ChatGPT & SEO: Analyzing Word Relevance for Large Keyword List
If you are handling SEO strategies for a large e-commerce business, you might have searched for ways to optimize keyword research processes. I have been trying to find a method to replace many hours of manual human review for each keyword and determine its relevance to specific product categories. In one of our e-commerce projects, we have around 450 subcategories of products (with thousands of lower-level subcategories). You can imagine how time-consuming it would be to optimize this using only manual human labor.
Initially, I tried looking for word embedding techniques that use vectors, such as Word2vec, and using Python scripts. However, since I don’t have strong technical or coding skills, I decided to set it aside for the time being. Later, while experimenting with ChatGPT, I came up with the idea of using it to determine relevance for a large set of keywords. I tested it on extensive keyword datasets in the Czech language, and the results were impressive! The accuracy is around 70%-80%, based on my subjective evaluation of the outcomes (I am working on a more accurate assessment, but for now, it is at least subjective)
Now, I have two options:
A) I can use ChatGPT for determining relevance — the results will be about 70% accurate, but this will significantly speed up our progress in on-page activities. Instead of reviewing/editing 2–4 subcategories per month, we can process 6–10 of them.
B) I can continue with the “old” manual process of selecting relevant keywords, which will be nearly 100% accurate but time-consuming.
I chose the less accurate but faster solution (we can still manually review it after finishing all categories). Now, I’d like to share an easy process that you can try right away :)
Using ChatGPT in Google Sheets to Assess Keyword Relevance
First, you need to install the ChatGPT plugin for Sheets and Docs. I’ve included a quick setup guide as part of my previous article (box bellow) — just look for the “ChatGPT Google Sheets Setup” section.
Once you have the ChatGPT plugin set up in your Google Sheets, prepare your data. You’ll need a list of individual keywords (for example, in column A) for which you want to determine relevance, as well as a group of keywords representing the topic of the category. I usually define a category with at least 10–15 keywords, but you can also try defining it with just one general keyword.
If you’re defining the topic of a category using a list of several keywords, place them in a separate cell and separate them with commas. If you’re defining the topic with just one keyword, you can use it directly in the prompt.
Here’s an example of a prompt:
On a scale from 0 (not relevant) to 10 (very relevant), determine the relevance of the word:"&A1&" overall to the group of these words:"&$C$1&". In the output, use only the number, do not write any word.Note: In cell C1, I use a list of keywords to define the category’s topic, with keywords separated by commas.
A few of my notes:
- Initially, I used a 0–100 scale, but since I only received numbers rounded to tens, it wasn’t helpful. However, you can choose what works best for you.
- Even when I ask ChatGPT to give only a number as an answer, it sometimes doesn’t do it. Instead, it may give a text response or show relevance keywords for each individual word in a group of words that describe a category or topic.
- When defining the category with just one word, the chances of getting word answers were much lower.
- For certain keywords, an error may appear instead of the output. In such cases, the function needs to be removed and re-entered.
- You might need to change the level of relevancy to decide if a word is related or not, based on your own words. I often remove words with relevancy lower than 7 (sometimes 6) automatically.
Your Feedback Matters: Tell Me What You Think
If you have any questions, suggestions, or ideas, please feel free to share them in the comments section. I’d greatly appreciate it if you’d follow my profile here on medium.com. Stay tuned for more intriguing articles on automation and the utilization of AI in SEO.
