How to Test If a Name Passes the Drunk Test
Can someone half-paying-attention say it, spell it, and find you? Here is how to know — and what it predicts.
Someone you just met at a party tells their friend they should check out this thing they read about. Their friend, three drinks in, is half-listening. Twenty seconds later the friend is on their phone trying to find your site. They will type whatever they thought they heard. If what they thought they heard does not match your URL, you do not exist for them. They are at someone else's site, or a parked domain, or a typo'd nothing. The drunk test — sometimes called the noisy bar test — is shorthand for a real diagnostic about whether your name survives the conditions in which it actually travels. People are not paying full attention when they hear about your company. They are at a party, on a treadmill, in a conversation about something else. The names that grow are the ones that work in those conditions. There is a way to test it before you launch.
What follows: how to run the drunk test, what it tells you, and what to do if the name fails. Then a tool that includes it in the full audit.
Record the name being said in a noisy environment
Play music at moderate volume. Have someone say your name into your phone, normally, the way they would mention it in conversation. Do not pronounce it carefully. Do not give context. Then play the recording for someone who has never heard the name and ask them to type what they think they heard. Whatever they type is what your customers will type. If it does not match the URL, you have a real problem.
Test on at least five people you have not coached
Friends who have heard you say the name fifty times will recognize it. Cold ears will not. Test on five people who have never been exposed to the name. Their typing accuracy is your real-world signal. Any pass rate under eighty percent is a yellow flag; under sixty is red. Do not average across coached and uncoached listeners. The cold listeners are the only ones whose data is meaningful.
Watch for the substitutions, not just the misses
When people guess wrong, look at what they typed. If they typed a different real word that was phonetically close, your name has a homophone problem — your customers will land on someone else's site. If they typed a typo'd version of your name, the problem is less serious; you can buy the typo domain and redirect. Substitutions are worse than typos. Diagnose which kind of failure you have.
Fix by changing one consonant, not the whole name
Most drunk-test failures come from one specific phoneme that gets misheard — typically a soft consonant that blurs into the next syllable. Identify that consonant. Try variants where it is sharper, harder, or replaced with a clearer sound. Often a one-letter change converts a failing name into a passing one. Whole renames are usually overkill; targeted fixes are usually enough.
If a name fails the drunk test repeatedly, accept it and move on
Some names cannot be fixed without losing what made them appealing. If you have iterated and the name keeps failing the drunk test, it is the wrong name for a brand that needs to spread by word of mouth. Some names work for paid-acquisition-only businesses where customers are typing from a screen. Most early-stage businesses are not those. The drunk test is a constraint, not an opinion. Respect it.
Stress-test your name across 12 dimensions before you commit.
Enter the name. Get the deep analysis: phonetics, memorability, the drunk test, global language scan, radio test, SEO, competitive landscape, plus live domain and social handle checks.