OpenAI’s ChatGPT: a revolutionary tool for enhancing SEO with AI meta data

Our previous guide covered ChatGPT, an Artificial Intelligence (AI) tool developed by OpenAI. We demonstrated in a detailed step-by-step guide how ChatGPT offers an innovative solution for business owners, marketers, content creators, and SEO professionals to create well-written, SEO-optimised content.

However, the GPT-3 language processing model provided by ChatGPT offers so much more and can be an invaluable aid for those in need of ideas, inspiration, and support. It’s important to note that GPT-3 has so many other incredible applications, including coding, translation, chatbots, and much more. In this guide, we’ll show specifically how GPT-3 and ChatGPT can be used as a powerful AI meta data tool to assist in the creation of meta titles and descriptions for SEO.

GPT-3: example asking for a definition and code

ChatGPT: an example using a simple prompt

If you are unsure what these tools are, please see here for more information on GPT-3, ChatGPT and the differences between the two.

How can GPT-3 and ChatGPT be used to create SEO meta data?

With the correct usage, both tools have the ability to improve SEO productivity enormously and can be used as a valuable resource for SEO professionals seeking new ideas.

With over 10 years of experience in writing SEO, we have a lot of experience in writing meta titles and des and identifying the good, the bad, and the ugly. We’ve seen it all and can honestly say that both GPT-3 and ChatGPT are powerful tools for creating meta data if used correctly.

Using natural language processing, machine learning, and other advanced technologies, GPT-3 can be used for different kinds of purposes One such application is analysing keywords to create well-optimised SEO meta data. This was touched upon in our guide to using ChatGPT for SEO, but this didn’t explore the full benefits relating to meta data creation. For example:

• Analysing keyword data using Google Analytics and Google Search Console CSV exports
• Generating new keyword ideas
• Revising keywords to create smaller lists
• Creating meta titles and descriptions
• Varying meta titles and descriptions based on prompts, style and personal preference
• Creating meta data in bulk

Before going any further, here are a few definitions explaining what meta data is, specifically when it comes to SEO. Please skip these sections to understand more about GPT-3 and how this can be used.

What is SEO meta data?

SEO meta data is a form of data specifically used to help search engines understand the content of a website. It includes a meta title tag, which is the title of the page, and a meta description, which is a brief summary of the page‘s content. These meta tags are contained within the HTML code of the website and not immediately visible to users, but they can be seen in search results, e.g. 

What is SEO meta data: shown as HTML code

Meta data is a critical component of SEO, as it helps search engines understand the relevance of a website and its content.

What is a meta title tag?

The meta title tag provides a brief overview of the page‘s content, and it is displayed in search engine results as shown below. This helps users quickly identify the topic of the page and determine which pages to click on. The meta title tag is also considered to be the most important HTML tag for SEO and inclusion of keywords.

What is a meta title: shown as a Google search result

What is a meta description?

The meta description also helps search engine crawlers and users quickly identify the topic of the page. It is also displayed in a search engine‘s results as shown below, helping to increase visibility and drive more traffic to the website. 

What is a meta description: shown as a Google search result

Keyword Opportunities Using ChatGPT with Google Analytics and Google Search Console

We won’t cover this in detail here as we’ve already created two comprehensive guides jam-packed full of useful prompts to generate keyword ideas using ChatGPT combined with Google Analytics and Google Search Console data.

Here are some of the most relevant sections:

• Identifying Trends, Gaps & Growth Opportunities using Google Search Console
• Identifying High-Value Keywords with Google Analytics
• Using ChatGPT to conduct keyword research and identify high-value keywords
• Analysing keyword performance and identifying optimisation opportunities
• Uncovering new keyword opportunities and expanding keyword strategies

AI meta data creation using ChatGPT

Here are some examples showing how ChatGPT can be used to create AI meta data. It’s important to stress that these are just examples to show what’s possible. There are no right or wrong ways to ask ChatGPT to help, but equally, there’s no substitution for experience and understanding how to get the best out of this powerful tool.

For the sections below, we’ll use the example of an eCommerce store selling flowers.

Generating and analysing keyword ideas

As well as using data from tools such as Google Analytics and Google Search Console, you can go straight to ChatGPT and just ask for keyword ideas.

Prompt 1: Please provide a list of keyword ideas for my….

Ask ChatGPT for keyword ideas

Prompt 2: Provide the most relevant and popular keyword (“primary keyword”)

An AI-generated primary (focus) keyword

Prompt 3: Provide 3 other relevant/popular keywords (“secondary keywords”)

An AI-generated list of three secondary keywords

For many novice business owners or website administrators, these simple prompts (or similar) will be enough to get started. However, as any SEO professional will know, a tool like Google Keyword Planner can be invaluable for checking and prioritising keywords according to the search volumes.

In this case, let’s see how good the initial list and subsequent responses from ChatGPT were. The data below shows that all AI-generated keywords have a search volume. This in itself is impressive as it means they all have the potential to get seen and generate clicks.

Evaluate ChatGPT keywords and search volumes using Google Keyword Planner

The next factor to consider is whether ChatGPT recommended the best primary and secondary keywords. There’s no right or wrong answer here, but most SEO professionals will use their experience to guide them. If we break them down, the four keywords overall are a good selection as they cover different angles and they are all relevant to the eCommerce shop as a whole, e.g.

1. Online
2. Flower Shop
3. UK
4. Fresh Flowers
5. Flowers Delivery
6. Florist

If we were to make a decision without using AI, the list would look a little different:

• Primary – Online Florist
• Secondary – Same Day Flower Delivery, Fresh Flowers, UK Flower Shop

The differences are subtle but we have taken a more search volume-led approach and have covered slightly more variations:

1. Online
2. Florist
3. Same Day
4. Flowers Delivery
5. Fresh Flowers
6. UK
7. Flower Shop

As well as keyword ideas, ChatGPT is a powerful tool for many kinds of content and marketing ideas. For more inspiration, see our guide here.

Creating meta titles and descriptions

At this stage, there does appear to be a difference between the quality of results when using ChatGPT (https://chat.openai.com/chat) vs the GPT-3 playground (https://beta.openai.com/playground).

Example showing the meta data results using ChatGPT

Prompt: Generate an SEO-friendly meta title and description for the 4 keywords listed above

Using ChatGPT as an AI meta data tool

The above AI meta data is acceptable but identifying good meta data can only be done with experience. A few general rules to consider are:

1. Are all keywords present in one form or another? – yes
2. Do keywords appear to be repeated necessarily? – yes (but this is bad practice)
3. Will the meta data attract the visitor’s attention – maybe (but it’s not great)
4. Is the meta data too long or short – too long, you can check using a tool like https://totheweb.com/learning_center/tool-test-google-title-meta-description-lengths

A rough score out of 10 for the above would probably give a 4/10 result. The main reasons are that the keywords are repeated in a way that we would not consider to be SEO-friendly (the word “flower” for example is mentioned three times in the title). There are no attention-grabbing calls to action, and whilst we don’t personally stick to a strict length, both the title and description are too long and this is confirmed below:

SEO meta data checker: Test title and description pixel length

Example showing the meta data results using GPT-3

Prompt: Generate an SEO-friendly meta title and description for the 4 keywords listed above

Using GPT-3 as an AI meta data tool

The SEO meta data generated above is definitely more to our liking, but why:

1. Are all keywords present in one form or another? – yes/mostly
2. Do keywords appear to be repeated necessarily? – no
3. Will the meta data attract the visitor’s attention – it’s better  
4. Is the meta data too long or short – adequate length

A rough score out of 10 for the above would probably give a 6/10 result. The main differences are that there is less repetition and both the meta title and description are shorter. On the downside, some keywords and personalisation are missing, e.g.

• The ChatGPT versions included the words “Delivery” and “Florist” in the title – these are missing above; however, GPT-3 captures the user intent correctly as the words “Buy” and “Online” imply delivery and we can probably assume Google understands that “Flower Shop” and “Florist” are the same thing
• ChatGPT uses phrasing in the description like “Hand-picked, hand-delivered and guaranteed”, “smile on your face”, and “beauty of nature” – these are great emotive keywords and provide a more human feel, which will help to attach clicks.

Example showing a hybrid/human-generated version

In a nutshell, both of the above examples are ok and they are better than having no SEO meta data at all. If we were to take both of these results one step further and add our own personal touch, the meta data would look more like this:

• Title: Buy Fresh Flowers Online | Flower Shop Delivery UK (maybe include “Florist” somewhere)
• Description: Shop for the freshest flowers in the UK, delivered from our online flower shop. Hand-picked, hand-delivered and guaranteed to put a smile on your face. Enjoy the beauty of nature from your home.

In reality, having to manually create meta data defeats the object of using AI and it’s unlikely that both tools would be used. Our suggestion would be to use the GPT-3 playground version here (https://beta.openai.com/playground) and experiment with different prompts to see what works best for you.

Varying meta titles and descriptions based on prompts, style and personal preference

With some additional prompts, we can get GPT-3 to automatically create AI meta data that is more like our hybrid/human-generated version. In the example below, we gave  two additional prompts:

Prompt 1 – Revise the above but make it more emotive and compelling whilst keeping the keywords

Prompt 2 – Revise the above but make it shorter 

Use GPT-3 to revise meta data and make more emotive

We can fairly safely say that GPT-3 has achieved both very well, and would probably score around 7-8/10 based on our rules above.

Bulk SEO meta data creation using GPT-3

We’ve covered a fair bit of ground above, showing how ChatGPT and GPT-3 can be used to come up with good keyword ideas, create SEO meta data and revise the meta data to achieve a style and balance that is good for SEO.

When optimising the occasional page this is great, but what about working at scale and bulk meta data is needed? Unfortunately, there’s no way to automate this entirely. We wouldn’t recommend this even if it were possible.

However, it is possible to work on more than one page at a time. Here is the first example using a simple list of pages:

SEO meta data created in bulk using AI

The AI meta data generated here is very good given it has been completely automated with a single keyword provided per page. It’s important to remember that GPT-3 remembers the previous questions and answers, so will often use this information when constructing new results.

As demonstrated in the above section, we can also ask GPT-3 to vary the results in bulk by providing different prompts, such as “vary more”, “add more”, “include”…

The process we have trialled with great success is to generate meta data in bulk by providing more information in the initial prompt, including keywords researched using Google Keyword Planner for example. This time, we will go back to ChatGPT and see what it comes up with when using the below prompt:

Prompt: For each of the 3 pages and keywords above, create SEO-friendly meta data and make emotive

Using ChatGPT to create meta data in bulk with supplied keywords

You will probably agree that this is an excellent example of creating meta data in bulk, especially given that it’s entirely generated by AI and follows our rules well. Most importantly, it seems to have understood the need to include different keyword variations and includes emotive language to give that human touch.

Conclusion: the future of meta data creation using GPT-3?

What we can do right now with these two tools is very impressive.

As covered in our full SEO guide to ChatGPT, it is so much more than just an SEO meta data tool. We can ask ChatGPT to brainstorm content ideas, suggest headings, and even fully write the content. We can ask it to create personalised web content, blog posts, product descriptions, targeted ads, and technical articles. We can even ask ChatGPT to evaluate and improve the quality of existing content

But where do we see this heading? Right now, GPT-3 can work using the “playground” area or APIs. In both cases, there needs to be a level of textural input in terms of outlining the pages and keywords. This can help massively with ideas and productivity, but it still feels a little lacking in terms of automation.

It’s possible that in the future, GPT-3 or future versions of the AI, will be able to connect to your website and potentially third-party tools as they become more integrated. In this scenario, we could provide prompts such as below:

Example Future Prompt 1: Suggest meta data for all pages on www.websitexyz.com

Example Future Prompt 2: Use Google Keyword Planner to provide the best 3 keywords for each page on www.websitexyz.com and provide the meta data for each page

Example Future Prompt 3: Review the meta data for each page on www.websitexyz.com and cross-reference to my competitors and Google Keyword Planner. Then enhance my meta data so I have a competitive edge

Ok, the last one is probably reaching a bit too far, but you get the idea.

As ChatGPT and GPT-3 become more connected and integrated with the rest of the Internet, our prediction is that this AI will become intertwined with everything we do. We can’t see a situation where content creators and SEOs will become redundant, but it will almost certainly change how we all work.

Savvy SEO readers may have concerns about duplicate content, or perhaps have fears that Google can detect and penalise AI content. If that’s you, see our latest article on AI content detection, which covers all of these questions and evaluates tools such as GPTZero and Open AI’s new AI Classifier tool.

We hope you found our guide to AI meta data creation useful. If you have any experiences or comments regarding using GPT-3 or ChatGPT as a meta data tool, please let us know in the comments below.