If you are wondering what ChatGPT is (which I highly doubt), it’s the new chatbot that everyone is crazy about, from students to professors to SEO experts! The entire world is talking about it.

Chat GPT concept. Business person chatting with a smart AI  using an artificial intelligence chatbot development. Artificial intelligence system support is the future
Chat GPT concept. Business person chatting with a smart AI using an artificial intelligence chatbot development. Artificial intelligence system support is the future

This chatbot seems to have all the answers to every question you can think of, and what’s more? It does it in a conversational way that makes you feel like you are talking to a genius coursemate or an intelligent friend.

When it rolled out, OpenAI made it publicly available to test out its functionalities. Now, this is where the SEO pros didn’t go to sleep. ChatGPT was made to carry out several SEO tasks that would usually make the whole process of optimization tedious and time-consuming.

The result? Astounding!

In this post, I will talk about the several ways ChatGPT has been tested for SEO, the viability of using it to write optimized content on your website, and also consider if this will be a good SEO practice to imbibe in your strategy for 2023.

Now that ChatGPT has gone pro, is it a worthy investment, or should we stick to the traditional method of SEO?

You might agree with me on many things at the end of this post.

What is ChatGPT?

ChatGPT (Generative Pre-trained Transformer) is a chatbot developed by OpenAI in November 2022 to optimize language models for dialogue.

It is a large learning model that uses artificial intelligence (AI) and data collected from several sources across the web to interact with users conversationally.

The incredible thing about this chatbot is that it can answer follow-up questions from users, acknowledge mistakes, and challenge incorrect inputs.

ChatGPT interface screenshot from OpenAI's official page.
Screenshot from OpenAI’s official page.

Here’s the official description from OpenAI. I highlighted the main points:

We’ve trained a model called ChatGPT which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer follow-up questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.”

Many predict that ChatGPT will take over Google and most search engines but maybe not directly.

True, ChatGPT has shown a different way searches can be done, and while Google isn’t there yet, there have been changes in how the search engine replies to various queries.

ChatGPT is excellent, but taking over Google, I’m not sure it’s there yet.

How ChatGPT was trained

Previous models like GPT-3 and Codex were trained using the InstructGPT, and while these were successes in their own rights, they had some amount of “harmful and untruthful” outputs.

ChatGPT is a fine-tuned version of a model from the GPT-3.5 series. It was trained using Reinforcement Learning from Human Feedback (RLHF). Although it uses the same methods as the InstructGPT, it has slight differences in how data is collected.

“ChatGPT is a sibling model to InstructGPT, which is trained to follow an instruction in a prompt and provide a detailed response.” – OpenAI.

Screenshot describing the difference between ChatGPT & InstructGPT
Screenshot describing the difference between ChatGPT & InstructGPT
Source: OpenAI

According to OpenAI, there were three crucial steps in training ChatGPT:

1. Supervised learning.

Human AI trainers played both sides of a conversation (user and AI) where they sampled a prompt (query), then demonstrated the desired output from model written suggestions.

This dialogue was then mixed with the InstructGPT and transformed into a format.

2. A reward model.

To create this reward model, data needed to be compared, so the collected comparison data, which included two or more model responses from the initial dialogue, was ranked by quality.

A labeler ranks the outputs, and this comparison data was used to train the reward model.

3. They optimized a policy against the reward model.

The reward model was optimized using the Proximal Policy Optimization (PPO) reinforcement learning algorithm. 

Screenshot showing how ChatGPT was trained.
Source: OpenAI

How ChatGPT works

ChatGPT is a large learning model, which means that it is practically trained to learn how to respond to different queries based on a large amount of data.

Unlike some might think, ChatGPT isn’t particularly trained to answer a question in a certain way, but the amount of data it has enables it to predict how it should answer a question. 

In other words, this model is trained to act independently and cognitively based on its data.

The higher the amount of data it is trained on, the more tasks it can carry out.

The incredible thing about large learning models like GPT-3 & GPT-3.5 is that increasing the amount of data they have can also lead to unexpected and surprising abilities. 

For example, Stanford University wrote about how GPT-3 could translate a text from English to French, although it wasn’t directly trained to perform that task. It even outperforms models that were explicitly trained for those tasks.

While this is impressive, it can also be an Achilles heel because it implies that this model—ChatGPT- relies entirely on the information it has. 

We’ve seen this happen in some of the user testing done. For instance, ChatGPT cannot provide information beyond 2021.

What happens when you input your query in the ChatGPT?

The AI chatbot gathers all the data directly or vaguely related to what you’re asking; it guesses your intent and gives you an answer it believes best satisfies your query.

You can also follow up with another question; it will understand what you mean, unlike other search engines and virtual assistants.

Sample Query
Source: Open AI

We must understand that the introduction of ChatGPT is a glimpse into what the future entails. 

We are just scratching the surface of this AI wonder! One thing I love about technology is that it keeps getting better, and that’s what’s to be expected of ChatGPT.

What can you use chatGPT to do?

ChatGPT can perform several tasks, including blog writing, coding, and even poetry. It can talk about your country’s political system and electoral issues and even help you write your final papers.

In fact, ChatGPT passed a Wharton MBA exam, and this is just the beginning. The NYC Department of Education has already banned the use of ChatGPT for fears of it being used to pass exams.

Here are several ways you can use ChatGPT, including:


You can translate your text into any language using this chatbot. This includes your keywords, entire blog post, and maybe your novels!


This is bliss for programmers and bloggers. You don’t need to outsource programmers to help you write your codes for off-page SEO. Just ask ChatGPT!


Most examples I’ve seen online have been basic calculations, but if this wonder bot can code, it should also do more complex mathematical problems.


You can write entire content using this bot, from blog posts to social media content. You also get to input points you want, the way you want them, and even create bullet lists.


This chatbot does incredibly well if you want a perfect summary for your content. You can add factors like length, tone, and reading level.

Sentiment analysis

When you’re done writing content, you can analyze its sentiment, which helps you know if you are meeting your audience’s intent and the purpose of your content.

Limitations of ChatGPT for SEO.

ChatGPT has its fair share of limitations, like every newly released technology, especially when everyone wants to use it for everything possible.

Here are some current limitations of ChatGPT:

Answers might be logically plausible but unrelated.

This is one of the major issues with ChatGPT because it seems that fixing it is challenging for the developers. According to OpenAI, they cannot train the model to be more cautious because it will avoid certain questions even though it can answer them accurately.

Also, supervised training misleads the model because it can only answer a query based on what it knows rather than what the demonstrator knows.

To work around this limitation, you should explicitly state how you want your answers, rephrase your queries, or avoid queries that might need updated answers beyond 2021.

Now that ChatGPT has gone pro, is it a worthy investment, or should we stick to the traditional method of SEO?

It can misunderstand prompts.

The model is susceptible to tweaks in phrasing or asking the exact prompt multiple times.

It means you can ask a question, and the AI might claim not to know the answer, but when you rephrase even slightly, it might be able to answer you correctly.

This limitation must be addressed because if incorporated into search engines or voice assistance, it might affect search results, especially voice searches where phrasing can be in vernacular.

It guesses the user’s intent instead of verifying.

When unsure of what your input means, this chatbot guesses your query’s intent and tries to answer it based on its conclusion rather than on what you want, unlike most ideal chatbots that will ask clarifying questions.

Sometimes, its conclusion might be correct, but other times, it might not be.

This will be problematic for users who need instant and accurate answers.

Cannot generate content on ‘inappropriate’ requests.

Based on trying to curate the type of prompts and responses on this model, the Moderation API was used, but it might not work as perfectly as intended.

The model can biasedly judge specific contents as unsafe and safe (false positive and false negative outputs).

It has in-built biases.

Another limitation is the in-built biases because of the training data used in developing this model. Trainers prefer longer answers because it looks more comprehensive and accurate, which might be misleading or contain repetitive and overused phrases.

It cannot browse the internet!

Unlike traditional search engines, ChatGPT cannot browse the internet, and its answers are based on the data it has been trained with; this is one of the reasons why its content abilities are not up-to-date.

The just-completed world cup or the newly released Google algorithm (2021, 2022) is an example.

OpenAI clearly states that regular updates will be made to the model to rectify most of its limitations and any other problematic outputs that users will observe during usage.

How to use ChatGPT for SEO.

ChatGPT can be used for SEO because it has quality responses. Here is how you can use ChatGPT for SEO:

Content Idea Generation

You can simply ask for ‘content ideas relating to (a specific keyword),’ and you will get a series of ideas and queries relating to that keyword.

Here’s what you need to understand about ChatGPT. It uses tons of data from the web, and for each query you ask, it searches through this data and gives you what you need.

For example, if you ask “a content idea relating to image SEO,” it will search through its data and give you a list of questions that are being asked about Image SEO. You can filter through these results and streamline them to a target audience or competition.

Generating content ideas - ChatGPT

Keyword Research

When using ChatGPT for keyword research, you can give it one or two focus keywords and ask for related keywords.

It will effortlessly give you a long list of related keywords based on its data. It can also determine if a particular keyword will be more competitive than another.


Keyword Intent

You can take your keyword research further by asking the chatbot to segregate each keyword based on searchers’ intent. For example, transactional keywords are different from informative and commercial keywords.

With this information, you can accurately tailor your content to meet the searcher’s intent.


Keyword Translation

For websites that cater to a large audience with different native languages, this comes in handy.

With ChatGPT, you can translate this keyword to any language you want, which helps you rank for foreign sites and audiences.

Keyword research, intent, and translation can be done as follow-up queries, so you don’t even have to move between tabs or applications.


However, I can’t say that ChatGPT is a more accurate keyword research tool than Ubersuggest and all other tools out there because there is yet to be a direct comparison.

Title Generation

I was teaching some interns how to write optimized posts, and when we got to title optimization, we all got exhausted. Optimized title generation is challenging because you want to avoid ending up with something dry or bland.

Using ChatGPT, you can generate several interesting topics that captivate any reader.


All it needs is a title and some form of context (optional), and you’ll have a sweet list you can use. You can also copy your content into the chatbot and have it generate a title based on what you have written.

Even if you don’t like any, it can serve as an idea bank for you to come up with more captivating headlines, but I can guarantee to a large extent, that you will see an SEO-worthy title from the list.

Meta Description

You can easily create a meta description for your content by pasting a part or all of your content on the chatbot and requesting a meta description. You can also ask for the exact tone you want your meta description to be and include as much detail as you wish.


Outline Generation

If I can use ChatGPT for SEO, I’d use it to generate an outline for my post. The ability of ChatGPT to compare search queries and know the user’s intent is vital. It can help you create an outline and, ultimately, content that covers all the searcher wants to know, needs to know, and what they didn’t even know they needed.

Starting from keyword intent, you can ask the chatbot to create an outline for a specific topic or keyword.


Content Writing

Yes, you can use ChatGPT to create good content in any niche. You can specify the number of words and the intent and even include a compelling CTA.

However, I recommend against using AI to write web pages or posts because they often lack originality and may contain subtle errors.

For example, OpenAI developers have indicated that ChatGPT might be able to perform specific tasks and give concise answers, but it can also provide ‘plausible but nonsensical answers.’



With the GPT-3 model incorporation, ChatGPT can improve your writing style and quality. It can automatically correct your grammar and spelling mistakes without you having to lift a finger.

I have a bias on this, though. It is difficult to trust an editing tool to proofread for me because there are other factors to consider, including sentence flow and that human insight, which an AI tool cannot just have.



ChatGPT can help you paraphrase your content to give it a different feel and tone. For example, you can turn professionally toned content and make it more fun or easier to understand.

Interestingly enough, ChatGPT is designed so that it cannot plagiarize itself, which means its output for the same query might differ in words but have the same message. 

“Go to the door” and “move toward the door” are different phrases but have the same meaning—this is what ChatGPT can do.


You can summarize the entire content using this chatbot. All you have to do is copy the post or content and ask it to summarize it for you in the exact way you want (of course, you have to specify this instruction).


FAQ Generation

Frequently Asked Questions are essential to all SEO strategies, and ChatGPT makes it easy for any SEO expert to gather related FAQs around a particular topic easily.

For example, if you want to publish content on voice search, you can ask “related questions on ‘how to do voice search optimization’” and you will get instant answers.

You can filter these queries and select the ones to include in your content. It might also give you more FAQ ideas.


Statistics Bank

When writing optimized content, you want to include statistics. For example, if I am writing a post on ‘content marketing,’ I want to have as many statistics as possible, which might require me to spend considerable time surfing through several posts. With ChatGPT, I can have all these statistics on a single page by asking for ‘content marketing statistics.’


Sentiment Analysis

If you ever wonder what emotion your reader feels when they read your content or what kind of tone your writing carries, you can analyze it using ChatGPT. 


Code Generation

When using ChatGPT for codes, you must carefully examine its output to spot mistakes and errors. Regardless, you can use ChatGPT for several SEO tasks involving simple codes.

For example:

Hreflang tags

Hreflang tags target audiences in specific parts, usually those with a different native language. It is typically used to incorporate multilingual SEO on a website such that your website displays in the native language of the user.

ChatGPT can help you automatically generate hreflang tags you can add to your website.


Redirection codes

You can generate a redirection code snippet using ChatGPT to redirect users to an updated page on your website. You must include all the correct information, including the original and new locations.


Robot.txt codes

These codes indicate the web pages on your website you want or do not want Google to crawl. You can easily generate a robot.txt file.


Schema MarkUp

You can generate the schema markup code of any information you provide for technical SEO optimization. Just make sure you provide exact details and appropriate context.

Schema Markup - ChatGPT

Can Google identify ChatGPT content?

Apparently yes!

Cryptographic watermarking by OpenAI researchers makes its contents detectable.

An article by TechCrunch explains that OpenAI is working on “statistically watermarking the outputs of a text (AI-generated content)” ~ Scott Aaronson.

An OpenAI engineer named Hendrik Kirchner developed a working prototype, and they hope to incorporate it into future models. The goal is to make it hard for users to pass off AI-generated content as humans.

This watermark will be an unnoticeable secret signal and a relief to educational systems worried about the extent of use of ChatGPT by students. This might come as a mixed feeling in the SEO world, but it is to be expected. 

Most, if not all, AI-generated content is detectable; content translated from another language and even content that uses anti-detection algorithms are included. It is expected as well that content generated by ChatGPT will also be detectable. 

On January 31st, 2023, OpenAI also revealed on their Twitter handle that they are working on an AI Classifier that will differentiate human-written texts from AI-written texts. The post also revealed that they are releasing an initial version so as to collect feedback and improve the system.

This Classifier is, however, not reliable as it can give false positives and negatives and should not be used as a primary decider. It is especially limited to long-form content (>1,000 words) and cannot identify contents that are AI-generated but has been edited.

Should you use ChatGPT for SEO?

Yes, but you should avoid using it to write your content. Instead, use it as leverage to help you save time and create optimum content.

I know you want to create content faster, but you should really avoid using ChatGPT to write your content, especially if you are uploading it on Google.


Because Google has explicitly stated that it views content created using only AI as spam, and it clearly violates Google’s guidelines.

Here’s what Google says:

Spammy automatically generated content is content that has been generated programmatically without producing anything original or adding sufficient value.”

With this, the issue isn’t that it’s automatically generated; the problem is that there is no added value or originality. It means you can safely use ChatGPT for any SEO process, but always make sure the human input is more than that of the AI instead of copying and pasting content.

Google algorithm can already identify AI-generated content; you don’t want your website penalized because of fast results. 

With talks of ChatGPT having a cryptographic watermark for its AI-generated content, you can be sure Google will not bypass any AI-generated content, no matter how well-written it is.

Is ChatGPT Plus worth it?

On the 1st of February 2023, OpenAI revealed on its Twitter page that they are piloting ChatGPT Plus, which is a subscription plan that offers a faster response plan and reliability during peak hours.

This ChatGPT Plus plan rolls out at $20/month for each user. The free version is still available, but with the premium version, you get priority access to new features, improvements, and faster responses. You don’t have to wait until there’s a free space.

Currently, it is only available to users in the United States, but it will be made available in other regions soon enough.

Is it worth it? 

That depends on you and what your needs are. There is no doubt that ChatGPT will make SEO easier, more accurate, and less time-consuming. The question is, can you do without it?

For SEO agencies with lots of customers and not enough SEO writers, consider using ChatGPT Plus to make work easier for your writers and meet deadlines easier and faster. However, since ChatGPT is still improving, I’d say wait till its SEO effectiveness has been tested and trusted before committing. 


This is just GPT-3; imagine what GPT-10 will do!

ChatGPT is changing how we view search engines, and the possibilities are endless. However, using it for SEO should be done cautiously and only as an aid.

You can keep experimenting with several ways ChatGPT can be used, and you may see a great use for it that will change the SEO world forever!

What do you think of ChatGPT for SEO? Is it something we should embrace or not?

READ ALSO: The best free SEO tools to optimize your websites – SARMLife 

Leave a Reply

Your email address will not be published. Required fields are marked *

You may also like