At the end of last year, Open AI released ChatGPT, a first-of-its-kind machine learning chatbot. The revolutionary chatbot experienced explosive growth, reaching 1 million users in just 5 days.
In January we posted our blog “How ChatGPT will impact search marketing”, here we explored some possible use cases for ChatGPT across the four core teams at Varn. We also wrote about some of the potential limitations of the technology. Over a month on, we have now had some time to experience using ChatGPT to aid in our SEO work.
As a team at Varn, we have identified three key use cases where we have seen the most value in using ChatGPT.
No matter if you love Regular Expressions or are terrified of them, this tip will help all of you!
Regular Expressions (Regex) can expand the possibilities of important SEO tools like Google Search Console, Screaming Frog and more. For example, we recently wanted to filter out all the URLs in Screaming Frog so that only those containing a query string would display. This is exactly where we would want to reach for a regular expression.
However, with all the rules and complexities, writing Regex is very difficult. Even using documentation, writing a simple regex string can be incredibly time-consuming. This is where AI tools like chat GPT can come in extremely handy. Because AI has an understanding of what a user is asking, and can use this to write the Regex for us.
Here is an example chat log with chat GPT:
As you can see, we simply asked the chatbot (in plain English) how we would like to sort our URLs. Within seconds it gave us the working regex code, along with a handy description of how it works. By copying and pasting this regex into the Screaming Frog filter, we were able to sort URLs so that only those containing query strings were showing.
The true power of this feature comes from its flexibility. As SEO professionals, we could easily be presented with any number of pages or queries that we may need to filter down. But with the analytical abilities of AI, we can convert any written command into a working regular expression in seconds.
This use case has improved the way we use Regex at Varn, but it does come with limitations. ChatGPT will always try to do the best job with the prompt you have provided, but it will rarely ask for more information if it feels your query is inadequate. This means that when writing your prompts for ChatGPT you should always be as specific as possible, leaving no room for a different interpretation.
Additionally, you should be very careful to not just copy and paste this code if it’s being used for something important, like a client-facing Looker Studio filter. Instead, you should use a tool like regex101.com to test the regular expression against any unusual edge cases it may encounter.
One of the most challenging parts of writing SEO-friendly blogs is coming up with an engaging idea. But Chat GPT can help!
Chat GPT is extremely versatile, and it is also able to assist with more creative processes, as well as technical tasks such as writing Regex.
Blogs can be a great way to improve the Search Engine ranking of a site. They also provide value to visitors which helps to keep them engaged. However, we will often find ourselves racking our brains for a blog idea that is both engaging to a user and will also tick all the boxes for SEO.
Here is an example of where we have used Chat GPT to assist in the ideation of new blogs on the Varn website.
Here you can see that as a result of our request, we have been provided with a bulleted list of ideas for potential blogs on the Varn site. It is important to remember to give a very descriptive prompt for chat GPT – In this example, we found that we needed to specifically state that we wanted the answers in a bulleted list. Otherwise, we were presented with completed blogs, rather than blog ideas. We also provided a list of keywords that we could potentially use, but needed to make it clear that using these keywords was optional. Otherwise, GPT would just cram as many keywords as possible into each bullet point. This extreme level of specificity is vital if you want to unlock the true power of chat GPT.
There are a range of ideas in this list of potential blog posts (some of which are better than others), but overall this list could be a valuable resource when coming up with new blog topics. For example, the 8th blog in the list brings up an interesting idea about deciding when to reach for PPC or SEO for your web promotion. There are some ideas in this list that are too broad, or which lack a certain context, but overall there are many good prompts that you could use when creating a future post; They may just need some work to progress them into a workable blog title.
Writing Metadata and headings with SEO in mind can be time-consuming, but Chat GPT can help streamline this process.
In the previous section, we showed how we have been using chat GPT to assist in generating blog ideas. However, we can also use Chat GPT to create content for these blogs.
Here is an example of how you can generate metadata and headings using Chat GPT.
Once again within this section, we found it was very important to be hyper-specific with our prompts to Chat GPT, even going so far as to state the desired length of the metadata we wanted. The results are very interesting, the meta and headings are all well related to the blog topic we gave, and they do a good job of including keywords we might want to optimise for. In addition to this, the content reads well and tells a good story of what kind of content the rest of the blog might include.
Despite the fantastic quality of this AI-generated content, there are some problems. Firstly, the metadata actually exceeded the character limit we provided, this is because GPT cannot directly count the number of characters it is generating. For this reason, the character counts are often not what we asked for, even if it’s usually not off by much.
The biggest limitation preventing us from using Chat GPT to generate content like this is the risk of search engines picking up on our use of AI-generated content. Historically, Google has explicitly given guidance to avoid the use of “automatically generated content”, and this is what makes us so cautious about using AI-generated content on websites.
However, over the past year, Google has been slowly changing their guidance on AI content to appear more in favour of its use. Last month, the company properly clarified their stance on AI-generated content, essentially saying that it doesn’t matter how the content is produced, it will be judged based on the content’s quality. This suggests that AI content could be usable in SEO, provided it is user-friendly and meets the E-E-A-T guidelines set out by Google. Despite this statement by Google, we still tend to air on the side of caution and remember that AI-generated content is not perfect. It is also important to remember that as SEO experts, we have a fully formed understanding of the specifics and will be able to produce higher-quality content than a chat bot. For this reason, we tend to use AI generation as a tool to help inform our own content and will never copy and paste anything directly from Chat GPT.
So far chat GPT has had an extremely positive impact on our SEO work at Varn. It has allowed us to speed up some tedious technical tasks and allows us to put more time into other areas of our work. Even now, we still believe we could find other uses of the technology to better improve our workflow, and we are excited to explore these possibilities.
However, we also believe that it is best to think of Chat GPT as an (albeit very powerful) tool in our toolbox, not a superpower that will do all of our work for us. This means it is still important to have a good understanding of the task you are trying to accomplish, so that you can review the quality of the response and provide Chat GPT with accurate information and context.
Varn is an expert & specialist SEO search marketing agency. Consultancy * Search * Analytics * Audits