Method
Evidence you can check
Most naming tools hand you a number. Memorability 9.0. Brandability 8.5. Pronounceability 9.5. The numbers look like measurements, so they get treated like measurements, and they decide things.
They aren’t measurements. A score is a pile of judgements compressed into one figure, with the judgements thrown away. Nobody can tell you what separates a 9.5 from a 9.0, including the model that produced it. And the compression destroys the part you could have used: that the word is three syllables, that it has a common misspelling, that a live federal trademark already covers it in your category.
So we show the evidence and skip the number. It asks more of you. It also means nothing on the screen is pretending to know something it doesn’t.
What that looks like in practice
Unknown is a real answer, shown as itself. A domain check returns available, taken, complicated, or unknown. Complicated means something specific: redemption period, premium pricing, registry hold. Unknown means the check didn’t complete. Neither one gets rounded to a green check or a red x, because rounding them is how you end up confidently wrong.
Registry status and acquisition evidence are separate. A name may be available for standard registration, shown in an aftermarket listing, offered through a make-offer page, registered with no listing we observed, or unresolved. These are factual pathways, not scores. A missing observed listing never claims that no listing exists.
Trademark data is evidence, never clearance. You get counts and the actual records: live exact matches, dead exact matches, close text matches, with links to the source. You never get a checkmark. A zero-match result is not permission to use a name, and no software can give you that permission. An attorney can.
Prices are shown with their shape intact. First-year and renewal prices appear separately and are never blended into an average, because a discounted first year presented alone reads as the ongoing cost. Where we list a figure we say it’s the lowest we checked, not the cheapest available, because our coverage isn’t exhaustive and saying otherwise would be a guess.
Absence of evidence is labelled as absence. A domain with no marketplace listing isn’t proven unavailable for purchase. A word with no dictionary entry isn’t proven to be a coinage. When we haven’t checked something, the screen says we haven’t checked it rather than leaving a blank you might read as a result.
One name, both ways
spandrel is a real word with a real screening behind it. Here is what a score would have told you about it, and what we show instead.
What a score gives you
That number is invented, and it has to be. No tool that publishes a score like this publishes how it was computed, so there is nothing to reproduce. That is the point rather than a complaint.
What we show
Screened 5 September 2026. A point-in-time snapshot, not a live lookup, and not a guarantee.
The 8.7 is not wrong so much as empty. spandrel genuinely is easy to say and easy to spell, and the .co is open, so a high score is defensible. What the number cannot tell you is that a live federal trademark already exists on the exact word.
And notice what the evidence lets you do next, which is the part that actually matters. That mark sits in class 016, which covers printed matter, and the record itself lists art prints and paintings. If you are naming a piece of software, it may not conflict with you at all. If you are naming a print studio, it plainly does. No single figure can carry that distinction, and neither can a green checkmark. Only the records can, which is why we show you the records.
The renewal price is missing because the registrar didn’t return one. We say so rather than estimating it or quietly showing the first-year figure twice.
Where AI is and isn’t used
The associations that fill the workspace come from dictionary and phonetic data, not from a language model inventing words it thinks sound good. There is an optional AI lens, clearly marked as one lens among several, and it suggests directions rather than ranking them.