability, needs work to give it meaning
Phonetic respelling	Lyft, Flickr, Tumblr, Fiverr	Frees up a taken domain, costs you the radio test

Notice that the blend row is the only one that gets both a short result and retained meaning. That is why it keeps appearing in the top of brand portfolios. Groupon tells you it is about groups and coupons in seven characters. Instagram is instant plus telegram, which was a precise description of the product in 2010.

The trade-off nobody mentions

Trademark law ranks names from generic (never registrable) through descriptive, suggestive, arbitrary, to fanciful (a coined string with no independent meaning). Fanciful marks get the broadest protection. Exxon and Kodak sit there.

Now hold that next to the memorability rule. The more of your source words survive in a blend, the more obvious the meaning and the easier the name is to explain. But the more obvious the meaning, the closer the mark slides toward descriptive, and the weaker the protection. Groupon is transparent and had to work hard to hold its mark. Verizon reads as invented and sits comfortably in fanciful territory.

So the honest statement is this: the better your portmanteau works as an ordinary word, the weaker it is as a trademark. Pick your side deliberately. If you are building something you intend to defend, bias toward the opaque end and budget for the explaining. If you are building something that needs to sell itself in a crowded marketplace listing, bias toward the transparent end and accept a narrower mark.

When you have candidates, run them through the five tests above before you get attached, and check the domain situation early with the domain name generator so you find out about the .com before you find out about the logo.