Test names with useful, specific questions
Turn observed confusions into deliberate revisions without pretending that taste is an objective score.
Ask about a task instead of general goodness
Somebody can like a name and still confuse it with another character. A form can feel unfamiliar initially and become usable after a good introduction. General liking is therefore one observation among several. The more useful development question is whether the name fulfils its intended task.
Choose a small realistic situation: a direction with three locations, a conversation with two roles or an equipment list. Supply the information a player would actually have at that moment. Explaining the entire background first tests memory of the explanation more than it tests the name.
Separate observation from explanation
Write down what happens before deciding why. A reader pauses at a boundary, reverses two forms or remembers only an opening word. Possible explanations come afterwards. The pause might concern an unfamiliar spelling, an unclear task or an interruption. Treating every hesitation as rejection produces misleading revisions.
Choose one question per short test. Recognition, pronunciation and cultural association require different conditions. A single conversation is not a comprehensive study. It can nevertheless reveal a particular difficulty that the team can address and test again.
Make a comparison fair
Use the same text and presentation for two candidate names. Do not simultaneously change the character, illustration and background. Otherwise several differences are being tested together. If several readers participate, vary the order of presentation so the first version does not always establish the initial context.
Have readers note their initial interpretation before discussion. Afterwards, ask which words stood out, what was remembered and which expectation formed. Avoid leading questions that supply the preferred answer. Asking what somebody expects at a place reveals more than asking whether a name sounds appropriately mysterious.
Revise one observed problem
In a fictional trial, two destination names share the same first word and differ only at the end. Each is readable alone, but the intended objective is later confused. Replacing another final letter may change little. A second revision changes the opening and links it to a visible feature, giving memory both a different word and an image.
Record the task and the actual confusion. Do not infer that the new form is universally superior or easier for everyone. The conclusion is narrower: under these conditions, the desired distinction became clearer. That precise record is useful when the interface or language version changes later.
Exercise: finish with a working decision
Not every unusual form needs to disappear after initial hesitation. An older or foreign-sounding name may deliberately belong to the world. Decide whether the difficulty supports that purpose or merely adds effort. A better introduction may help an important form; an unimportant complication may be better removed.
End the test with a clear action: keep, revise a defined feature or check again under different conditions. Record the reason and open questions. Generating more candidates then addresses a specific task instead of becoming an endless search for a supposedly perfect name.