Claude vs ChatGPT for SEO
ChatGPT has been around for two years now, while Claude (also known as Claude AI) only appeared in market in March of 2023.
Plenty of marketers have started to incorporate AI into their workflows to minimize busy work — and LLMs like these two are no exception.
So, how do the two compare when it comes to SEO?
Keyword Research
Claude
When prompted to do keyword research, Claude can provide a generalized guideline.
It can't access real-time keyword data and makes assumptions based on pre-existing information instead.
If you ask for more specific data (such as longer-tail, lower-competition keywords), Claude is once again operating on assumptions and examples.
While these guides could be helpful, they're not very specific or accurate.
ChatGPT
When given the same prompt as Claude, this is the output ChatGPT gives you when you ask for keyword data.
Just like Claude, it can't access real-time data and instead references a more general approach to keyword strategy.
ChatGPT, like Claude, is also not able to give real instances of longer-tail, low-competition keywords but instead gives advice on the types of keywords you should target.
It also includes sources.
Neither Claude nor ChatGPT are of much use here — you'd be better off going directly to a keyword research tool like Ahrefs or Semrush.
Search Intent Analysis
Claude
Understanding search intent can be the difference between winning and losing a conversion.
This is Claude's response to the request, "Analyze the search intent for "how to rank a product page" and what the searcher is looking for."
You can do this in bulk for a set of keywords by uploading a spreadsheet or file for Claude to read, which is possible with its free version.
This also applies to keyword clustering and other SEO basics.
However, it's important to note that its breakdown of what to cover:
- doesn't take into consideration the pages that Google has ranked highest
- doesn't include the format you should use (for example, video or text)
- isn't based on real-time search data but rather on assumptions
More on that in the next section.
ChatGPT
Like Claude, ChatGPT can give you the most likely search intent for given keywords based on the information it has access to. It can also cluster keywords for you and arrange the information in whatever format you like.
Its response to search intent analysis for "how to rank a product page" was fairly similar to Claude's response, with one or two deviations.
The free version of ChatGPT (3.5) isn't able to read files, though, so if you don't have the paid version you'll have to upload these keywords manually.
Just like Claude, ChatGPT doesn't prioritize this information according to what's ranking, which means its outputs are more generalized.
Topical Coverage in Generated Content
Claude
I prompted Claude to write me an outline for an article titled "How to Rank a Product Page: A 2024 Guide."
I specifically asked it to reference the top pages on Google for this keyword to make sure that I talk about what's already covered today.
In return:
- Claude told me it can't read top pages out there today
- I got a general idea for what to write for each section
This response is so generalized that I have little to no chance of ranking on the SERPs for this keyword — especially since I have no idea what top competitors are writing or what Google has prioritized.
This same logic applies to fully-fledged drafts.
To get to parity with top pages with Claude, you'd have to manually feed it individual pages from the SERPs today.
ChatGPT
When I gave ChatGPT the same prompt for an outline, it did a better job of referencing top articles from its most recent update on the internet (namely, Semrush).
But it also acknowledged that it doesn't have real-time access to what's ranking on the SERPs right now.
Again, I'd have to feed it those articles individually.
Gap Analysis and Proprietary Data
You need to bring net new information and perspectives to the table in order to outrank your competition on Google.
I asked Claude to do gap analysis on the keyword "how to rank a product page" and to provide me with unique expertise on the subject.
In response, Claude gave me:
- fictional expert quotes
- suggestions for new angles and approaches to take based on common SEO best practices
While its suggestions for new angles could inspire unique information, it's incapable of giving me access to real expertise that is totally unique and impossible for my competitors to copy.
I asked Claude if it could maybe automate a process to gather expert quotes in real time.
TL;DR: You can't rely on Claude for nuanced opinions or unique expertise since it's entirely dependent on training data from the internet at random intervals.
ChatGPT
When I asked ChatGPT to perform gap analysis and provide me with expertise, it simply referred me to tools to use and gave generalized advice.
Like Claude, it is simply unable to access real-time expertise or give me credible quotes to work with.
It could give suggestions for new angles to take, but once again, this is based on assumptions.
You could build a GPT model yourself that can access this data for you, but that requires coding expertise (or expensive developers.)
Technical SEO
Claude
Can LLMs take care of SEO "busy work" for you?
These are the prompts I used on Claude and its responses.
1. Extracting and Analyzing Keywords
I asked Claude to extract keywords from given text. You can also upload files for it to analyze.
I then instructed it to cluster those keywords.
I asked it to identify the buyer journey stage for each of the keywords.
Claude held up pretty well here — although its worth noting that the information on the above output is more contextual versus based on real buyer journey data.
2. Generate Metadata for This Article
Claude was able to write metadata for a given article.
You might need to check that this isn't repetitive and that no information gets cut off on Google.
3. Turn This Video Into Text
I tried to upload a video to Claude via hyperlink and via the attachment option and to turn that video into text.
This is to help me quickly repurpose content.
Claude was unable to do either (keeping in mind I'm using the free version.)
4. Site Architecture Suggestions, Core Web Vitals, and More
Claude is unable to access direct links in real-time, and is incapable of helping you improve your site architecture.
I asked it to:
- give me internal linking suggestions for my site
- give me improvements core web vitals
- find me opportunities for SEO refreshes
All to no avail. This makes Claude useless for specific technical SEO improvements on-site.
5. Generate Schema Markup For My Site
Claude was able to generate schema markup for a web page that I provided, although some data was missing or was given placeholder data.
It's important to run this through schema.org.
6. Find Opportunities for On-Page SEO Improvements
Claude was completely incapable of suggesting technical SEO fixes for an SEO article provided, severely limiting its use.
ChatGPT
We actually wrote an article on the best ChatGPT prompts for SEO busy work.
Following the same use cases as above, here are ChatGPT's responses.
1. Extracting and Analyzing Keywords
Like Claude, ChatGPT has no problem categorizing and organizing information when given that information.
I prompted it to cluster these keywords.
I prompted it to identify the buyer journey stage for each keyword. (Remembering that this is based on context versus real understanding of searchers.)
2. Generate Metadata for This Article
If you give ChatGPT access to an article, it can write metadata for you.
You'll need to make sure that neither of these are cut off on Google — something that ChatGPT doesn't natively optimize for.
3. Turn This Video Into Text
ChatGPT pro is unable to access videos and transcribe them.
4. Site Architecture Suggestions, Core Web Vitals, and More
Just like Claude, ChatGPT is unable to directly access your site to help you find and improve site architecture or core web vitals.
You'll have to do this yourself with web crawlers, or spend hours feeding ChatGPT databases.
5. Generate Schema Markup
ChatGPT was able to create this schema markup for an article that I gave it.
However, the output is populated with example text at times and will need to be thoroughly edited and checked with schema.org.
Essentially, ChatGPT gives you a jumping off point for schema data.
6. Find Opportunities for On-Page SEO Improvements
Although ChatGPT does reference the article I gave it to improve for technical SEO, its suggestions are still generalized.
The language is also generic.
In this context, ChatGPT and Claude are best used to correct grammar and syntax to improve readability.
Claude vs ChatGPT
In the end, Claude and ChatGPT are exceptionally similar when it comes to their SEO capabilities.
Capabilities | Claude | ChatGPT |
---|---|---|
Keyword Research | Cannot access real data | Cannot access real data |
Search Intent Analysis | Based on contextual assumption | Based on contextual assumption |
Topical Coverage | Based on contextual assumption | Based on contextual assumption and latest knowledge of SERP |
Gap Analysis | Based on contextual assumption; cannot give unique data | Based on contextual assumption; cannot give unique data |
Technical SEO | Can make generic on-page suggestions; doesn't have any access to your site in real time | Can make generic on-page suggestions; doesn't have any access to your site in real time |
LLMs Have a Limited Place in SEO
From the above analysis, it's pretty obvious that LLMs have severe limitations which will impact just how much of your SEO strategy you can safely pass off to them.
Namely:
- Without plugins or extensive manual input, LLMs do not have access to real-time data. It cannot give you property valuations or keyword volume
- Their information is based on the internet. LLMs cannot discern between a credible article from a Nobel Prize winner and an unqualified person's additions to Wikipedia. So, you can get wildly inaccurate information at any given time
- Their outputs are largely assumptions. LLMs feed you information that you most likely want to hear according to previous prompts and tokens. This doesn't tell you what your prospective buyers are really searching for, nor does it give you top search results or an indication of what Google values in responses
- They're completely incapable of giving you unique information or expertise. This gives you no advantage in terms of standing out against thousands of other pages saying the same thing
- They're incapable of making accurate technical SEO suggestions or improvements to site architecture. LLMs don't have free access to your site's information. You would have to consistently feed them huge datasets for any suggestions to be useful or accurate
Using LLM outputs as-is for SEO will cost you credibility and revenue. They're not helpful to real users and ultimately lead them nowhere.
What Can You Safely Use LLMs For?
LLMs have access to decades of data on simple, straightforward subject matter like definitions.
This means you can safely use them for glossary pages in bulk and for programmatic SEO, if this strategy would drive real buyers in the context of your business model.
This works because no fresh perspectives or nuance is needed. The opposite can be said for SEO, which is why using LLMs for SEO will not get you anywhere with Google or with your buyers.
LLM Alternative for Smart SEO Automation: Letterdrop
That is not to say that there is no AI out there that can accurately help you rank higher on Google in a way that is of value to customers.
Letterdrop is capable of aggregating the following into full-blown research dumps, outlines, and drafts for SEO:
- Real-time analysis of search intent
- Real-time analysis of top SERP results and features
- Real-time gap analysis and unique suggestions for adding net new information
There's also a workflow that enables you to easily embed subject matter expert quotes into your blogs, helping you rank higher with more unique content.
It also has access to your site in real-time. It's capable of not only recommending on-page and site-wide fixes, but of automating them entirely.
This includes:
- Internal links
- Fixing broken links and redirects
- Over 60 other on-page SEO rules
Want a closer comparison of ChatGPT vs Letterdrop? Check out our recent blog article.
Reserve LLMs for Glossary Pages
You get the most out of AI when you play to the strengths of its model.
Assuming that programmatic SEO is a viable strategy for your company, tools like ChatGPT and other LLMs are excellent.
The same goes for glossary pages.
Don't compromise the quality of your SEO articles, which are meant to be of genuine help to prospective buyers, for perceived convenience (and cost saving.)
If you're interested in using an AI tool built for helping you rank higher, reach out to us.
Subscribe to newsletter
No-BS growth strategies and content marketing tactics in your inbox twice a month.
Related Reading
Some other posts you might find helpful