The short answer: build a shortlist of three to five names yourself, then show each one to readers who have not heard you talk about the book, in a short scene, and ask what they picture before they know anything else. Record first impressions, pronunciation and recall a day later. The name that is remembered, said the same way by everyone, and pictured closest to your character wins.
Most naming advice stops at the list. The list is the easy half. What you cannot do alone is hear a name the way a stranger hears it, because you already know who the character is. Testing with readers fixes that, and it takes an afternoon, not a survey budget.
Why a name needs testing at all
A reader meets a name before they meet the person. In those first few lines the name carries age, class, region, era and sometimes temperament. If it carries the wrong ones, you spend chapters correcting an impression you created on page one.
Three failures come up again and again, and none of them is visible to the author:
- Pronunciation drift. Half your readers say it one way, half another, and the audiobook narrator picks a third.
- Collision. Two names in the cast share a first letter or a sound, and readers mix them up in dialogue.
- Borrowed baggage. The name belongs to a celebrity, a famous fictional character or a brand, and readers bring that person with them.
Step 1: Build the shortlist yourself
Start from the character, not from a generator. Write down, in one line each, where and when they were born, who named them and why, and what they would prefer to be called. Those four facts rule out most names before you look at a single list.
Then gather candidates from sources that match those facts:
- Period name records. National statistics offices publish the most popular given names by birth year. A name that was common in the year your character was born reads as true; one that only became popular decades later reads as an anachronism.
- Regional and family sources. Local newspapers, parish and school records, and name-meaning dictionaries for the culture the family comes from.
- Your own cast list. Write every existing name in the book on one page. Strike any candidate that shares a first letter, a syllable count and a stress pattern with a name already there.
Say every survivor aloud in a full sentence of dialogue: shouted across a room, whispered, and used by someone who dislikes the character. Keep three to five. More than five and readers stop reacting and start comparing.
The author's job is the shortlist. The reader's job is the verdict. Mixing the two, by asking readers to invent names, gives you their taste instead of your character.
Step 2: Write one neutral test scene
Do not test names as a bare list. A name on its own gets judged for sound; a name in a scene gets judged for fit. Write a short passage, 150 to 300 words, where the character enters, acts and speaks, without anyone describing them. Then make one copy of the scene per candidate name, changing nothing else.
Keep the scene neutral. If the character does something heroic, readers will like whichever name they read first. Ordinary behaviour, like ordering coffee, answering a phone or arriving late, shows what the name alone contributes.
Step 3: Choose readers who have no context
The best testers have never heard you describe the book. Friends who know the plot are already attached to a version of the character. Five to ten readers is enough to see a pattern, and reading one short scene takes each of them a few minutes.
Give each reader only one version of the scene. If one person reads all five versions, the first name they saw colours the rest. Rotate the versions so every name is read by roughly the same number of people.
Step 4: Ask the same five questions every time
| Question | What it tells you |
|---|---|
| How old is this person, roughly? | Whether the name fits the era and age you intend. |
| Describe them in three words. | The personality the name suggests before you have written any. |
| Say the name out loud. How would you spell it? | Pronunciation drift and spelling confusion. |
| Does it remind you of anyone? | Borrowed baggage from real or fictional people. |
| (Next day) What was the character called? | Recall. A name nobody can remember will not survive a 90,000-word book. |
Ask the questions in writing and in that order, so an early answer does not steer a later one. Do not explain the character until every reader has answered.
Step 5: Read the answers for agreement, not votes
Resist counting favourites. A name can be everyone's favourite and still be wrong for the character. Look instead for agreement between what readers pictured and what you intend:
- Did most readers put the character in the right age band?
- Do their three-word descriptions overlap with yours?
- Did everyone say the name the same way?
- Could they recall it the next day without prompting?
A name that passes all four is safe. A name that fails recall or pronunciation is a problem you would keep paying for in every chapter, even if readers liked it.
- Readers guess the wrong decade: the name belongs to another generation. Go back to the period records.
- Readers spell it three ways: simplify the spelling or pick a different name.
- Readers mention a celebrity: unless you want that echo, drop it.
- Readers confuse it with another cast member: change whichever name has less history in the draft.
When to test, and when not to bother
Test the names readers will see most: your protagonist, the antagonist and anyone who appears in most chapters. Walk-on characters do not need a panel; a quick read-aloud and a cast-list check are enough.
Test early. Once a name has lived in a draft for months, you will defend it against the evidence. The cheapest moment to change a name is before the first full draft exists.
Keeping the test organised
You can run this with a document and a spreadsheet. If you already share chapters with beta readers, the same private links work for a test scene. In Write in a Click, beta reader links let a reader open a scene without an account and leave notes that come back grouped by chapter, so each name version can sit in its own short chapter with its own link.
- Write your protagonist's four naming facts: where, when, who named them, what they prefer to be called.
- Build a shortlist of three to five names from period records and your cast list.
- Write one neutral 200-word scene and make a copy per name.
- Send one version each to five readers with the five questions.
- Ask the recall question the next day, then compare answers with your intent.
Frequently asked questions
How many names should I test at once?
Three to five. Fewer and there is no real choice; more and each name gets too few readers to show a pattern.
How many readers do I need?
Five to ten is enough for one character. You are looking for strong agreement or clear confusion, not statistical proof.
Should I tell readers what the test is for?
Not until they have answered. Knowing it is a naming test makes people judge the name as a word instead of reading the scene.
Can I test names for a fantasy world?
Yes, and it matters more there. Invented names fail on pronunciation and recall far more often, so the spelling and next-day questions do most of the work.
