The ultimate guide: ChatGPT for the keyword research

Keyword research is a crucial part of SEO because:

  • it helps you understand what your audience is searching for in Google
  • it helps you understand how to optimize your content to rank higher in Google

ChatGPT can be used to streamline the Keyword research process. Let’s see how! 

Caution: keyword research is a very specific area that requires a deep understanding of the audience and current trends; also, it usually requires access to specialist tools. ChatGPT can be really helpful, but use it with caution.

Using ChatGPT to understand search intent

Understanding user intent is critical to ranking on Google. Google is constantly working to understand what users want to know and what kind of content will best satisfy their needs. Usually it’s confusing – for instance, Google may not be sure what to do when users type “Apple”: 

  1. Are they looking for fruits or a popular company?
  2. If they are looking for the IT company named: “Apple” – are they interested in investing in Apple stocks, do they want to know the company’s history, or find a way to buy its products? 

So, as a content creator or marketer, it’s essential to consider user intent when developing your SEO strategy. Your content should satisfy the user intent directly. By adjusting your content to the search intent of your target audience, you can improve your chances of ranking higher on Google and ultimately, driving more traffic to your site.

Obviously, ChatGPT can help! Here’s a sample query:

I have a list of queries that I pasted below.

Please show me in a table format what the user intent is for these keywords. Here is the list of keywords: {list of queries}

I asked ChatGPT what the user intent is behind such keywords as “Buy iPhone 13” or “How to clean a car”. (If you want the same prompt as I did, visit this Notion page).

ChatGPT generated the table I asked for. I trimmed it a bit for brevity.

If you want to use this method, ensure that ChatGPT can correctly detect keyword intent for the keywords in your niche.

To do this, conduct an experiment by asking ChatGPT to determine user intent for a given list of keywords, and at the same time, ask an experienced SEO to do the same.

  • If ChatGPT’s output is similar to the experienced SEO’s output, it may be worth using ChatGPT to speed up the process.
  • However, if ChatGPT produces different results, you can teach ChatGPT about your specific keyword intent to help it adjust accordingly. I explain this in more detail in the Improving ChatGPT results by teaching it chapter of the upcoming ebook.

Using ChatGPT to predict if fresh content is needed

Sometimes, your website may struggle to rank for certain queries if your content doesn’t contain up-to-date information. For example, if you have a blog about smartphone technology and only write about older phone models, you may not rank well for searches related to the latest smartphone releases.

If you want to try it, just check Google to see what pages are ranking for “iPhone 13”.

I bet you won’t find ANY page touching on iPhone 11 that ranks in the top ten results for this query.

You can use ChatGPT to help identify which queries require fresh content to improve your rankings. Here is a sample prompt to accomplish this task:

Can you create a table that displays which queries need fresh content?

Add a short explanation about your choice.

List of queries:

{list of queries}

I asked ChatGPT for its opinion on ten keywords:

I must say, ChatGPT performed impressively by providing me with a detailed explanation for its choice:

Quality control

If you want to use this method, check if ChatGPT can correctly judge the need for fresh content in your niche.

To do this, conduct an experiment by asking ChatGPT to determine whether there is a need for fresh content in the case of a given list of keywords, and at the same time, ask an experienced SEO to do the same.

If ChatGPT’s output is similar to the experienced SEO’s output, it may be worth using ChatGPT to speed up the process. However, if ChatGPT produces different results, you can teach ChatGPT to help it adjust accordingly. I explain the process in the Improving ChatGPT results by teaching it chapter of my upcoming ebook.

Checking if the query is seasonal

Some searches depend on the season: in winter people might look for “top winter coats” or “skiing spots,” while in summer they could search for “beach trips” or “summer style ideas.”

You should definitely consider seasonal keywords when doing keyword research.

How can you use ChatGPT to find out if a search term is seasonal?

Here’s a sample prompt:

For each of the Google queries listed below, give me information as to whether the traffic is seasonal or not. Present the data in a table format.

List of queries: {list of queries}

In my case, ChatGPT presented me with data in a neat table:

If you’re looking for a reliable tool to double-check your results, Google Trends is an excellent option. With this tool, you can easily check the popularity and seasonality of your keywords over time.

What if you’re not happy with the table columns?

You can ask ChatGPT to fill specific columns. As Nadeem Haddadeen, one of the reviewers for this ebook, suggested: 

Determine when users are looking for authoritative content

For some queries, Google scores authoritative sources higher than more relevant sources. It happens when Google assumes that users are looking for authoritative content.

ChatGPT can determine whether users are looking for authoritative sources. Let’s see how!

Here’s a sample prompt:

Create a table that displays the queries for which only authoritative sources are expected. Provide a brief explanation for your selection criteria. List of queries: {list of queries}

And ChatGPT gave the answer not only with a judgment but also with an explanation of its choice:

Important note: Always check if this kind of detection works properly within your niche.

It’s important to ensure that any detection methods you use are effective for your specific niche. One way to do this is by training ChatGPT.

Thanks to this training, you can be sure that the ChatGPT output is customized to your specific needs and is providing the best results possible. Give it a try!

Learn how to train ChatGPT easily. Sign up for my free “Mastering ChatGPT for SEO” ebook.

Join the waitlist

Query Deserves Diversity

Sometimes Google provides diversified results for general searches. It’s intentional.

For example, when someone searches “yoga,” they could be looking for poses, tutorial videos, or even equipment to purchase. Google then tries to satisfy these various user needs.

To help you better understand user intent and tailor your content accordingly, you can use ChatGPT to create a simple query intent detector. 

Here is a sample prompt to make a simple query intent detector using ChatGPT:

Could you please create a table that displays the queries in a case in which search results are expected to be diversified? Kindly provide a brief explanation of your selection criteria.

List of queries: {paste a list of queries}

I asked ChatGPT to classify keywords like: “how to clean a car” and “benefits of plant-based diet”. If you want to use the same input as me, copy it from the Notion page.

And ChatGPT provided me with a nice table with concise explanations. (I limited the screenshot to four results for brevity).

As always:


Determine keyword difficulty

Ranking high in search results can be a challenge for some queries, while for others, it can be relatively easy.

To access their chances of ranking high, SEOs try to detect keyword difficulty. Fortunately, there are a variety of tools available that can help you predict the difficulty of a keyword. They are pretty advanced and take into consideration many factors, such as the number of links to your competitors.

In situations where you don’t have access to these tools, and if you’re okay with an answer that’s currently FAR FROM IDEAL, ChatGPT can be a helpful alternative.

Here’s a sample query for ChatGPT:

I would like to rank on Google for the list of keywords below. For each keyword, could you provide me with its difficulty level?

Please display the results in a table and include a brief explanation of your selection criteria.

List of queries:

{list of queries}

I asked ChatGPT to measure the SEO difficulty level of five keywords:

ChatGPT gave me a nice table:

  • It’s important to keep in mind that ChatGPT has limitations: for instance, it doesn’t take into account important factors such as accurate keyword popularity or who is currently ranking in the top 10 for those keywords. This can make it challenging to accurately judge keyword difficulty using ChatGPT.
  • In many ways, relying solely on ChatGPT for keyword research is like trying to navigate using Google Maps with your eyes closed. Not a good idea, right? While it may provide some helpful insights, it’s definitely not an ideal solution.

So use this feature at your own risk 🙂

Show keyword popularity

Before trying to rank for a particular keyword, SEOs try to measure its popularity. Nobody wants to rank for keywords that can’t bring any traffic.

ChatGPT can show you information about keyword popularity.

But use it with extreme caution as this data is just an estimate and it doesn’t come from any reliable source.

If you accept the risk, here’s a sample prompt:

I have a list of queries. For each query, provide me with information about their search demand.

List of queries:

{list of queries}

ChatGPT quickly gave me the expected answer:

Again, you should be extremely careful with this feature as ChatGPT made up the statistics.

When you ask ChatGPT for the methodology, it tells you that the data comes from Google’s planner. No, it doesn’t.

I explain this in more detail in the ChatGPT Lies chapter of my upcoming ebook.

Classifying a keyword as long tail or short tail

It is important to know if a keyword is long tail or short tail because it can impact both your SEO strategy and the likelihood of ranking for that keyword.

  • Short-tail keywords are typically shorter phrases that are broad and general in nature. They have high-search volume and high competition, making it more difficult to rank for these keywords. For example, “shoes” or “coffee shop” are short-tail keywords.
  • Long-tail keywords, on the other hand, are longer phrases (usually four or more words) that are more specific and targeted. They have lower-search volumes but lower competition, making it easier to rank for these keywords. For example, “men’s running shoes for flat feet” or “organic coffee shop in Brooklyn” are long-tail keywords.

ChatGPT can help you detect whether a keyword is long tail or short tail. Here’s a sample prompt:

Please classify each of the queries listed below as either “long tail” or “short tail”, and provide a brief explanation for each decision in a table format.

Here’s a list of queries I want to check:

{list of queries}

I used the same sample keywords as in the previous prompts:

ChatGPT provided a detailed explanation of its choice:

As always:

  • Check if ChatGPT produces data that is useful for your niche.
  • If it doesn’t produce useful results, try to teach the model what a long tail is and what it isn’t. And then check again if the model performs better. I wrote a separate chapter on ChatGPT’s custom training in my upcoming free ebook about ChatGPT for SEO.

Calculating keyword density

I never really believed in the idea of keyword density, but I know there are still some SEOs who use it. So why not make their lives a bit easier with ChatGPT?

Just give ChatGPT a sample of your content and the keyword you’re targeting, and it will calculate the keyword density for you. No more tedious manual work! Here’s a sample prompt:

For the following text, show the most common keywords, including keyword density.

Here’s the text: {paste the article}

I asked ChatGPT to generate keyword density for an article about marketing:

If you want to use the same prompt as I did, copy the prompt from the Notion page.

ChatGPT gave me the desired output:

Using ChatGPT for keyword categorization

When you conduct keyword research, you’ll often end up with a large list of keywords. It’s practical to group them logically to make sense of them.

  • You can do this task manually, but it can be tedious. On the other hand – if done by a skilled worker, it can yield the most accurate results.
  • Regardless, you can try using ChatGPT to see if it’s a good solution for your needs.

As an experiment, I gave ChatGPT 100 popular SEO-related keywords to categorize. The prompt I used:

I have a list of 100 popular keywords related to SEO. Categorize them.

Here’s the list of keywords:

{list of keywords}

If you want to use the same prompt as I did (including sample keywords), use this Notion page.

In my case, I would say that the keywords were categorized pretty well. You can see the output on the same Notion page.

However, as always, it’s important to test it out in your own niche.

If necessary, you can train the ChatGPT model by providing it with suggestions and see if it improves its results.

Generate a list of popular questions related to a given topic

When researching, it’s important to understand what your audience needs and what problems they want to solve.

This can be achieved by talking to your target audience. A useful alternative is to check the popularity of questions related to your subject using tools such as SEMRush or Ahrefs.

However, if you don’t have access to these tools or don’t need high accuracy, ChatGPT can be a helpful option.

Here’s a sample ChatGPT prompt:

Generate a list of popular questions related to {topic} 

For each of the queries, try to predict if users expect authoritative or fresh content. Format the data in a table.

I asked ChatGPT to generate popular questions related to SEO:

With regard to the output, ChatGPT returned a list of questions, with some additional notes about whether readers would expect to have authoritative, or fresh content.

I was impressed by the results produced by ChatGPT this time.

However, when you want higher accuracy, always use tools that estimate the real number of searches, such as Google Keyword Planner, SEMRush, or Ahrefs.

Teaching ChatGPT to think like Google

Making use of ChatGPT’s pre-trained models is pretty awesome! You can get help with tasks like content rewriting and generating topic ideas.

But the real magic happens when you fine-tune ChatGPT for your specific needs.

In this section, we’ll explore a really cool idea. What if we use Google’s criteria for determining high-quality content and see how our content stacks up?

We’ll be using the 20 questions recommended by Google for websites hit by a core update. These questions are a great starting point for assessing the quality of your content. You can find them here.

So I asked ChatGPT to act like a content quality reviewer.

Sample prompt:

I would like you to evaluate my content based on specific criteria. 

  • Then, I will provide you with a text, and you can create a table where you assign scores for each of the criteria. 
  • Please ensure that you provide an explanation for each decision regarding the scores assigned. If you do not have enough data, please use a question mark instead of making up a response. Is that clear? 

If so, here is a list of criteria I would like to evaluate my text on:

{list of criteria}

If you want to copy my prompt, go to the Notion page.

Obviously, ChatGPT accepted the challenge!

I then fed my ChatGPT evaluator with a content fragment from one of my articles: Ultimate Guide to JavaScript SEO.

ChatGPT gave me a detailed explanation of its judgment in 20 various criteria. That’s huge!

A few important notes:

  • Have in mind you can evaluate your content against any criteria you want. Just tell Chat what these criteria are. As I keep repeating in this ebook, the sky is the limit 🙂
  • Always validate the output!
  • Play with the queries. For instance, ask ChatGPT to be more objective and strict, or the opposite – more forgiving. It depends on your creativity.
  • In the initial prompt, I asked ChatGPT: “Is that clear?”. This question is essential because, rather than saying, “I’m not sure, can you clarify?”, ChatGPT tends to provide a response even when it’s uncertain. By asking if the question is clear, it increases the likelihood that ChatGPT has correctly understood the prompt. If not, ChatGPT will ask for clarification.
  • Final word: don’t treat ChatGPT like an oracleAssuming Google uses the criteria listed above, it can still have different interpretations. Think of an analogy with two teachers. When two teachers grade the same test, they might give different scores. This is because each teacher has their own biases, preferences, and understanding of the material. Similarly, Google and ChatGPT may judge websites differently because they have different algorithms.

Getting useful statistics related to a given subject

Statistics are useful for written articles and/or business meetings, as well as for many other reasons:

  1. Including statistics can help establish credibility and authority in your writing by demonstrating that you have researched and analyzed relevant data.
  2. Statistics can be persuasive because they provide concrete evidence to support your arguments and claims. This can help convince readers to take action or change their opinions.
  3. Statistics can be attention-grabbing and interesting to readers, increasing engagement with your content.

You can ask ChatGPT to generate some interesting statistics. But be extremely careful with ChatGPT output. I will explain it in more detail below.

But first, let me show you how to collect some statistics related to a given subject. Here is a sample prompt:

Please show me 10 interesting statistics about {topic}. 

  • Please include the source.

In my case, ChatGPT gave me pretty interesting statistics. I didn’t even know that email marketing has such a huge ROI (return on investment)!

ChatGPT can lie about statistics

There is a famous saying: “Lies, damned lies, and statistics”.

I would rephrase it: Lies, dammed lies, and statistics made up by ChatGPT.

Here is why:

  • You have to be extremely cautious with this type of output. As I showed in other parts of this ebook, ChatGPT can eagerly make up data. Sometimes it goes to the extreme. For example, when prompted to “Show me popular articles on the topic of International SEO,” ChatGPT provided me with five article titles, all of which were incorrect


So I recommend following a simple, three-step process:

    1. Ask ChatGPT to provide you with statistics on a given topic.
    2. Select which statistics are interesting to you.
    3. Verify the statistics by checking them in Google (or Bing, in case you’re reading this in 2026 and Bing is now the most popular search engine ;).

Asking ChatGPT to present famous quotes about a given subject

As I mentioned earlier, putting quotes into content makes it more interesting and adds credibility to it.

You can easily ask ChatGPT for quotes on a given topic. Here’s a sample prompt:

Please show me {number of quotes} interesting quotes about SEO. 

Include the source.

I asked ChatGPT about interesting quotes about SEO:

It’s important to be cautious about the answers ChatGPT provides. Sometimes ChatGPT may make up answers and even attribute quotes to the wrong people.

For example, let’s take a look at this quote: “SEO is not about optimizing for search engines. It’s about optimizing for people.” attributed to Bryan Eisenberg, co-founder of (source:

When I searched for this quote on Google, I couldn’t find any results. This is a clear indication that ChatGPT made it up. Bryan Eisenberg never said this. Moreover, it seems that the quote was posted on a list of top SEO experts published on SearchEngineJournal, but there are actually no quotes on that page.

I didn’t even bother checking the remaining four quotes, but I suspect the story may be similar.

To test ChatGPT in other niches, I asked it for famous quotes about Conor McGregor, one of the most popular mixed martial artists and all the quotes turned out to be made up. However, when I asked it for quotes about traveling, the answers seemed totally legit.

So, always use ChatGPT with caution and double-check the quotes!

Final note on using ChatGPT for keyword research

Yes, ChatGPT can be helpful for Keyword Research.

Remember that this feature of ChatGPT is a bit experimental and should be used as input for brainstorming, not for the final output.

For instance a lot of SEOs use ChatGPT to generate a list of related keywords, with prompt like this: Generate a list of popular keywords related to {topic}. If you use this, make sure that you don’t rely solely on ChatGPT.

My experience is that ChatGPT usually gives you keywords with zero popularity. So check keywords generated by ChatGPT in specialized tools such as Ahrefs, SEMRush, or Google Keyword Planner.


Sign up for my free ChatGPT for SEO ebook

Did you like this content? Wait for my ebook! It’s coming in June 2023 and contains so much more information about the ChatGPT for SEO use. And I am giving it 100% free! 

Join the waitlist

Hi! I’m Bartosz, founder and Head of SEO @ Onely. Thank you for trusting us with your valuable time and I hope that you found the answers to your questions in this blogpost.

In case you are still wondering how to exactly move forward with fixing your website Technical SEO – check out our services page and schedule a free discovery call where we will do all the heavylifting for you.

Hope to talk to you soon!