Content Optimization Prompts

By Amit Patel

Optimization prompts work on a draft that already exists — paste the real text in, every time. Asking a model to “optimize my article about [topic]” without the actual text produces generic advice you could have guessed yourself. These prompts assume you’re pasting in the real, published or drafted content.

Topical Completeness

• “Review this article [paste full text] for topical completeness against the keyword ‘[keyword]’ — what subtopics or questions are missing?”
• “Does this article [paste] answer the actual question implied by its own H1 in the first 100 words?”
• “Given this article [paste] and these ‘people also ask’ questions [paste], which questions does the article NOT currently answer?”

Sentence-Level Clarity

• “Identify vague or generic sentences in this paragraph [paste] and suggest more specific, concrete rewrites.”
• “Rewrite this intro paragraph [paste] to state the specific value in the first two sentences instead of building up to it.”
• “Identify passive-voice sentences in this text [paste] and rewrite them in active voice.”
• “Does this paragraph [paste] use unsupported superlatives (best, ultimate, #1) that should be replaced with something specific?”

Structure and Scannability

• “Does this article’s heading structure [paste headings] let a skimming reader find what they need without reading every paragraph?
• “Where in this article [paste] would a table, numbered list, or comparison format improve clarity over a paragraph?”
• “Suggest a better closing section for this article [paste] — one that gives a clear next step, not just a summary.”
• “Suggest 3 places in this article [paste] to add a specific number, date, or named example instead of a vague claim.”

E-E-A-T Signals

• “Does this article [paste] read as written by someone with firsthand experience, or could it have been written without ever using the product/process it describes?”
• “What claim in this article [paste] would benefit from a citation or source link to be more credible?”
• “Suggest one place in this article [paste] where a specific, named example would replace a generic statement.”

Guardrails

When a model suggests adding a “specific example” or “named case study,” it will sometimes invent one if you don’t stop it. Always follow up with: “only suggest examples I can actually verify or that come from data I’ve provided — don’t invent a case study, statistic, or quote.” Treat every suggested addition as a placeholder to fill with something real, not text to paste in as-is.

Frequently Asked Questions