Products SEO & GEO Agent Paid Ads Agent Website Agent Conversion Agent
How it works Pricing FAQ Blog Book a call

ChatGPT prompts for blog writing that sound like you

Updated · 8 min read

Short answer Don't ask ChatGPT to write the post. Ask it to interview you, then to turn your answers into a draft. Use one prompt per stage: topic, outline, interview, draft by section, edit, title and meta description, fact-check list. The facts and stories come from you, so the post sounds like you. Check every claim before publishing.

Want help applying this to your business? Talk it through on a free 30-minute call.

Book a call

The short version: one prompt per stage

A blog post written by ChatGPT from a one-line request sounds like everyone else's, because the only facts in it are the ones every site already has. The fix is to split the job. You supply the knowledge; ChatGPT supplies structure, speed and polish.

StageYou giveChatGPT gives back
1. TopicQuestions customers asked this month10 post ideas, ranked
2. OutlineThe chosen questionHeadings in the reader's words
3. InterviewSpoken or typed answersQuestions, one at a time
4. DraftYour answersOne section at a time, in your words
5. EditThe draftCuts, plain words, AI tells removed
6. Title and metaThe final postOptions to pick from
7. Fact-checkThe final postA list of every claim to verify

Stage 3 is what makes the post sound like you. Skip it and you get a generic post with your name on it.

OpenAI's own advice backs the two habits this rests on: give context, and refine in rounds.

"Ensure your prompts are clear, specific, and provide enough context for the model to understand what you are asking." (OpenAI Help Center: Prompt engineering best practices for ChatGPT)
"Start with an initial prompt, review the response, and refine the prompt based on the output." (OpenAI Help Center: Prompt engineering best practices for ChatGPT)

Do all seven stages in one chat, so ChatGPT keeps your answers in view. If you haven't saved your voice rules yet, do that once first; ChatGPT prompts that keep your brand voice shows how.

Prompts for blog topic ideas

Start from real questions, not "blog ideas for a dentist." Paste what customers actually asked: emails, call notes, review comments.

Prompt: "Here are questions my customers asked this month: [paste]. I run a [business] in [town]. Suggest 10 blog post topics, each answering one question a customer would type into Google. For each, give the question in the customer's words and one line on why a reader would care. Rank them by how close the reader is to hiring someone."

Pick a topic you can answer from experience. If you'd need to research it from scratch, ChatGPT will too, and that's where generic posts come from.

Prompts for a blog outline

Prompt: "Write an outline for a post answering: [question]. Put the direct answer in the first section. Use the follow-up questions a customer would ask next as headings, in plain words. Mark any section where my own experience is needed with [ME]. No introduction section, no conclusion section."

Check the outline against the order a customer asks things on the phone. Move or delete headings until it matches. For the general steps of a post, see how to write a blog post.

The interview prompt: how to make ChatGPT sound like you

This is the one prompt most guides skip. Instead of writing, ChatGPT asks and you answer.

Prompt: "Using this outline, interview me to write the post. Ask one question at a time and wait for my answer. Ask for specifics: prices, times, mistakes customers make, a job I remember, what I tell people that other businesses don't. If my answer is vague, ask a follow-up. Stop after 10 questions."

Answer the way you'd talk to a customer. Use the microphone in the ChatGPT app if typing is slow. Rough answers are fine; your phrasing and your facts are what the draft will be built from.

Google's people-first content questions ask about exactly this:

"Does your content clearly demonstrate first-hand expertise and a depth of knowledge (for example, expertise that comes from having actually used a product or service, or visiting a place)?" (Google Search Central: Creating helpful, reliable, people-first content)

Your answers to the interview are that first-hand expertise. ChatGPT can't invent it.

Prompts for drafting the post, one section at a time

Long drafts drift into filler. Ask for one section, read it, then ask for the next.

Prompt: "Write section 1 using only my answers above. Keep my wording where you can. Short sentences. Don't add facts, numbers or examples I didn't give you; if something is missing, write [NEED] instead."

Then, for the rest:

Prompt: "Next section. Same rules. Use the story I told about [job] here."

For the opening and the ending:

Prompt: "Write a two-sentence opening that answers the question right away. No "In today's world", no question to the reader."

Prompt: "Write a last line that tells the reader the one next step, in my voice. No summary of the post."

The [NEED] tag matters. It turns a guess into a gap you can see and fill.

Prompts for editing an AI-sounding draft

Paste the full draft back and run these one by one:

  • "Cut this by a quarter without losing a fact or a story."
  • "List every sentence a competitor could publish unchanged. Rewrite each with a detail from my answers, or delete it."
  • "Replace words a customer wouldn't use with plain ones. List what you changed."
  • "Remove these words everywhere: [your banned list, such as delve, elevate, seamless, unlock, navigate]."
  • "Point out where I contradict myself or where a step is missing."

Then read it aloud once. Anything you wouldn't say to a customer goes. To use tone words, OpenAI suggests plain adjectives: "Use descriptive adjectives to indicate the tone" (OpenAI Help Center: Prompt engineering best practices for ChatGPT). "Friendly, direct, no hype" works better than "make it engaging."

Prompts for the blog title, meta description and keywords

The title is what people click in search. The meta description is the short summary that may appear under it.

Prompt: "Suggest 8 titles for this post. Start with the words a customer would search: [question]. Under 60 characters. No exaggeration, no clickbait."

Prompt: "Write 3 meta descriptions under 155 characters that say what the reader learns and for whom."

Google's guidance says the fact-check and review it asks for also covers title elements, meta descriptions, structured data and image alt text, since all of them can show in search results (Google Search Central: Guidance on using generative AI content).

Skip the old prompt "rewrite this to include keywords X, Y and Z." Forced keywords read badly, and Google's spam policies name "Repeating the same words or phrases so often that it sounds unnatural" as keyword stuffing (Google Search Central: Spam policies). If the post answers the question in your customer's words, it already contains them. More on this in keyword stuffing and on-page SEO for blog posts.

The fact-check prompt: what ChatGPT gets wrong

ChatGPT fills gaps with plausible guesses. OpenAI lists the kinds of mistakes it makes, including:

"Fabricated quotes, studies, citations or references to non-existent sources" (OpenAI Help Center: Does ChatGPT tell the truth?)

So the last prompt is an audit, not a rewrite:

Prompt: "List every fact in this post that a reader could check: numbers, prices, dates, laws, rules, product names, quotes, statistics. For each, say whether it came from my answers or from you."

Everything marked "from you" gets checked against the original source, or cut. Google's guidance is plain:

"It is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing." (Google Search Central: Guidance on using generative AI content)

Should you say ChatGPT helped write the post?

Google doesn't require a label on every AI-assisted post. Its helpful content page suggests a disclosure where readers would wonder how a post was made: "AI or automation disclosures are useful for content where someone might think 'How was this created?'" (Google Search Central: Creating helpful, reliable, people-first content). If the post is built from your interview answers and you edited it, it's your post. If ChatGPT wrote most of it, a line on how it was made is reasonable.

What does hurt: publishing many thin AI posts. Google's spam policies list "Using generative AI tools or other similar tools to generate many pages without adding value for users" as scaled content abuse (Google Search Central: Spam policies). One good post a month beats twenty generic ones. See does Google penalize AI content?

Example

Illustrative cases, round numbers.

A plumber's water heater post (US dollars)

He asked ChatGPT for "a blog post about water heaters" and got 900 words any plumber could post. Second try: the interview prompt. He answered ten questions by voice in 15 minutes: tank vs tankless, the $1,500 to $3,000 range he quotes locally, the sediment problem he sees in his town's hard water. The draft used his phrases. The fact-check prompt flagged two numbers ChatGPT had added; he cut one and confirmed the other.

A bookkeeper's tax deadline post (Canada)

The draft listed a filing deadline ChatGPT supplied. The fact-check prompt marked it "from ChatGPT." She checked it against the Canada Revenue Agency page, corrected the date and added the late-filing mistake her clients make most.

A dog groomer's outline

ChatGPT's first outline started with "The history of dog grooming." She told it to use only questions clients ask at the front desk. The new outline opened with "How often should a doodle be groomed?"

FAQ

What is the best ChatGPT prompt for writing a blog post?

The interview prompt: ChatGPT asks you questions one at a time, then drafts only from your answers. It adds facts no other site has.

Can ChatGPT write a whole blog post for me?

It can, but a one-prompt post has only generic facts and may invent some. Use it to structure and polish what you know.

Will Google penalize a blog post written with ChatGPT?

Google targets content made to manipulate rankings, however it was made. A checked post built from your experience is fine; many thin AI posts are not.

How long should a ChatGPT blog prompt be?

As long as the context needs. A few lines with the reader, the question, your facts and the limits beat one vague sentence.

Different for you?

Keep learning

NextLLMs in SEO work: tasks they speed up and the errors to catchWhich SEO tasks ChatGPT, Claude or Gemini can speed up for a small business, the mistakes each makes, and a quick way to check the output before it goes live.

Do it yourself

AI SEO agent review: 8 checks before you chooseEight tests that show whether an AI SEO agent does real work on your site or only sends reports, using your own Search Console and website accounts.

Ask us one question

One line is enough. We reply by email.

Sources

  1. OpenAI Help Center: Prompt engineering best practices for ChatGPT
  2. OpenAI Help Center: Does ChatGPT tell the truth?
  3. Google Search Central: Guidance on using generative AI content (last updated 2026-10-01)
  4. Google Search Central: Creating helpful, reliable, people-first content (last updated 2026-10-05)
  5. Google Search Central: Spam policies for Google web search (last updated 2026-08-28)