Before the shortlist goes to legal, know it actually works.
Trademark and legal checks tell you whether a name is available. They don't tell you whether it fits the brand, whether people can say it without stumbling, or what it brings to mind for someone hearing it cold. That's a separate question, and it's worth answering first.
Internal opinion isn't a validation process
By the time a naming shortlist reaches final sign-off, everyone in the room has usually said each name out loud a hundred times and stopped hearing it fresh. Real respondents, hearing a candidate name for the first time, are the only honest test of whether it actually lands the way it's supposed to.
What gets tested per candidate
- Fit: does the name feel right for what the brand actually is and does
- Ease: can people say and remember it without having to think about it
- Association: what does it bring to mind unprompted, catching unintended meanings before they become a launch-day problem
- Timing and hesitation on each answer, often the clearest signal that a name isn't quite landing even when the stated answer looks fine
Comparing a shortlist directly
Run the same standardised question set against each candidate and compare results side by side. The AI report reads a naming campaign specifically and writes a verdict about fit, ease, and association for that name, not a generic summary that could apply to any question set.
About brand name testing
When in the naming process should I test?
Once you've narrowed a longlist to a realistic shortlist, before it goes to trademark and legal checks. Testing earlier means feedback on names that might not clear those checks anyway; testing later means less room to change course.
How many names can be on the shortlist?
As many as make sense to test, run each as its own campaign or structure a single campaign to compare a shortlist directly.
Does this help make the case internally?
That's often the real value: a client or stakeholder who's attached to a name they picked is easier to redirect with real respondent data than with another round of internal opinion.