Now that I have some terms with difficulty and search volume, I can go back to ChatGPT to help add some information for these terms:
ChatGPT had some issues generating this table, but ultimately the output is pretty impressive:
Now, I can also get some other interesting keyword data on my new list of terms by asking, “Can you tell me how many posts are in each sub-topic and how many fit into each level of search intent?”
Many times you may balance your content output across categories, business lines and support for different tools or products. This is a handy way to sum up that information quickly.
Keyword list generation, expansion and cleanup
Finding specific keywords to target and appending different modifiers to your terms to grow your list are critical aspects of keyword research.
A common application is to generate a list of geo-modifiers and marry those with a target term:
You can also get a list of size modifiers and apply that to different terms:
As you can see, you then need to layer in some additional prompts, give examples, and ask ChatGPT to clean things up to make these usable, but it can help you skip a step and give you some new ideas for modifiers.
Page-level keyword research
Many tools will create content briefs and grades for your content based primarily on what’s currently ranking well in search engines.
I use multiple tools in my day-to-day work, which can be pretty expensive.
We can use ChatGPT to compare our keyword set to the terms being targeted on posts ranking for specific terms.
Again, it is not a native SEO tool, so data will differ from what those tools use.
And you can get more specific advice with a follow-up prompt:
Here, I could run through multiple articles ranking on the first page for this search term, ask ChatGPT to pull out themes and common terms used in the headers for each, and then surface the most popular phrases across the articles:
For a smoother and more scalable experience in prompt-heavy tasks, you may want to use the OpenAI API. Have ChatGPT generate the code for you (now using Canvas):
ChatGPT prompts for keyword research APIs
Reminder: Proceed cautiously when dealing with programmatic APIs that cost money or credits. QA everything the tool spits out.
Another keyword research function for ChatGPT is to help SEOs interface with different keyword tool APIs:
For my new pickleball site, I could grab an extensive list of low domain authority sites ranking for some pickleball terms and then run them through this script. (Always make sure you extensively QA any code ChatGPT generates for you.)
If you’re not very technical, you can get step-by-step instructions and ask the tool how to complete a generic task:
Not relevant for my new pickleball site, perhaps, but if I want to do some keyword research on my data, I can get ChatGPT’s help with the Google Search Console API to find search queries that may need their page or article:
The GSC API isn’t the only way to get keyword data using ChatGPT.
ChatGPT keyword research regular expressions
Like the GSC API, using regular expressions (regex) within Search Console can be a great way to unearth keyword targeting opportunities:
Build your own tools
ChatGPT can instruct you on how to build specific tools for keyword research. (Again, always QA and proceed with caution!)
ChatGPT keyword research QA tips
The examples above demonstrate how valuable ChatGPT can be for keyword research. The price is free to $200 a month, so it is worth trying out.
In my experience, the platform tends to fall flat in three core areas of keyword research:
Not search-specific: ChatGPT is a native keyword research tool. It does not have direct access to keyword data and can’t perform key functions for keyword research.
Confident but wrong: The platform occasionally provides answers emphatically that aren’t true, delivering data presented as a fact but is flat-out wrong.
Broken code: ChatGPT’s code doesn’t always work. If you rely on it to interface with an API or build a tool to automate certain aspects of your process, assume there will be bugs and issues.
Here are a few specific suggestions to help:
Ensure you use dev environments and sandboxes when working with ChatGPT code or instructions.
Check any keyword suggestions against search and competition data to confirm you’re not spinning your wheels on subjects that don’t make sense for your site and your bandwidth.
If you’re asking the tool to analyze something, take the output with a grain of salt. Remember that ChatGPT could be wrong, cannot crawl URLs and has an outdated data set.
Use examples of desired outputs, give specific feedback on the output you get from the tool (with a grade), and ask the tool itself for feedback and suggestions on prompts.
Broader ChatGPT keyword research strategies
Caveats aside, SEOs will likely find useful keyword research applications with ChatGPT.
Ask yourself these questions to determine how to use the platform best:
How can you leverage the ChatGPT API?
If the chat character limit keeps you down or you have a larger-scale keyword or topic ideation task, you can use the API for various functions.
Are there other applications or APIs you want to use but can’t dedicate the time and energy to? What about a mash-up of multiple APIs (marrying data from e.g., GSC, Ahrefs or Semush)?
What are your more tedious and time-intensive keyword research tasks? Can ChatGPT help with these?
Think of ChatGPT as a virtual assistant. Understand that you’re ultimately responsible for the implementation of any data or code it generates.
Consider the tasks you need to complete, the things it’s likely capable of and try them out when it makes sense.
Contributing authors are invited to create content for Search Engine Land and are chosen for their expertise and contribution to the search community. Our contributors work under the oversight of the editorial staff and contributions are checked for quality and relevance to our readers. The opinions they express are their own.