Back to cookbook

High-Converting SEO Blog Post

Updated
Share

A prompt for producing a first draft that is already structured for search. It fixes the word count, where the primary keyword must appear, paragraph length and the meta description, so you edit a solid draft instead of rewriting a generic one.

It works best when you supply the real search intent and a few secondary keywords, because the model can then match the article to what readers actually want.

Prompt template

Act as a top-tier SEO copywriter. Write a 1,500-word blog post.

Topic: [TOPIC]
Primary keyword: "[KEYWORD]"
Secondary keywords: [KEYWORD 2], [KEYWORD 3], [KEYWORD 4]
Audience: [WHO IS READING]
Search intent: [informational / how-to / comparison]

Constraints:
- Include the primary keyword in the H1 and at least two H2s.
- Keep paragraphs under 3 sentences.
- Use one bulleted list and one short FAQ section (3 questions).
- Do not invent statistics. Mark anything unverifiable with [VERIFY].
- End with a 155-character meta description labelled "Meta description:".

Example input

Topic: How to reduce LLM token costs
Primary keyword: "reduce token usage"
Secondary keywords: prompt compression, context window, LLM cost
Audience: developers building AI features
Search intent: how-to

Example output

# How to Reduce Token Usage in LLM Apps (Without Losing Quality)

Every extra token costs money and latency. Here is how to cut both.

## Why reducing token usage matters
Token bills grow with every request. A 30% smaller prompt is a 30% smaller input bill.

## 1. Compress the prompt
...

## FAQ
**Does prompt compression hurt accuracy?** ...

Meta description: Learn practical ways to reduce token usage in LLM apps: prompt compression, smarter context windows and caching, with examples.

When to use it

  • Producing first drafts for a content calendar
  • Refreshing thin posts that rank on page two
  • Making briefs consistent across freelance writers
  • Generating the meta description alongside the article

Best practices

  • State the search intent (informational, comparison, how-to).
  • Provide 3 to 5 secondary keywords and ask for natural use, not repetition.
  • Give the audience and a tone reference.
  • Fact-check statistics; ask the model to flag any figure it is unsure of.

Common mistakes

  • Asking for keyword stuffing ("use the keyword 20 times").
  • Skipping the audience, which produces bland, universal copy.
  • Publishing without adding original examples or first-hand experience.

FAQs

How long should the article be?

Match the length of the top-ranking pages for your keyword. The template defaults to 1,500 words; change the number to fit.

Will AI-written content rank?

Search engines rank helpful content regardless of how it was drafted. Add original insight, fix facts and edit for voice.

Can I use it for product pages?

Yes. Replace the blog constraints with product-page ones (benefits, specs, objections) and keep the keyword rules.

Found this prompt useful? Share it.

Share