Can You Really Use AI for SEO? A Scientific Experiment
Is AI going to kill search engine optimization? Or is AI going to make SEO super easy?
Let me start by pasting a very illustrative disclaimer that Alli AI, a website that calls itself “the best AI-driven SEO tool,” has at the bottom of its homepage:
“*Performance, traffic and ranking representations made by AlliAI.com, Alli AI for SEO, [Alli Services, Alli Consulting] and their advertisers/sponsors (collectively, “Alli AI LLC”) are aspirational statements only of your earnings potential. Results on this page are OUR results from years of testing. These results are not typical and results will vary.” (Emphasis mine.)
This statement demonstrates to me that there's a lot of hype around AI for SEO content generation, keyword research, content creation, content optimization, and more. And companies are using that hype bandwagon to try to sell an AI powered SEO tool… that doesn’t necessarily get results.
TL;DR: Can you use AI for SEO? I tested multiple AI tool types on my blog. I found that with one exception, the best results came from no AI intervention whatsoever.
This box-and-whiskers chart shows the distribution of clicks per article, separated by articles created almost entirely with AI, outlined with AI, and written with no AI assistance whatsoever.
This longer article is for the people who are really curious, who don't just want the quick answer, who want the thorough answer into whether you should use AI for SEO. Let's get started.
Prefer watching? Catch the full video here, including some funny bloopers and behind-the-scenes shots:
[insert video]
The SEO War Behind the Scenes
I never really used to think about SEO strategy. I would just get on Google and I would type “best chocolate chip cookie recipe,” click on the answer, make my chocolate chip cookies, and move on with my day.
Behind the scenes, thousands of chocolate chip cookie recipe authors/ marketers were vying for my attention, for my click, for my eyeballs. This is because my eyeballs, and yours, are valuable.
My eyeballs represent an ad view, which means the person who wrote that blog post gets paid because I've looked at an ad. I might sign up for their newsletter where they can sell me additional recipes, or I might be tempted to buy a chocolate chip cookbook.
Three ads and a newsletter sign-up link. I’m not better than this! I have ads on my website content, too. This is the model of content that SEO and Google incentivize.
So all these SEO professionals are trying to get Google to rank them first. That's the war.
Over 100,000,000 results for the keyword “chocolate chip cookie recipe.”
Now, the rules of ranking are simple, but they're secret. Google doesn't want us to know exactly how it decides what article goes on top. If we all knew exactly how it worked, there would be nothing stopping us from gaming the system.
Indeed, many people do game the system. There are any number of SEO strategy experts (or SEO "experts" 🙄) who will try to sell you on the secrets of their SEO success.
Google's Objective
Important: Google isn't trying to trick anybody. It's just trying to make sure that I get the right chocolate chip cookie recipe for my needs.
If I don't get good results when I use Google's search page, I'll stop using Google and I'll take my very valuable eyeballs elsewhere.
Why does Google care about my eyeballs? Because Google is not a search engine company, despite the common misconception. It makes money from ads.
Google is an advertising company. It doesn't make money from me paying to use the search engine. It makes money from the companies who pay Google to show me ads on website content. And if I create content that ranks, and you click on my site and see ads, I earn a portion of those profits, too (as well as being able to monetize attention in other ways, like affiliate links).
This is where SEO strategy comes in. Google employees obviously don't read every single recipe that gets posted to the internet every single day. Instead, they have what I call “reader robots” that crawl the website to give it a search engine ranking.
Robots can't read, but they can scan for a keyword selection and get an idea of what the article is about. Then when you search for a keyword, and that article contains those keywords, Google shows you that result.
There are a couple other things, but that's basically all you need to know about how Google works with SEO content to make sure that it is showing the right content to the right reader.
Why use an AI SEO tool?
To make sure that you keep up with Google's reader robot, it's not enough to just post one amazing recipe and leave it at that.
The most reliable way to get SEO hits on your blog is by posting a lot of content, at least once a week. The more SEO content you produce, the better you're gonna rank.
SEO content basically equals money. Getting hits on your SEO content means you're earning from ads, cookbook sales, all the rest.
In 2022, I posted a six-ish articles of SEO content designed to rank for SEO. In 2023, so far I've already written 28 articles that are SEO'd.
A simple illustration of the rule more SEO content = more hits. This graph shows, in pink bars, how much traffic I got in 1,000s of hits. The yellow line shows the cumulative number of SEO articles I posted.
I learned that more SEO content = more traffic.
So imagine you're in my shoes. “All right. I have to put content out there if I want to rank. But I don't want to be an SEO monkey my whole life."
The problem? More SEO content is how you rank.
People get around this problem in a couple of different ways. Sometimes they just write crappier stuff that still has a keyword in it. Sometimes they outsource this to people from the global South, who they can get away with paying a lot less money. Or they'll use content mills to create content.
And lately, people have used an AI SEO tool to crank out content.
With AI, you can write a “high quality” article with a click of a button. You can conduct tons of keyword research. You can, allegedly, do content optimization with a tool that's AI-powered.
Note: By “high quality,” I don't mean something that you could read in the Atlantic. I just mean that artificial intelligence can now string together a series of words in a sentence shape. It “knows” what a preposition is and what's most likely to come after one.
Suddenly it's very easy to create content with AI content tools. And you don't even have to pay the pennies a day that you would pay somebody from the global South to do it. You can press a button and get free AI content, an article a day easily. And why stop there? Hell, you could have five articles, ten articles, one hundred articles a day with very little effort.
Does this follow the rule of more content = more hits? Does AI content rank?
Does AI content rank for SEO?
We already know the rule of “more content = good.”
My original theory was that AI technology would be really good at SEO.
Why?
SEO is robotic by nature. Nobody does keyword research because their users care about keyword research. To rank for SEO, you're not writing for humans. Remember, those Google employees are way too busy to review every single one of your chocolate chip cookie recipes. Instead, it's left to a robot.
Who better to write content for a robot than an AI algorithm?
Plus, most generative AI models like ChatGPT have been trained on a huge body of text that contains a lot of chocolate chip cookie recipes that rank on Google. ChatGPT already “knows” the structure of winning SEO article.
There are issues, of course. Almost every tool has been caught plagiarizing and making stuff up wholesale.
But even with those downsides, you can still understand how tempting a tool like this would be to someone who believes more content = more hits.
Well, confession time: I was that person. I used artificial intelligence to generate articles. I didn't do it once – I tested a bunch to find the best AI SEO tool. I wanted content optimization at a click. And I did it all for science.
Results of my SEO AI experiment
I went into this experiment with my theory that AI would actually be pretty good at SEO content creation.
But my SEO journey started before AI was really a thing. I used tools like MarketMuse (affiliate link because I love and recommend MarketMuse to anyone) and Yoast SEO, but nothing fancy.
I said to myself, “You know what, Zulie, it's time to take SEO seriously. Time to learn how to use Google Analytics, and the Google search console. Time to getting more hits, more views, more ad revenue, and more email newsletter signups.”
So I started taking it seriously. Almost instantly, I ran into two very real problems.
I wanted to do one article a week. I didn't suddenly have more time; I had to find time to write an additional weekly article.
A lot of SEO-perfect content is boring to write.
For example, the keyword “content marketing and AI.” I just couldn't think of anything interesting that I personally had to say about content marketing and AI, because I don't think AI is really good at it - -you need a thoroguh understanding of the human psyche, which AI doesn't have.
But I knew that I could potentially rank for it.
When ChatGPT rolled around, I decided to try it for my SEO efforts, for science. I kept a record of what tool I used for AI generated content, which was the best AI tool I would keep, and how well artificial intelligence worked for SEO optimized content.
I primarily used the free version of ChatGPT, because it's free and I found it to have the best AI user interface. I also tested out a bunch of other AI writing tools, mostly because those keywords rank, e.g. “scalenut review,” “jenni.ai review,” “contentshake review.”
I then repurposed the content I created through the trial to post on my website.
Here are my results:
That same chart illustrating how for me, the best content didn't use AI.
As you can see, the mostly AI written content didn't perform well. The outlined content did okay, especially when using ChatGPT or SEOwriting.ai (affiliate link because it’s the best I’ve found so far). But articles with no AI optimization at all did consistently best of all.
Here is a full screenshot of my results.
There were some exceptions: I used ChatGPT to help me add context and history to my best word fonts article. And my contemporary fiction article is getting some hits, too. But mostly, articles with no AI involvement did best.
I also spoke with others to see if their strategy worked. For example, Zaid Hashmi was able to get his keyword “Customer acquisition through WhatsApp” on the first page of Google, just below the featured snippet.
[Note: Zaid shared his results with me in exchange for me linking back to the site.]
He used a combination of ChatGPT and Surfer SEO. Here's the strategy he shared:
Finalized keywords for the blog
Punched in the keywords on Surfer SEO and analyzed the top ranking blogs for the keyword
Went to ChatGPT & put in the prompt to analyze the sites and give the structure of the blog
Gave the prompt to generate content for the structure. Heavily proofread and made necessary changes, added data and reports wherever required.
Ensure ChatGPT included the keywords while generating content.
Published it on the website after adding images, videos etc.
He gets ~10-15 clicks per month on this article.
Screenshot of hits on this article in the last three months. Screenshot shared with permission.
Another SEO writer, Bhavik Sarkhedi, shared that he was able to get his AI-generated content ranking “Blockchain development companies in India” on the third page of Google. (Again, this was shared in exchange for a backlink to that article.)
This is the prompt he used:
ChatGPT, Write a comprehensive, SEO-optimized article with the title 'Top 10 Blockchain Development Companies in India' The article should be at least 1500 long and
1. Discuss factors such as technology advancement, limitations, upcoming trends, technologies, alongside the summary of the particular companies.
2. If required, add relevant analogies during the article if required - don't start the analogy with 'Picture as' or 'Imagine' as it looks totally robot-ish.
3. Write this whole article in the first person who is a professional expert of SEO, blockchain development and software technology.
4. The keywords for these articles are Blockchain Development, Blockchain Development Company, Top Blockchain Development Companies in India. - should be used 3 times each in the whole article.
7. Don't start the paragraphs with prepositions or '-ing' form, it looks AI generated. You will first give me the total 5 title options and the article outline.
8. It will be like a very structured format for all top 10 companies, so maintain a tone and style like a professional writer.
9. Don't make up the company names; I need the list of real 10 company names, not imaginary.
10. The article should have a proper beginning and ending. After I approve it, you can start writing the article step by step as you may need many outputs. If there is a change in the title or outline, I will suggest it and then you can continue. Do you get it? Do you have any questions? If yes, please ask. If not, start
So it's clear that AI can help—at least a little bit.
But for me, writing articles myself still results in better optimization than just pressing generate. Even outlining with AI doesn't have consistently great results. As a content creator, I'm relieved that I have an extra incentive to just do the work myself.
Ultimately, I think my results prove that AI isn't going to take over SEO. Here's why.
Why AI isn't going to kill SEO
“In the 10 plus years I've been doing SEO, I think SEO has died about every six months,” Kyle Roof, SEO expert told me. “Somehow it survived. If you were around in February of 2023, SEO also died then with with the new AI stuff. And yet here we are, filming in September, and it somehow hasn't hasn't quite kicked the bucket just yet.”
Note the date: February 2023.
Why hasn't AI killed SEO?
Remember: Google's trying to surface helpful content to the reader. Google wants us to have a good user experience on the search, because if we don't like Google, then we go to DuckDuckGo.
If everybody's using AI SEO software to generate content, Google's gonna have a really hard time differentiating between AI blogs.
You know who is going to be differentiated? The person writing their own blog post.
Now, I don't think that's going to stop people from using AI extensively. It's tempting to just press that generate button and hope that this is going to get your site more hits with very little effort involved.
But I don't think that that AI somehow is going to precipitate this huge revolution. Because we're already in that revolution.
We were already in a race to the bottom. People were already trying to “generate” content with optimization as cheaply as possible, either with content spinners, or underpaying writers, or flat-out plagiarizing.
AI is just the latest iteration of how people try to cheat SEO.
The issue is that this kind of behavior has already caused a bad user experience for search. That's the whole point behind the Helpful Content Update. People were adding search terms like “reddit” and “quora” to their queries to surface content written by real people, not SEO marketers trying to rank for a keyword. Google hasn't yet explicitly dinged AI content, but because of this race at the bottom, people were already finding that a lot of search optimized content was not actually answering their question, or it was hard to trust.
I loved what Kyle had to say about why AI isn't great for the content generation use case: “[AI] is just a fancy article spinner. It cannot really give specifics. It can't give tone. It can't do branding...[W]hile I think it can speed up the process, what I've found is that, well, it might help you do SEO faster. But if you're bad at SEO, it's just helping you do bad SEO faster."
Human-written content is being more heavily promoted in the search algorithm. And people who are using AI to generate content exclusively, who aren't taking care to make sure that they're creating genuinely helpful content, will find their SEO performance suffering.
The future of the AI revolution? Human voices will be uplifted.
My theory of the AI Revolution is that it's actually going to put a premium on real, human, opinionated advice.
That's why my website was not dinged by the Helpful Content Update. When I write blog posts, even when I do it for SE results, I put as much opinion, and thought, and care, and experience into each article. Because if not, what's the point? I'm bored writing that kind of stuff.
That's why I outsourced to AI in the first place – and why I'm relieved to see that there's no point grappling with most AI optimization software, because it doesn't even rank.
There are exceptions. But for the most part, articles that I care about, as cheesy as it sounds, perform best in the search engine. Typically, the less AI I use, the more personality comes through – and that's what ranks as “helpful” content.
Human voices + AI optimization
You've heard this a thousand times already. But to finalize the answer to whether you should use AI for SEO, the future of SEO is a combination of AI and humans.
SEO tasks AI is good at:
keyword research
finding relevant keywords
creating a content brief
Things AI is not good at:
content creation
improving user experience
content marketing
social media
quality content
creating content to match user intent
Getting your content to consistently show up on search results
Personally, I will continue to use AI to generate outlines and find good keywords to rank for, maybe some minor optimization. At least then, even for the really boring keywords, I've got something in front of me rather than that dreaded blank page syndrome.
But AI isn't going to fully replace content writers, or social media experts, or user experience marketers, or anyone—except the people who do extremely bottom-barrel SEO stuff.
If all content is AI, there's not going to be a huge audience for that kind of content anymore. I haven't found a reliable AI detector, but I have found that most people don't really need one. You read a piece of work and form an opinion. Whether or not it's written by AI, most humans don't care for unopinionated, bland, voiceless writing. And that's what AI creates.
There are real downsides to this flood of garbage AI content. But you know what? We were too late for world exploration. We're too early for space exploration. I think this kind of technology exploration is a pretty neat point of history to be living through.
