List preference in writing

By Jack Stovell · published 2026-09-24

List preference is the format a writer uses for several items: bullets, numbered steps, inline in a sentence, a mix, or no lists at all. A ScriptGrain voice profile stores it as one of five labels (bullets, numbered, inline, mixed or avoids), judged by the extraction model from your samples. It does not move the voice match score.

Analyse your list preference free

Your free voice profile measures list preference among all 45 attributes, from a sample of your own writing. No card needed.

Try the six-signal check first

At a glance

What list preference measures

List preference measures how you present several items when you write, including whether you avoid lists entirely. Some writers reach for bullets the moment they've got three things to say. Others fold everything into a sentence: "you'll need a laptop, a decent mic and about an hour". Both are valid choices, and the attribute just records which one is yours.

The attribute sits in the punctuation and format group alongside things like comma density and ellipsis usage. It's a judged field, meaning there's no script counting your bullet points. Claude Sonnet 5 reads each sample in pass one, then a synthesis pass picks the label that best fits across everything you uploaded. That label gets snapped to one of five canonical values at save: bullets, numbered, inline, mixed or avoids. No in-between readings, no percentages. Just the closest fit.

How list preference is scored

List preference does not move your voice match score, full stop. The voice match engine has no comparator built for this attribute, so however your list habit gets read, it sits in the profile purely as description. It doesn't get compared against a draft, and polish tooling can't target it either.

It is one of 17 attributes that never move the score. When a draft is scored, code counts features such as sentence length and comma density; nothing counts list format. The label still goes to the model with every ScriptGrain draft. For the mechanics of what does move the score, the voice measurement framework explains the split, and the writing voice attributes reference lists all 45.

How to analyse list preference in your own writing

Analysing your own list preference takes about five minutes and a folder of your last few emails or drafts. Here's how:

  1. Pull ten to fifteen pieces of your own writing: emails, posts, reports, whatever you've got lying around.
  2. Count how often you present multiple items as bullet points versus how often you write them into a sentence.
  3. Check whether ordered steps get numbers, or whether you just bullet everything regardless of sequence.
  4. Look for inline lists, phrases like "you'll need X, Y and Z" buried in a sentence rather than broken out.
  5. Note whether you avoid lists almost entirely, favouring continuous paragraphs even when a list would be tidier.
  6. Tally the pattern. Whichever format dominates is your preference; if it's roughly even, you're "mixed".

That's the manual version. ScriptGrain's free voice profile reads this from a writing sample among all 45 attributes, without you tallying anything.

Examples of list preference in real writing

Example (bullets): a project update that reads "Blockers: waiting on design, waiting on legal, waiting on budget sign-off" as three separate bulleted lines.

Example (numbered): a recipe or tutorial that writes "1. Preheat the oven. 2. Mix the dry ingredients. 3. Fold in the eggs" because order genuinely matters.

Example (inline): a sentence like "the flat needs a new boiler, a lick of paint and probably a better landlord" that never breaks into a list at all.

Three writers, three completely different instincts, and none of them wrong.

How to make AI use lists the way you do

To make AI use lists the way you do, tell it explicitly which format fits the content and which fits your habit; left alone, it picks its own.

  1. Ask for numbered steps specifically when order matters, and bullets only when it doesn't.
  2. Keep list items parallel and short; don't let one bullet run four lines while its neighbours run four words.
  3. Tell the model outright which format you use: bullets, numbered, or inline. Don't make it guess.

A ScriptGrain profile sends your list preference inside the voice profile JSON block of the system prompt, under "Treat every attribute as a hard constraint, not a suggestion." There's no separate rule bolted on for this attribute; it travels with the rest of the profile. The same profile reaches ChatGPT, Claude and other tools through the API and the MCP server.

How to make AI stop using bullet points

To make AI stop using bullet points, ask directly for continuous prose with no bullet points at all, and then reinforce it by banning the habits that creep back in anyway.

  1. Ask explicitly for continuous prose, no bullet points, full stop.
  2. Fold any short list into a single sentence instead of breaking it out.
  3. Ban bolded topic sentences and headings on short pieces; that's often where the list instinct sneaks back in disguised as structure.

A profile that reads "avoids" goes to the model as a hard constraint like every other label. ScriptGrain's SGR-001 study shows what a profile can do to format: the bare arm produced a listicle with four headings and fourteen bolded topic sentences; the profiled arm, given the same brief, produced continuous prose with no headings. It compared whole profiles, not this one attribute.

Questions

What is list preference in writing?

It's the format you default to for presenting multiple items: bullets, numbered steps, inline within a sentence, a mix, or avoiding lists entirely. In a ScriptGrain profile it's one of five judged labels sitting in the Punctuation and format group, read from your writing samples rather than counted by any script.

Bullet points vs prose: which is better?

Neither is inherently better; it depends on the content and your own habit. Steps with a strict order suit numbers. Loose collections of related items often read better inline, as a sentence. Reports and updates often suit bullets. The honest answer is: match the format to what you're actually writing, and to how you naturally write.

How do I make ChatGPT stop using bullet points?

Ask for continuous prose with no bullet points, and be specific about it; a vague "write naturally" won't cut it. Fold short lists into sentences yourself as an example, then ban bolded topic sentences and headings on short pieces, since that's usually where the listicle habit sneaks back in.

How do I make AI write in paragraphs instead of lists?

Tell it plainly: continuous prose, no headings, no bullets. Give it one example paragraph in the shape you want, since models often follow a demonstrated pattern better than an abstract instruction. If it still reverts, explicitly ban headings and bolded lead sentences too, since those often smuggle list structure back in.

How do I make ChatGPT write without headings?

State it directly in your instructions: no headings, continuous prose only. This pairs well with banning bolded topic sentences, since bold lead-ins can stand in for headings: in ScriptGrain's SGR-001 study, the bare arm produced four headings and fourteen bolded topic sentences on one brief.

Related

Sources