Prompt

6 minute read

Content Brief Prompt: Plan a Page Around the Reader

A reusable content brief prompt that turns a topic, real queries and verified notes into a short brief with a promise, an outline, sources per section and clear scope.

Abstract illustration for Content Brief Prompt: Plan a Page Around the Reader

This content brief prompt turns a topic, your research notes and a few search queries into a brief a writer can follow. It defines the reader, the question the page must answer, the outline, the sources to use and what to leave out. You approve the brief before anyone starts drafting.

The page explains each variable, gives the full prompt and shows one real run on an invented topic. It also lists the checks that keep a brief grounded in evidence. The prompt never invents search volumes or statistics, because a model cannot know them.

What this content brief prompt does

The prompt starts from the reader's task, not from a keyword list. It writes a one-sentence promise for the page, an outline in which every heading answers a real question, and a list of the sources each section should cite.

It also names what the page should not cover, so the article stays focused. Finally, it flags claims that need a source you have not supplied yet, which keeps the writer from filling gaps with guesses.

When to use it

Use it before writing a new article or guide, especially when several people work on one piece. It saves the editor from rewriting a draft that answered the wrong question.

  • A topic from your content calendar that needs a clear angle.
  • A freelance writer who needs one page of instructions.
  • A cluster plan where each page must stay within its own intent.

However, if the topic overlaps an existing page, update that page instead of briefing a new one. Our content refresh brief prompt handles that case.

Inputs and variables

Replace every bracketed variable before you run the prompt. The research notes matter most, because the brief can only be as accurate as its sources.

  • [TOPIC]: the subject in one line, for example "how to winterize a road bike".
  • [READER]: who will read it and what they already know.
  • [QUERIES]: real queries from Search Console or a keyword tool, with any numbers labelled by source.
  • [RESEARCH_NOTES]: facts you have verified, each with its source URL.
  • [EXISTING_PAGES]: related pages on your site, so the brief can link to them and avoid overlap.

Use public sources and your own data. Do not paste unpublished plans or client information into a public AI tool.

Reusable prompt — version 1.0

Copy the prompt below as it is. It asks for a source next to every section, which makes unsupported sections obvious.

You are helping an editor write a content brief about [TOPIC] for [READER].

<task>
Write a brief a writer can follow to produce one useful page.
</task>

<rules>
1. Start with the reader's task and a one-sentence promise the page must keep.
2. Build the outline from the questions in <queries> and the task. Every heading must answer a real question.
3. Next to each section, name the source from <notes> it should use. If no source supports it, mark NEEDS SOURCE.
4. Never invent search volumes, statistics, quotes, product details or dates.
5. List what the page should not cover, and which pages in <pages> it should link to instead.
6. Keep the brief under 400 words.
</rules>

<output_format>
1. Reader and task
2. Promise (one sentence)
3. Outline: heading | what it answers | source
4. Out of scope, with links to existing pages
5. NEEDS SOURCE list
</output_format>

<queries>[QUERIES]</queries>
<notes>[RESEARCH_NOTES]</notes>
<pages>[EXISTING_PAGES]</pages>

Rule 3 is the safeguard. A section marked NEEDS SOURCE waits for research instead of reaching the writer as a guess.

How to use it

Any capable chat model can run this content brief prompt. Short, verified notes produce better briefs than long, unchecked ones.

  1. Collect real queries and verified facts with their sources.
  2. List the related pages already on your site.
  3. Fill in the variables, paste the prompt into a new chat and run it.
  4. Resolve every NEEDS SOURCE item, then send the brief to the writer.
  5. Check the finished draft against the promise and the outline.

Example input and output

We ran version 1.0 once on Claude Opus 5.5 on November 15, 2026. The topic and notes are invented for this example; the output is trimmed, and trims are marked with "…".

TOPIC: how to winterize a road bike
READER: commuters who ride year-round and do basic maintenance
QUERIES: how to winterize a bike, winter bike chain lube, store bike in winter
RESEARCH_NOTES:
- Wet lube suits rain and snow (manufacturer guide, northwind.example/lube)
EXISTING_PAGES: /guides/fix-a-flat-tire, /guides/adjust-disc-brakes
1. Reader and task: a year-round commuter who wants the bike to survive
   wet, salty roads with basic tools.
2. Promise: after reading, you can prepare a road bike for winter in
   under an hour.
3. Outline:
   | Clean and protect the frame | what to do about road salt | NEEDS SOURCE |
   | Choose a winter chain lube | wet vs dry lube | manufacturer guide |
   | Storing a bike you will not ride | how to store it | NEEDS SOURCE |
   …
4. Out of scope: flat repair (link /guides/fix-a-flat-tire).
5. NEEDS SOURCE: salt corrosion advice; storage advice.

In this run, the model marked two sections that had no source instead of filling them in. That single run is an illustration, not a benchmark.

How to check the output

Review the brief before the writer receives it.

  • Promise: the one-sentence promise is specific and something the page can keep.
  • Sources: every section names a real source you supplied, or is marked NEEDS SOURCE.
  • No invention: there are no volumes, statistics or quotes without a source.
  • Overlap: out-of-scope topics point to existing pages instead of repeating them.
  • Length: the brief stays short enough that the writer actually reads it.

Troubleshooting

These fixes address the bad outputs we see most often.

  • The outline is generic: add more real queries and a sharper description of the reader.
  • It invents statistics: repeat rule 4 at the end of the prompt.
  • Every section needs a source: do more research first; the prompt is working as intended.
  • The brief is too long: lower the limit in rule 6.

Limitations

The prompt cannot check whether your research is complete or what competing pages cover. Therefore, the editor still decides the angle and approves the brief. It also cannot measure demand; any numbers must come from your own tools.

Outputs vary by model, version and input. Treat the brief as a draft for an editor, in line with Google's guidance on helpful, people-first content.

Version history

Each version carries a date, so you can tell which prompt produced an older brief.

  • 1.0: November 15, 2026. First version, tested once on Claude Opus 5.5.

Frequently asked questions

Quick answers to the questions readers ask most about this topic.

Which AI models does this content brief prompt work with?

Any capable chat model that follows structured instructions, such as current models from OpenAI, Anthropic or Google. We tested it once on Claude Opus 5.5.

Can the prompt find keywords and search volumes for me?

No. A model cannot know search volumes. Paste real queries from Search Console or a keyword tool, and label any numbers with their source.

What does NEEDS SOURCE mean in the brief?

It marks a section that none of your verified notes supports. Research it before the writer starts, so the draft does not fill the gap with guesses.

Should I brief a new page or update an existing one?

If an existing page already serves the same reader task, update it instead. Use our content refresh brief prompt for that case.

Sources

These pages support the content guidance and the prompting approach above.

  1. Google Search Central: Creating helpful, reliable, people-first content developers.google.com
  2. Google Search Central: Guidance on using generative AI content developers.google.com
  3. Anthropic: Claude prompting best practices platform.claude.com