Introduction
In my previous article “What is GEO (Generative Engine Optimization)?”, I shared the story of how one of my blog posts was cited by Google AI Overview, Gemini, and ChatGPT in just 7 days.
Even more surprisingly, just a few days ago another blog post I wrote for the same website was cited in Google AI Overview in only 2 days—Day 1 indexed → Day 2 cited.
The screenshot below shows how the keyword “heim joint torque” appears and is cited in Google AI Overview:
To my surprise, I also received a bonus: AI Overview directly displayed our brand name—and even added a clickable link!
At this point, you might be wondering: “Were these articles all written word by word by hand?”
The answer is—not exactly. In fact, most of them were created with the assistance of ChatGPT.
Naturally, the next question many people have is:
“Wait, doesn’t Google’s algorithm treat AI-generated content as spam?”
The answer is
No—that’s a misconception.
Take a look at the screenshot below from Google Search’s official guidance on AI-generated content, which makes Google’s stance very clear:
In short: as long as the content provides real value to users, it will not be penalized simply because it was generated by AI. So yes, you can confidently use ChatGPT to assist your writing.
That said, this doesn’t mean you can blindly throw a topic at it and then copy-paste the generated text straight onto your website.
In the following sections, I’ll walk you step by step through the topic “ChatGPT Writing: How to use ChatGPT to write a GEO-friendly article?”
1. ChatGPT Writing Starts From A "Good Topic"
The very first step in blogging is choosing a topic. Many people simply write based on intuition, but if your goal is to get your article cited by Google AI Overview, Gemini, or ChatGPT, you need to be strategic: you must choose a good topic!
A good topic, even if the article itself is just average, can still be picked up and cited by AI. So, the question is: what makes a “good topic”?
(1) Choose a Low-Competition Topic
On one of our project websites, we have a blog post that has been online for many years. When we search for its seed keyword (not even a precise long-tail keyword), it frequently appears on the first page of Google search results. And when Google AI Overview was introduced, this same article would occasionally get cited as well.
But here’s the interesting part: this article only has a single paragraph of about 100 words. No headings, no structure, no images, no videos. Clearly, the reason it ranks so well is because the topic itself faces very little competition.
So how do you find low-competition topics? One effective method is to use SEMrush for keyword research to identify a “good topic.”
For example, in SEMrush, you can enter a competitor’s domain and look at their keyword list. From there, filter by keywords with low keyword difficulty (KD) but still reasonable traffic. This often reveals low-competition, high-potential topics.
The screenshot below from SEMrush illustrates keyword traffic and KD data:
For more details on how to conduct keyword research, you can refer to my guide: How to Find SEO Keywords.
(2) Topics Around User Needs
Ask yourself: What are my target readers searching for? What are their biggest concerns?
For example, if you run a SaaS website, you should pay close attention to common user pain points such as “price comparisons,” “how-to guides,” or “best practices.”
Some simple ways to find these topics include:
- Typing a core keyword into Google and checking the “People Also Ask” section — this is often a goldmine for topic ideas.
- Entering a keyword on Reddit or Quora to see which threads generate the most interaction. These are the questions real users care about.
- Searching for the core keyword on social media platforms to discover high-engagement discussions.
(3) Topics Related to Your Products or Services
Start from your own products or services and dig deeper.
You can expand core topics (your product or service) using different angles:
- Framing them as questions (what is, which are, how to),
- Providing usage guides,
- Exploring application scenarios,
- Creating comparisons or “X vs. Y” type posts.
You can also:
- Ask ChatGPT to brainstorm a list of topics related to your product/service, then filter them according to your business needs.
- Use tools like Deep Market to input your core keyword and generate extended topic ideas.
All of these can come together to form a Topic Cluster, surrounding your product or service.
In this model, the central blue node is your core topic (your product or service), while the surrounding nodes are various sub-topics. If some sub-topics are also target keywords, you can promote them into new core topics and continue extending further.
But keep in mind: this method has both pros and cons.
- Pros: It ensures your content covers the topic comprehensively.
- Cons: It can easily become an echo chamber, producing many articles that don’t truly attract user interest.
(4) Trending Topics
You can also look for trending topics:
- Use Google Trends to track how the popularity of a topic changes over time.
- Tools like Exploding Topics or AnswerThePublic can help you discover emerging questions.
Choosing trending topics often increases your chance of being cited by AI models, since they tend to prioritize the freshest content when generating answers.
2. Defining Keywords
Once you’ve chosen a topic, the next step is to define the keywords for that topic.
A keyword is simply the search term someone would type into Google or ChatGPT. When your article’s content is closely aligned with those terms, it’s much more likely to be indexed and cited.
So, what types of keywords are there? We usually work with four main categories:
- Focus Key Phrase: The “main character” of your article — the keyword you most want to rank for.
Example: For this article, the focus key phrase is ChatGPT writing. - Related Key Phrase: Words or phrases closely connected to the focus keyword, often appearing together.
Example: AI writing, SEO blogging. - Long Tail Key Phrase: More specific and longer search queries. They usually have lower search volume but higher precision.
Example: How to use ChatGPT to write a GEO-friendly blog. - Semantic Key Phrase: Not synonyms, but contextually related words that naturally appear when covering the topic.
Example: When talking about ChatGPT, semantic phrases might include prompts, generative AI, search engine optimization.
A common question is: Do all of these fall under LSI (Latent Semantic Indexing) keywords?
- Strictly speaking, Focus and Long Tail do not.
- Related and Semantic are often treated as LSI.
But in today’s large language model (LLM) world, nobody really cares whether something is “LSI” anymore. What matters is this: does your article cover the full semantic field of the topic?
So, how can we use ChatGPT to help with keyword discovery?
First, you must decide on a focus key phrase yourself. This should come from your own research during the topic selection step — not simply by asking ChatGPT.
Then, you can give ChatGPT a clear instruction, for example:
“You are an SEO expert. You are helping me write a blog titled ‘How to Use ChatGPT to Write a Blog Easily Cited by AI.’ I have set the focus key phrase as ‘ChatGPT writing.’ Based on this, please generate a list of related keywords, long-tail keywords, and semantic keywords.”
ChatGPT will provide you with an initial list. After that, you should validate it with keyword research tools such as Ahrefs, Semrush, or Google Keyword Planner to check search volume and difficulty. This way you ensure accuracy while not missing valuable long-tail opportunities.
One last but very important point: exact keyword combinations are no longer critical.
In the past, SEO often meant obsessing over exact matches. But today’s large language models are smart enough to understand semantics. As long as you fully cover the topic, even if you phrase things differently, your content can still be recognized and cited.
3. Gathering Materials (Feeding ChatGPT a “Resource Pack”)
After defining your keywords, the next step is to gather materials. Since we’re planning to use ChatGPT to assist in writing, we need to reduce the risk of it “hallucinating” by preparing reference materials in advance and feeding them into the model to give it a clear direction.
The process is simple:
- Take your target topic and search it on Google.
- Collect the top 5 ranking articles.
- Save them as Word documents.
- Upload them to ChatGPT.
- Ask ChatGPT to summarize the structure of each article.
This way, you’ll quickly get an overview of how the best-ranking content in your niche is structured.
4. Building a Clear Article Structure
With your materials ready, the next step is to create the skeleton of your article — the structure.
Remember: most people online scan rather than read every word. If your article is just a long wall of text with no hierarchy, readers will bounce in seconds. That’s wasted time for them — and wasted effort for you.
So what makes a good structure? You can use a simple first-principles test:
If you glance at the article, can you immediately understand what it’s about?
If even you find it confusing, your readers won’t stand a chance.
Some may ask: “Isn’t blogging just about SEO? Why should I care about the reader’s experience?”
Wrong.
If your article turns readers away right after they land, your bounce rate will skyrocket. While GA4 now reports this as “engagement rate,” the principle is the same: if users don’t like your content, Google won’t like your page.
In other words: a clear structure serves both your readers and your SEO. If you want stable rankings, you first need to earn your reader’s time and attention.
During the material-gathering step, we already asked ChatGPT to summarize the structure and key points from 5 reference articles. This gives us a strong foundation.
The next step is to let ChatGPT remove duplicate sections across those articles and merge the unique parts into one complete structure — the goal being to create something more comprehensive than any single one of those 5 articles.
From there, it’s time to add the “human element.” That’s right — the human element — such as:
- Inserting your own opinions, even if they’re still rough ideas (ChatGPT can help you expand them).
- Offering counterarguments, as long as you can support them logically.
- Adding supporting data, with clear sources cited.
- Pulling in company resources, documents, or case studies, and asking ChatGPT to integrate them.
- Enriching the text with images, videos, or comparison tables.
- Editing ChatGPT’s draft structure — because its outlines are often verbose or a bit messy.
By doing this, you’ll create a structure that is both AI-assisted and human-driven. At this point, your article framework is essentially complete.
Here are the key structural elements to keep in mind:
(1) Start Strong with the Introduction
The first paragraph is the entry point.
- Tell the reader what problem this article will solve.
- Pose a question or example to spark curiosity.
For example, at the beginning of this very article, I used a real case: “My article was cited by Google AI Overview in just 2 days.” This immediately grabs attention.
(2) Build Up Step by Step
The body is your core value delivery. A practical approach is to break it down into H2/H3 modules, like building blocks:
- Background/Definition: Define the concept. e.g., an H2 + short paragraph: “What is XXX?”
- Steps/Methods: Provide actionable steps or tips. e.g., an H2: “5 Methods of XXX” with 5 H3 subsections.
- Case Studies/Data: Share real-world examples. e.g., an H2: “2 Cases of XXX” with 2 H3 subsections.
- Summary/Tip: End sections with a short takeaway sentence.
This makes the content easy for readers to skim — and easy for AI models to parse.
(3) Close with Impact
Never end weakly. Your conclusion should:
- Summarize the main points.
- Give readers a call to action (comment, bookmark, share).
- Provide a next step (related links, product/service intro).
(4) Feed AI a Clean Structure
Why does structure matter so much? Because both search engines and AI models prefer content with clear logic.
When your article uses structured headings (H2, H3), short paragraphs, and lists, AI can easily identify what problem each section addresses.
Put simply: a clean structure is like pre-labeling your content for AI.
Tips
- Before writing, sketch an outline — even just a few bullet points. It will save you time and prevent chaos.
- Keep paragraphs under 200 words. Long, unbroken blocks of text are exhausting to read.
5. Enrich Your Content Formats
Text: Don’t make your article one big block of plain text. Use bold, italics, or highlights to emphasize key points and guide the reader’s eye.
Bullet Points: When you have long paragraphs, try breaking them down into bullet points for easier scanning and better readability.
Images: Add images to illustrate important points in your article. Remember to:
- Use descriptive, human-readable file names.
- Add Alt text, Caption, and Description.
- Mention the purpose of the image in the text itself.
For example, the image below demonstrates how to optimize image SEO.
Videos or GIFs: If images can explain a concept, then videos or GIFs can do it even better. If possible, create your own. If not, you can also embed or reference high-quality videos from others.
Comparison Tables: When covering a large section (e.g., an H2 with multiple H3 sub-sections), try wrapping it up with a comparison table plus a short summary paragraph.
Trust me, AI models love this format because it makes the content structured, clear, and easy to reference.
6. Start Writing with ChatGPT
After completing the steps above, you’ve already achieved the following:
Your ChatGPT thread now contains the blog topic you want to write about, so you don’t need to worry too much about “hallucinations.”
The thread also includes suggestions for related key phrases, long-tail key phrases, and semantic key phrases, and you’ve validated them with tools.
You’ve prepared various non-text assets.
You have a clear structure that includes human input, not just ChatGPT output.
Now—let’s do it!
Wait, how? Should we just ask ChatGPT to generate the whole article?
Of course not.
We’ll control the output with a series of prompts. Here’s the prompt I use:
You are a copywriter helping me write a blog post (topic: XXX). Please follow these requirements:
Topic: XXXXX
Language: English
Blog type: Technical blog
Accepted tone: Relaxed, readable, “industry friend” tone (as if chatting with the reader)
Rejected tone: Stiff, textbook-like tone
Rhetoric: Keep it simple and direct; avoid flowery metaphors
Audience: Industry buyers, procurement engineers, automotive/mechanical modifiers
Focus key phrase: XXXXX
Related key phrases: as listed above
Semantic key phrases: as listed above
Long-tail key phrases: as listed above
Word count: 1,500
Keyword placement
The opening and the conclusion must include the focus key phrase and 1–2 related phrases.
Sprinkle focus/related/semantic/long-tail keywords naturally across headings and paragraphs. No stuffing.
It’s fine to place the focus keyword in some headings, but not every heading.
Title structure
One H1 (post title)
Several H2
Add H3 where appropriate—not under every H2 (avoid rigidity)
Content requirements
Go deep—no shallow overviews
Use bullet points where it feels natural (not in every section)
Use pros & cons selectively (not in every section)
Keep each paragraph under 200 words
Use transition words to improve flow
Outline
Introduction without a heading—briefly define XXXXXX in ~80 words
First H2: XXXXXXX
Second H2: XXXXXXX
- First H3: XXXXXX
- Second H3: XXXXXXX
- Third H3: XXXXXXX
- (Adjust the number of H3s as needed)
- Add a comparison table plus a short summary paragraph of the H3sThird H2: XXXXXXXX
Fourth H2: XXXXXXXX
- (Add more H2s as needed)Conclusion: ~80 words
Do not write yet. If you fully understand, reply yes.
After ChatGPT replies, use the next prompt:
You will write my blog in segments. First, write the introduction and the first H2 section.
When ChatGPT outputs content, review everything: ensure language and word counts meet your requirements. In my experience, ChatGPT tends to use overly flowery metaphors—tell it to remove them and keep the style simple and direct. Once the first segment is correct, continue section by section.
Why segment the writing instead of generating the whole article at once?
Because AI is less likely to hallucinate on short outputs than on long ones. Segmenting may feel tedious, but it’s effective and produces more controllable results.
After you’ve fine-tuned and approved each segment, we’ll move on to further polishing the entire article.
7. Refine Your Article Further
- Identify the most important parts of your content and highlight them using quotes, bold, italics, or background highlights.
- Insert the non-text assets you’ve prepared: images, videos, GIFs, and tables.
- Give the entire article to ChatGPT and ask it to suggest 1–2 spots to insert external authoritative links. These links can be:
Supporting data references
Explanations of specific terms
Relevant related articles
For authority, prioritize websites ending in .org (nonprofit), .gov (government), .edu (academic). Wikipedia or high-authority .com sites are also good choices.
- Edit your URL into a clean, readable format that includes the focus key phrase. A great approach is to use the blog’s main title as the URL.
Write a click-worthy SEO title that includes your focus key phrase.
- Write a concise, compelling meta description. You can ask ChatGPT to generate several versions and then choose the best one.
- Write a summary or excerpt that captures the main point and place it at the very beginning of the article.
- Once the draft is ready, use an SEO plugin to run a full check. I recommend All in One SEO Pro, which evaluates keyword distribution and provides an overall SEO score for your page.
Example screenshots:
8. Final Optimization Before Publishing
Create a blog cover image. Once the article draft is done, ask ChatGPT to generate a 16:9 cover image based on the entire content. By this point, after all the back-and-forth, ChatGPT will fully understand the context of your article, so the accuracy of the image is often very high.
Set up Schema tags. This is a step many people skip. Schema markup is a form of structured data that tells search engines (including Google and AI models) what type of page you have and what the content is about. I recommend using All in One SEO Pro’s Schema feature to configure:
- Article title
- Excerpt/summary
- Cover image
- Keywords
- Publish and update dates
Once Schema is configured, you can use Schema Markup Validator to confirm that your structured data is correct.
Conclusion
And there you have it — mission accomplished. You’ve learned not only the ChatGPT writing technique, but also how to handle the entire workflow: from choosing a topic, defining keywords, gathering resources, and structuring content, to final optimization and publishing.
Of course, the real results will come from practice and iteration. Different websites, industries, and audiences may require different best practices. The key is this: create content that serves both search engines and real readers.
If you follow this process consistently, each article you publish will be more structured, more professional, and more likely to be cited by Google AI Overview, Gemini, and ChatGPT.
Now, it’s your turn to put this into practice. 🚀
