ChatGPT for SEO: Best (and Worst) Use Cases
I'm sure you've seen tons of ChatGPT use
cases because they're literally everywhere. ChatGPT is insanely cool but most of
the use cases you've seen in Twitter or LinkedIn Threads or YouTube videos
are mostly hype up for engagement bait. So in this video, I'm going to share some
of the best and worst ChatGPT use cases for SEO – prompts, responses, and all. Let's start with one of the "best" use cases
for ChatGPT and that's to construct regex.
Regex, or regular expression is a sequence of
characters that's used to find patterns within text. And ChatGPT is great at it. For example, if we want to find keywords
phrased as a question in Search Console, I can ask my buddy Chat… "What is the regular expression to show matches
that contain any of the following words in it: how, what, who, when, where, why, do? By the way, these words should
be the first word in the match." Now, armed with this answer, I'll go to Search
Console, paste that custom regex into the query filter and set my Positions filter
to greater than 10.
Sort the list in ascending order and now we
have a list of informational keywords we can work on bringing from page 2
to page 1 of Google. Now, beyond Regex, ChatGPT can actually
write full out automations for you. And this is one of my favorite ChatGPT use
cases for SEOs because it's going to make your work more efficient. Here's an example: Let's say you're working on
a link building campaign. You've collected the first and last names of
authors you want to contact and of course you have the URL and domain. The inefficient way to find these emails
would be to go through them 1 by 1 in a web app or with a Chrome Extension. The efficient way would be to code something in
Appscripts to connect to an email finding API. The ChatGPT way would be to have
the program write the code for you. So I'll type in: "Using Hunter.io's "Email Finder"
API, write a function called findEmail in Google Appscripts to return a person's email address.
If no email exists, then return
"They don't want to be contacted". I will provide the first name,
last name, and domain name. Btw, my API key is this. Go!" Now I'll take that code, and open
Appscripts from the extensions menu. Let's paste in the code from my
good pal Chat and save the file. Give permissions as needed and
we're now back to the sheet. In the email column, I'll type in "findEmail",
open parentheses, click the first name, last name, and the domain. Hit Enter and you're all set. As for the rest of your list, just drag the formula
down and you're now finding emails at a much faster pace. There are literally hundreds of throwaway or quick
hacky tools you can use ChatGPT for to make your SEO more efficient but the show must go on.
Because the next use case is one of
the worst that I've seen from ChatGPT, yet tons of people are raving about it. And that's doing AI-driven keyword research. So a common prompt people use on Twitter
and LinkedIn is to ask ChatGPT for long-tail or "easy-to-rank-for" queries. And at first glance, this list in
the golf niche doesn't look so bad. But you throw them into a keyword research
tool and you realize that pretty much none of them have search demand. And Google Trends shows
the same trend – or lack thereof. As for the "easy-to-rank-for" part, this just literally
makes no sense because keyword difficulty will mostly come down to backlinks, SERP
competition, and the actual content itself. But ChatGPT isn't even able to browse the web. All this negativity has got me feeling down
so let's talk about a use case that I've been using quite a lot recently and that's
coming up with clickworthy titles.
Just ask ChatGPT to write 10 click-worthy
titles for your blog post on a topic. For the best results, give it a working
title especially if you have a specific angle you're going for to match search intent. In fact, you can even ask it to frontload
keywords if something isn't compromisable for you. Want more personality? You can even ask ChatGPT to make
your titles sound like a Mr. Beast video. I like this one. So let's actually use this for our next use
case which is to create quick outlines. In my opinion, this is one of the best use cases
for ChatGPT simply for efficiency purposes. The point isn't to get a ready-to-go outline
because you're feeling lazy – it's to get your creative juices flowing. For example, if I were creating a blog
post on this topic, I'd talk about this part a lot sooner, probably cut this
part, and add in something like this. In fact, I don't even like the title anymore so
I'd run with the angle of "Best and Worst Chat GPT Use Cases for SEO" – which
is the video you're watching now.
Now, I am curious to see what kind of use cases
ChatGPT would offer me so let's ask it to expand on use cases for point III. And uh-oh… It's actually offered two awful use cases
right here in the content generation section. Let's talk about this one on generating
so-called "high-quality content" – more specifically, long-form content creation. When you ask ChatGPT to write you a full
article, the content is almost always going to come out as boilerplate content. It'll be readable, but the quadrillion people
who are using ChatGPT will be creating the exact same content. In fact, if I ask it to create a full 1,500-word
blog post based on the title we chose using the outline that was generated, and the
use cases I agreed with, it returns this. Like I said, it's readable
and still crazy impressive. But what exactly are we impressed by? For me, it's the fact that AI is able to stay
topically relevant, complete well-structured sentences, and in some
cases, write better than me. Where it falls short is
the actual content itself. It's clear there's no expertise and there's
no way this content would actually satisfy people's intent for clicking through to
a ChatGPT use cases for SEO post.
And if you think that writing content paragraph
by paragraph would be better, you would be right. But it doesn't change the fact that expertise
is lacking, and therefore, the content is just a bunch of rambling about nothing. Now, I do think that AI has its
place in content generation. And from my experience, it's mostly with
short-form content like meta descriptions. For example, I ran a crawl for our website
and it turns out we have around 300 pages with missing meta descriptions. So if I want a meta description for this page
on how to use Ahrefs' Keywords Explorer, I can ask ChatGPT: "Write a meta description
that's no more than 156 characters for a page titled: "How to use Ahrefs' Keywords Explorer." And that's good enough for me! Now, another consistently good use
case for ChatGPT is proofreading.
For example, last year, we tested the quality
of freelance writers based on their rate. And the cheapest writer we worked
with charged us just $0.02/word. Now, if we take a part of the $0.02/word
post and ask ChatGPT to proofread it, then in seconds, it makes small
but accurate improvements. Beyond proofreading, ChatGPT is pretty
good at writing extremely concise sentences to improve clarity in your content –
which I think is super underrated. In fact, you can use this to optimize
your pages for features snippets.
For example, let's say that I'm looking
through our blog's keyword rankings and notice that we're not ranking for the
featured snippet "301 vs. 302 redirect." I can ask ChatGPT a question like,
"what's the main difference between a 301 and 302 redirect?" To which it gives a decent response. But it's not exactly optimized
for a featured snippet. Looking at the actual featured snippet itself,
I've made some mental notes to take back to ChatGPT to improve our chances
of taking over that snippet. I'll ask it to be more concise and to start
the sentence off with "The main difference" to directly address the query
"301 vs. 302 redirect". This is definitely more concise and it starts
with "the main difference" just as I asked. But this information is kind of incorrect
because there's a bit of nuance to how 301s and 302s pass link equity – which
again is why you shouldn't rely on any GPT to write full content or content
that requires expertise for you.
This brings us to another bad use case of
ChatGPT which surprisingly a lot of SEOs are raving about – and that's
search intent classification. Now, you might be able to look at a keyword
like "how to tie a tie" and immediately know that the search results will
be dominantly informational. But doing it at scale, especially for keywords
with less context is a very different story. In order to identify search intent, you need
to actually look at the SERP, the top-ranking pages' content and things like SERP features to
come to a reasonable conclusion on intent class. But as far as I'm aware,
ChatGPT doesn't have SERP data. So check this out. Several months ago, Patrick, Josh,
and I picked five random keywords and attempted to assign an
'intent bucket' to the keywords. And as you can see, we had slightly
different opinions on each keyword. But we all chose the same dominant intent.
When I put ChatGPT to the test, it did okay and
scored three out of five for the same keywords. But because it can't look at SERPs for these
keywords, it doesn't realize that an acronym like "AMA" which has multiple
meanings is navigational. And the same goes for "crockpot." In fact, it's horrible at identifying
navigational queries. And forget about keywords with local intent. If you can't trust its suggestions, you're
going to end up checking SERPs anyway, as you should, so there's no point in using
ChatGPT for this somewhat trivial task. But what it's really good at doing is
whipping up quick snippets of code for things like Schema markup,
and hreflang snippets.
Now, you should have a good idea of some
effective ways to use ChatGPT for SEO and some things you may not
want to waste your time on. Here's a full list of the ChatGPT
use cases that we went through. Take a screenshot, try these prompts for
your own site, and subscribe to meet again in the next tutorial..
