SEO is dead and ChatGPT killed it — or so they’re saying. Is it true?
While AI-generated content can result in short-term gains, over the medium-to-long-run, websites that rely on generic AI content will tend to suffer large traffic losses.
In this article, Jonathan Finegold, Ignite Visibility’s Digital Strategist, digs further into these points and concepts below, focusing on E-E-A-T and why content differentiation is key in 2023.
What You’ll Learn:
- Google’s Opinion on AI-Generated Content for SEO
- Why Human-Driven Content Differentiation Matters More Than Ever
- 10 Ways to Leverage AI as a Tool for Ideation
- Some Final Thoughts on AI-Generated Content
What Google Says About AI-Generated Content for SEO
The world’s premier search engine has an opinion or two on AI-generated content.
Technically, AI-generated content is not against Google’s guidelines.
However, Google’s main objective is to provide the best experience for users by delivering quality search results, based on search intent.
To follow through with its mission, Google focuses on rewarding high-quality, reliable and helpful content with top-ranking positions.
Unfortunately, AI-generated content isn’t always accurate. It’s not guaranteed to be original either.
Here’s an example: say two companies use the same prompt to ask ChatGPT to create an article on the best email marketing tools. Chances are, ChatGPT could produce very similar content for both queries. Since Google values unique and helpful content, these companies can expect lower rankings in search engine results.
Summarizing:
- Google’s main objective is to reward high-quality content, regardless of the production process.
- Quality content demonstrates experience, expertise, authority, and trust, also known as E-E-A-T.
- Google has dealt with mass-produced low-quality content for a long time
- It can recognize spam, including AI-generated spam
Rule of thumb: produce people-first content that showcases your experience and expertise on a topic.
Always Defer to Google’s Quality Rater Guidelines
Website performance on Google search is algorithmically determined. While there is a certain irony in that Google also vets the performance of its search algorithm against human input.
It does this by hiring teams of humans to rate content across the web against the Search Quality Rater Guidelines. In case you’re not keen on reading a 170+ page PDF, there is a shorter overview document — it’s only 36 pages long.
These guidelines tell us what the ideal piece of content looks like (and what it doesn’t look like).
And although the guidelines don’t have a direct impact on search rankings, they often tell us what will. Updates to the guidelines tend to precede updates to the algorithm.
Pro-tip: the guidelines include a changelog at the very end.
Key Takeaways
- Deceptive and/or harmful content will receive the lowest rating.
- Your-Money, Your-Life (YMYL) topics — e.g. finance, healthcare, politics, etc. — will be more strictly graded against the criteria.
- Content quality takes into consideration the amount of effort put into making sure the content satisfies the searcher, the originality of the content, the talent and skill of the creator, and its accuracy.
- The reputation of the content creator matters and for best performance, the creator must show thought leadership across different platforms: their website, LinkedIn, guest blogs, etc. It’s also beneficial to have relevant credentials and training.
- Content quality is shown through E-E-A-T: experience, expertise, authority, and trust. Does the creator have first-hand knowledge and experience? Is the information accurate? And does the creator have a broader recognition as an authority on the subject?
The highest quality content shares these characteristics:
- The purpose of the page is beneficial and has no potential for causing harm.
- The accuracy of the information is carefully scrutinized and of the highest possible standard.
- The content shows originality, talent, and skill, and conveys first-hand and/or trained experiences.
- The author and website domain are associated with a very positive reputation.
- The page title (title tag and H1) accurately summarizes and reflects the content.
- There is adequate information about the content creator.
- There are no ads or other elements (e.g. interstitials and pop-ups) that significantly interfere with the consumer’s experience.
Does Google Matter Less Now? What About Bing?
ChatGPT spurred competitors to push their own products to market, initiating the “AI Wars.”
Microsoft invested heavily in Open AI, making Microsoft Open AI’s exclusive cloud computing provider. In exchange, Microsoft is extending Open AI’s technology across its own platforms, including Bing.
Google recently launched Bard in response.
While I’m sure that both technologies will continue to see development and that more alternatives will pop up, the truth is that, to date, Bing’s OpenAI tech is a step above Google’s.
Interest in Google Bard has seemed to have dropped off after very short spurts in press coverage.
Does this matter? Does this have any implications for SEO?
Big picture: no.
The purpose of a search engine is to show a human the highest quality result.
Google has invested a lot of resources into defining and measuring what high-quality content is, and so it’s one of the best guides available on how to create content that will rank well regardless of the search engine.
The main thing to look out for is the diversification of search engine usage. Per Statcounter, Google is still the top dog and has in fact seen a slight growth in the share of search since the beginning of 2023:
This is worth keeping an eye on as Bing hits new user milestones, anyway.
Getting back to the main point: Google’s definitions of quality are still the most important.
Why Human-Driven Content Differentiation Matters More Than Ever
One thing is for sure: AI has lowered the average cost of generating content, more of it will flood the market than ever, and AI-generated content will become increasingly saturated and devalued.
With more content out there, the more important it is to differentiate your content. Let’s talk about how to do that from an E-E-A-T perspective.
E-E-A-T Criteria: Only Humans Have the Experience to Bring a Unique Angle
Tools like ChatGPT can respond to prompts in a way that mimics an expert or authority, as long as the prompt is crafted well.
It mirrors this voice through what it can take from its database and predictive analysis. The output can be good — certainly impressive for a non-human content generator and even better than a lot of the generic human-generated content out there.
Again, content spam isn’t new and humans were never very good at it, anyway. It’s not surprising that ChatGPT can produce decent generic, predictable content.
Excellent content — content that stands out — isn’t predictable.
Your experience and expertise, mixed with creative thinking, give your content an advantage over the AI competition. Only a human with experience and expertise on a topic can:
- Share recent, diverse, and previously unpublished examples, case studies, and data which are true. AI, of course, can make anything up. But content that’s factually incorrect doesn’t tend to rank well.
- Push the envelope on what is known about a topic, including the provision of analysis versus regurgitating known facts.
- Analyze a topic from a unique, fresh perspective.
- Relate the content to what their readers truly care about.
AI Can’t Do SEO Content Strategy
The quality of a specific piece of content is only one piece of the puzzle. The bigger picture is how the piece fits in with the overall content strategy. There might be a place for AI within the overall strategy, but AI can’t replace the strategy (or the strategist).
AI can also disincentivize good strategy. It could, for example, lead to content overproduction and dilute the average quality of your content, the trustworthiness and authority of your domain, and your website’s perceived relevance to the search query.
None of this means that AI, like ChatGPT, doesn’t have a place. That place is under the supervision of a human who is curating the content, keeping it faithful to an overall strategy, and giving it the E-E-A-T elements that only a human can offer.
AI-Generated Content is Factually Unreliable
Another issue with AI-generated AI is factual accuracy. AI has access to a database of information, and part of the solution to factual accuracy is the quality of the database.
When that database doesn’t have the answer to a question, the AI’s language model has to predict what the answer is. This is where things can go awry.
The example that always comes to my mind is when ChatGPT was asked for the most cited economics paper of all time.
Its answer was pretty credible at first glance: “A Theory of Economic History” by Douglass North and Robert Thomas. “Economic” and “theory” are the two most cited words in popular economics papers, and the most likely word to follow from those two is “history.”
Who is the most cited economic historian? Douglass North. The average economics paper has two authors, too. ChatGPT’s answer was therefore pretty smart.
It’s also blatantly wrong — that paper doesn’t exist.
Getting basic facts wrong has been a major, documented problem with ChatGPT. ChatGPT has even fabricated entire news events. Google Bard famously erred on its first public demo.
These confidently outputted, but factually incorrect responses to basic questions are known as AI hallucinations.
Getting basic, known facts right is just part of accuracy, truth, and credibility. What has set humans apart is our ability to explore the unknown and provide new information.
What the above examples show is that when AI is needed to cover gaps in knowledge, they tend to provide inaccurate answers.
Pushing the envelope, providing new insights and examples with a dedication to the facts, and doing a better job at engaging your readers are what defines excellent content — and AI like ChatGPT or Bard can’t do these things for you.
Sites That Rely on AI-Generated Content Tend to See Big Losesses in Search Traffic
The fact that AI-generated content can’t truly achieve excellence in quality hasn’t stopped people from trying to use AI-generated content to rank for keywords and acquire organic traffic.
What has been the verdict? After all, the proof is in the pudding.
Caveat: the following data is often anecdotal and is not a systemic analysis of all web properties that rank on Google. These examples are often also devoid of context to protect the domain’s and brand’s identity. There is certainly more data to collect and a larger picture still coming together as our experience with ChatGPT- and other AI-generated content grows.
All of that being said, there is certainly a lot of evidence against the use of AI-generated content for SEO.
Gael Breton, the founder of Authority Hacker, analyzed Bankrate and Cnet AI-generated content against their expert-written content and found that human content outperformed its AI-generated counterpart by a very, very significant margin — up to 776%.
The length of the AI-generated content didn’t make a difference:
Furthermore, even when AI-generated content earns organic positioning and traffic, the effects are often short-term and then drop back precipitously. Consider these examples:
See Code+SEO’s full AI-site graveyard thread here.
The evidence suggests that AI-generated content does not typically outperform content written by a human expert.
Even when AI-generated content ranks and drives organic traffic to a domain, the benefits are very short-lived and disappear almost entirely.
Relying on AI-generated content is simply not a sustainable SEO strategy.
10 Ways to Leverage AI as a Tool for Ideation
Although ChatGPT and tools like it can’t outperform expert human writers, that doesn’t mean that these tools don’t have a place in our SEO workflows. There are some pretty creative ways of using ChatGPT and like-tools to assist in research, editing, and directional guidance. I’ll cover a handful of examples — and caveats:
- Blog topic recommendations — always check against keyword data from a reliable tool such as SEMrush and make sure you’re not cannibalizing keywords from an existing piece of content.
- Content gaps — ask ChatGPT to find semantically related topics that are missing from a specific piece of content. As always, double-check against the resource for accuracy. In the example below, ChatGPT recommends including content on email automation when the analyzed page clearly already discusses that topic.
- Internal linking recommendations — ChatGPT is great at recommending related articles to link to. An expert can help you determine which internal links would drive the most value and what the optimal anchor text should be.
- Keyword recommendations — AIs can’t give you actual keyword search volume data unless they are pulling it from a third-party database. The recommendations are directional and can give you ideas to fuel your own research when you pull search volume data from tools like SEMrush. If you use the right prompt, you can also leverage AI to help find synonymous and semantically related keywords for further inspiration.
- Cluster keywords based on semantic relevance and search intent — AI can help you group relevant keywords together when you’re researching for your content strategy. There will be some manual clean-up and reorganization, but AI can help at least get you started.
- Create Unique Headlines – You only have a few seconds to capture your reader’s attention and get them to click on your webpage. Use AI to help you generate a set of headlines that are both relevant and attention-grabbing.
- Generate FAQs — AI can generate good FAQs to answer. As an expert, you will need to clean up the recommendations and add your own, as you know your audience and what they care about best.
- Email Marketing Copy – AI tools can help you speed up your email marketing process altogether. While it’s clear you should not rely on AI-generated content, it’s a great place to start gathering some ideas. You’ll need to revise and edit to make sure the copy is consistent with your brand voice and tone.
- Social Media Copy – When prompted correctly, AI can generate catchy sentences that will drive your audience to click on your content over competitors. Use AI to help you craft your captions, but beware of spelling and grammar issues and always be sure to fact-check.
- Paid Ads Copy – If you have a difficult time summarizing the value behind your deal, AI can help you. As a marketer, relying on AI to do the hard work will give you more time to ensure creative is up to standard.
A common theme in every case is that the recommendations need to be reviewed and turned into actionable insight by an expert. AI cannot replace the expert, it can only help automate some of the more routine tasks and help make the process faster.
These recommendations are just a handful of ways you can leverage AI as an SEO assistant. While I focused on ChatGPT, there are so many other AI-powered tools available to you.
Be Careful With What You Expose to AI
Everything that’s available to you through AI like ChatGPT, Bard, and any other tool is available to its other users. It’s important to be very selective about what data you feed into it to condition the results. Never upload proprietary data or information unless you’re ready for that to be accessible to the broader public.
Pro-tip: You can block plugins that leverage ChatGPT to crawl your site by disallowing the ChatGPT-User agent on your robots.txt file:
Think About the Larger Conversation Around the Topic You’re Writing On
Traditionally, we’ve thought about SEO and marketing in terms of stages of the marketing funnel: awareness, consideration, and conversion. And a comprehensive SEO strategy will create content for each of these stages within specific verticals of your industry and market.
If you sell gas grills, for example, you’ll want how-to content to establish expertise in your space, draw high-funnel visitors, and make them aware of your brand.
You can then use mid-funnel content, like an industry study, best-of guide, or infographic, to segment users who are in the consideration stage.
You’ll drive users in this segment to your transactional content. This still matters, AI might just change how users move through these stages.
The entire conversation around a purchase might now happen through an AI assistant on Bing or Google.
Conversations are fluid and can jump around. Asking Bing AI to recommend a good gas grill between the $500 and $1,500 range might prompt it to recommend the Weber Summit S-670 6-Burner Natural Gas Grill. You might then ask if Weber grills are hard to maintain if they are warranted, and which vendor has the best shipping and return policies.
As of the middle of 2023, AI search assistants don’t seem good enough to have productive conversations around these questions in a way that saves the consumer time by aggregating data from multiple websites.
For example, asking Bing AI about who has the best return policy only gives me information from the Weber website. I can’t get Bing AI to compare the return policies of the top 10 online gas grill sellers.
The experience is bound to improve, and quickly, if there is demand for this type of search assistant.
Final Thoughts on AI-Generated Content
- AI-generated SEO content can be decent, but it can often be factually incorrect and it’s quite generic. AI-generated content is not excellent.
- Excellent content is what it takes to own the top positioning on Google and other search engines over the long term and in a sustainable way.
- Excellent content is written by experts and contains unique insights, data, and interpretations. It shows the author’s experience and expertise.
- AI can be leveraged as a research and assistance tool in the content-production process.
AI is also rapidly improving. It will change digital marketing.
AI can help scale your email marketing through AI-driven personalization, especially by influencing the way we segment our email lists.
Chatbots will become much better as they incorporate new AI, giving your business an additional tool for engaging with prospects and qualifying leads before they reach your human team.
Machine learning is already used to improve digital analytics, through improved attribution modeling and predictive insight. And machine learning algorithms are already at work in your Google Ads and other paid advertising platforms, allowing you to focus on big-picture strategy while the AI optimizes the tactical landscape.
AIs can be leveraged to improve marketing performance.
FAQs on AI-Generated Content
1. Should you use AI to write all your blogs and website copy?
No. Rely on AI to get you started on the creative process, help you build content outlines, and write enticing titles. You can use AI to help you brainstorm, however, you should get into the habit of fact-checking AI-Generated results.
2. Can Google detect AI content?
Yes, Google can detect AI-generated content through its machine-learning algorithm. If your content is created for people first, it won’t affect your rankings. However, Google has been detecting low-quality content and spam for years. So make sure the intent behind your content is to deliver high-quality, valuable information.
3. Is all AI-generated content bad?
No. But if you’re not fact-checking AI-generated content or making edits, you could be in trouble. The bottom line is to stay away from poorly written content, so edit, fact-check everything, and follow E-E-A-T guidelines.
