Blog Outline Prompt: Search-Ready Outlines with Claude
A blog outline prompt for Claude that builds a search-ready article outline from your keyword, audience and sources, with H2/H3 structure, angle and FAQ ideas.
By Claudekit Team Published 2 min read
- Category
- Writing
- Use case
- Plan a blog post or guide before drafting
- Tested with
- Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5
Use this prompt before writing any article, tutorial or guide. It produces an outline that answers the reader’s real question, uses your sources and first-hand experience, and is structured for both readers and search engines.
The prompt
You are a senior content strategist and editor for a technical audience. Create an outline that a writer can draft from without further briefing.
<sources>
{{SOURCES}}
</sources>
<brief>
Primary keyword: {{PRIMARY_KEYWORD}}
Related terms: {{RELATED_TERMS}}
Target reader and their goal: {{AUDIENCE}}
Content type and target length: {{FORMAT}}
First-hand experience, data or opinions we can add: {{EXPERIENCE}}
</brief>
<instructions>
1. State the search intent (informational, how-to, comparison, commercial) and the one question the article must answer.
2. Propose 3 title options under 65 characters that contain the primary keyword naturally.
3. Write a 150-160 character meta description.
4. Build an H2/H3 outline. Under each heading, add 2-3 bullets on what it must cover and which source or experience supports it.
5. Put the direct answer near the top. Order sections by what the reader needs first.
6. Include places for code blocks, tables or screenshots where they help understanding.
7. Add 3-5 FAQ questions people are likely to ask next.
8. Use only facts from the sources. Mark claims that still need a source with "[needs source]".
9. If the brief is missing the audience or the goal, ask before outlining.
</instructions>
<output_format>
## Intent and core question
## Title options
## Meta description
## Outline
## FAQ ideas
## Internal link ideas
## Gaps to research
</output_format>
How to use it
- claude.ai: paste source excerpts or links’ content into
sources. Keep sources above the brief. - Claude Code: handy if your site lives in a repo, because Claude can check existing posts to suggest internal links. Save it as a skill at
.claude/skills/outline/SKILL.mdwith the keyword as$ARGUMENTS; see the skills guide. - API: works well in a content pipeline as the step before drafting.
Variables
| Variable | What to put there |
|---|---|
{{SOURCES}} | Official docs excerpts, notes, data |
{{PRIMARY_KEYWORD}} | The main query you want to rank for |
{{RELATED_TERMS}} | Secondary keywords and synonyms |
{{AUDIENCE}} | Who reads it and what they want to achieve |
{{FORMAT}} | “how-to guide, about 1,500 words” |
{{EXPERIENCE}} | Your tests, screenshots, results, opinions |
Example output (illustrative)
## Outline
H2: What is an MCP server?
- One-paragraph definition from the MCP docs (source 1)
- Diagram idea: client, server, tools
H2: How to add an MCP server to Claude Code
- claude mcp add example with --transport http (source 2)
- Our test: connecting a GitHub server [screenshot]
Tips and variations
- Competitor gaps: paste the headings of the current top results and ask “what do all of them miss?”
- Refresh mode: paste an existing article and ask for an updated outline with outdated sections flagged.
- For more techniques, see Claude prompting best practices.
Frequently asked questions
Why outline first instead of asking for the full article?
An outline is cheap to fix. Agreeing on the angle, structure and sources before drafting avoids rewriting thousands of words because the piece answered the wrong question.
Can Claude do keyword research with this prompt?
No. It works from the keyword and related terms you supply. Use your own SEO tool for search volume and difficulty data, then paste the results into the prompt.
How do I stop the outline from sounding generic?
Fill in the first-hand experience variable. Real examples, data or opinions you can contribute are what make a piece different from every other result, and the prompt builds sections around them.