Illustration created by Midjourney with the prompt “a highly stylized drawing of an AI-coworker designed to write blog posts, colorful, vibrant –v 4 –ar 3:2”
Can you really use ChatGPT to write better blogs? First we need to define what we mean by “better.” For our purposes, a better blog post will have a clickbait title, have a killer topic sentence, incorporate every possible keyword and keyphrase that might help win “the box” on Google, and then, and only then, be informative, enlightening, and entertaining.
Is ChatGPT up to the task? No. Not by itself. But I wasn’t ready to take no for an answer, so I started experimenting. I decided to try a combination of tools to see if the AI-assisted work product would outperform my purely original work. Unsurprisingly, the work done in partnership with my AI-coworker outperformed work I did alone. Here’s the process.
The Initial Content
First, you need a story. You can write it yourself (which is what I do, and what I did for this experiment). Alternatively, you can describe the persona you want ChatGPT to assume, describe the content you need, and ask ChatGPT to write it. Or you can copy and paste the most popular article on your subject into ChatGPT and ask the AI to reword it for you.
The ethics (or legality) of asking ChatGPT to paraphrase, reword, or rephrase someone else’s work are outside the scope of this writing, but the software is capable of rewriting at a very high level.
No matter how you generate your content, step one is getting a first draft from ChatGPT. If you are character limited, just type “please continue” or “keep going” into ChatGPT and it will continue from where it left off.
This is where many people would stop. But I was not happy with the output. It was written at about a 7th-grade level and it violated Strunk & White’s Rule 17, “Omit needless words.” For this experiment, the extra words didn’t yet matter because the writing was also missing the highest-ranking keywords and keyphrases.
Keywords and Keyphrases
I searched for the keywords and keyphrases that were most associated with the topic and made a list. Then I crafted a prompt for ChatGPT that asked it to “rewrite my article to incorporate the following list of keywords and keyphrases.” This took several tries. I found myself regenerating individual paragraphs – in truth, I could have written most of the SEO-enhanced sentences faster by myself.
Help from the Pros
Next, I pasted the article into my favorite blog-scoring tools and incorporated most (but not all) of the suggestions. I won’t tell you which SEO scoring tools I used, because I’m not shilling for a particular brand here, but there are many to choose from: SEMRush, Ahrefs, Yoast SEO, Google’s SEO tools, etc. You should choose the one that fits your needs (and your budget).
In my case, additional keywords were generated, so I repeated the process of having ChatGPT rewrite the article to include all of the keywords that made sense. I also paid some attention to information about search intents generated by my various SEO tools.
As a final step, I pasted the work into a plagiarism checker to see if it would get past Google’s very keen filters. Then, just for fun, I had a different AI product rewrite the finished post so I could do an A/B test.
Was the Juice Worth the Squeeze?
The results were fine. I think we’ll know more after a few more tests. In practice, we would need to generate the content using ChatGPT for this test to help us understand if we could use AI to do the heavy lifting. In practice, we already pay careful attention to SEO best practices, so there wasn’t enough room for improvement to move the search needle in a short-term test.
However, I can easily imagine a team dedicated to affiliate marketing or some other transactional business creating a very profitable (almost fully automated) workflow and process for the creation of well-SEO’d content marketing pages. You could probably automate the process with a no-code solution like Zapier – the steps are that easy.
Where Does This Go?
School systems and teachers are already up in arms about plagiarism and cheating using tools like ChatGPT. Will the business community become equally alarmed? Even if people do get their ire up, is there anything to be done? My guess (one person’s opinion) is that we are about to see an explosion of applications built on top of large language models. A little more than 10 years ago we saw mobile-centric and mobile-only applications (such as Uber, Instagram, Snapchat, etc.) emerge as smartphones became ubiquitous. More recently, as cloud computing became ubiquitous, we saw companies (such as Notion, Slack, Airtable, Figma) profit by building applications on top of cloud infrastructure. What’s next? The obvious answer is the application layer for AI.
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. I am not a financial advisor. Nothing contained herein should be considered financial advice. If you are considering any type of investment you should conduct your own research and, if necessary, seek the advice of a licensed financial advisor.