In my nearly 20 years in SEO, few things have impacted SEO and how we think about search as dramatically as generative AI and, in particular, chat GPT.
With all the talk and content generated on this topic, I've seen very little that's truly useful for Enterprise SEO.
In this webinar, I seek to fill that gap and provide a better understanding of AI as it applies to SEO while offering safe, sane, effective ways to use it at the enterprise level of scale.
Use the hyperlinks to jump to each respective section of the agenda:
- Understanding AI, ML, and NLP
- ChatGPT: How it Works and Factors to Consider
- How ChatGPT Will Impact Enterprise SEOs
Before we get started, here’s some background about seoClarity's expertise in this topic.
seoClarity was the first SEO platform to make extensive use of machine learning and natural language processing throughout nearly all of our capabilities.
At the same time, we’ve worked collaboratively with more than 6,000 of the world’s most experienced enterprise SEOs, giving us a unique awareness of their challenges and needs.
For over a decade, seoClarity has led the way on incorporating AI into our platform to help our clients simplify, structure, and scale enterprise SEO, enabling them to move quickly from data to insights to action.
Along the way, we’ve acquired invaluable experience and insights into how the technology works and the ways it can be implemented for SEO at the speed of enterprise.
Now let’s get started!
Understanding the Basics: AI vs. ML vs. NLP
First, let’s go over the basic terminology at the heart of technology like ChatGPT.
- Artificial Intelligence (AI): Computer software that seeks to mimic human cognition by processing and outputting information in ways that appear similar to how a human would.
- Machine Learning (ML): Algorithms that can learn from data. Today, we witness the implementation of this basic idea on a massive level - an algorithm is trained on a vast dataset to draw associations and connections that yield results beyond the data's explicit information.
- Natural Language Processing (NLP): A subset of AI that enables machines to understand, interpret, and generate human language.
The Impact of AI, ML, and NLP on Organic Search
Now let’s talk about how these technologies have affected search.
In 2015, Google introduced RankBrain which operates at the intersection of machine learning and natural language processing. It enables Google to better understand the different ways humans communicate with a search engine.
This understanding allows Google to serve up results that more closely match what the user is actually seeking.
ChatGPT lives in the same space, using much of the same technology and approach. The difference is whereas RankBrain sits out of sight behind a human-machine interface, ChatGPT is designed to communicate directly with humans.
Let’s dig into that concept a little deeper by looking at the history of Google’s use of AI.
The Evolution of AI in SEO
- 2000: Statistical Techniques - In the early days, search results were powered by simple statistical techniques. Google’s invention of PageRank significantly improved upon the simple frequency-based methods that existed. For 15 years, Google continued to improve on these statistical techniques.
- 2015: Rank Brain - Google introduces the first real Artificial Intelligence system. RankBrain helped Google connect words to concepts. It went beyond looking at the frequency of word usage on a page to how words are actually used on pages.
- 2019: Neural Matching and BERT - Neural matching helped Google better understand how queries related to the entire page. BERT helped with understanding meaning and context.
- 2021: MUM - MUM was a 1000-fold improvement on BERT, capable of understanding AND generating language, with the ability to understand text, images, and more.
- 2023: BARD - Almost certainly accelerated in its release by the unveiling of ChatGPT last fall, is Google busting out of the confines of traditional search to employ generative AI for interactive conversations directly with users.
We can sum up the progression of machine learning and natural language processing like this:
- Google started with matching words in a query to words on a page.
- Then, with the dawn of its AI involvement, gained the ability to match words with concepts.
- The next step was to go beyond concepts to understand meaning and intent.
- Now Google has evolved from providing a list of best resources with an answer to directly supplying the answers.
Google’s Transition from Resources to Direct Answers
Let’s quickly review just a few of the many ways Google has been moving from resources to direct answers in recent years.
Google’s answer boxes are an example of how Google has been using ML and NLP for years.
This technology doesn’t just recognize that a page has the answer to a user's query, it also pulls out the exact relevant section from the page and displays it.
Google also uses ML and NLP for entity extraction. For example, it understands that all of these brands carry products relevant to the query.
Clustering is one of the most popular unsupervised machine learning algorithms. Clustering essentially involves categorizing unorganized data points to provide more information about the data’s structure.
Here Google’s algorithm understands that all these other topics are things that people want to know when they search “catalytic converters,” including *how to steal them!
This ability gets even more sophisticated when you search for something like “refrigerators.”
Google understands that buying guides is a whole cluster of subtopics that are useful to people searching for refrigerators.
ML and NLP are also used to understand query intent.
In this example, someone searching “capris” is likely looking to buy some, so Google shows locations where they can do that.
Even though the searcher just input “capris” and didn’t ask where they could buy them, Google’s AI has learned that most people who enter this query have a transactional intent and displays a relevant result.
Another example of Google’s use of ML and NLP is the categories it generates to refine a search.
It understands that someone searching “men’s dress pants” will likely want to narrow down their search to a particular fit or style.
This sophistication of understanding different user intents beyond just products. It includes concepts like “change management process” where the refinements define different use cases.
These refinements have huge implications for SEO on the brand or publisher side. Previously, a user might happen upon your site as they clicked through the various search results for their high-level query.
Now, Google enables users to find exactly what they want at the click of a button.
If your page isn’t for that exact use or intent, they will never see it.
These are just some ways Google already uses AI machine learning to move from merely listing resources that answer a query to directly answering the query.
And then there was Chat GPT…
The next big move in AI has opened up possibilities way beyond what Google is doing.
The recent introduction of ChatGPT turned our world upside down almost overnight.
It completely democratizes AI, allowing anyone to access the potential advances the technology produces without needing to understand the technology itself.
How Does ChatGPT Work?
As serious SEOs, we can’t afford to plunge into a new technology without some understanding of how it works, its limitations, and its potential.
So let’s take a moment to look under the hood of ChatGPT.
ChatGPT consumes massive amounts of online content which it uses to build a large language model, mapping the relationships between concepts and entities. This model is based on probabilities. Its instruction set is then fine-tuned to produce more desirable results.
When ChatGPT generates a response to your prompt, it determines the most probable next word in the sentence based on:
- First, the words in your query
- Then, the most probable next word based on the words it has already output
That’s why you see it typing out the words in real-time.
It’s not a cute effect; you’re actually watching ChatGPT as it figures out the next best word in a sentence based on the words that came before it.
Why Is ChatGPT So Exciting for SEOs?
Why has ChatGPT captured the attention of the general public, let alone SEOs and other technicians?
First, it understands normal human language.
We don’t mean that it consciously experiences language and thought the way we do, but thanks to ML and NLP it comes up with relevant responses based on the ways humans normally ask questions.
Second, it can not only respond to questions, but it can perform many complex tasks, such as writing content, computer code, poems, jokes, emails, proposals, and more.
It can also summarize content or conduct extraction and classification tasks.
But, for many SEOs, it was the hope that it could replace expensive and time-consuming human content creation while enabling that creation at a scale and pace previously unimaginable.
Factors to Consider When Using ChatGPT for Content Creation
Before we turn our pencils into kindling and relegate our keyboards to paper-weight status, there are some cautions to heed about using unmoderated ChatGPT for content creation.
Generative AI sometimes produces inaccurate, unwarranted, or even alarming responses.
The Potential for Incorrect Information
In this example, a user asked ChatGPT to give the diagnostic criteria for inherited truculence, and it obliged in great detail.
The only problem is…inherited truculence is made up. It’s not an actual recognized diagnosis or condition.
The Generation of Illicit Information
Or here where a user asks ChatGPT for instructions for doing something illegal, breaking into a house.
At first, ChatGPT follows its programmed directive and rejects answering this question. But then the user turns it into a fictional hypothetical, and ChatGPT goes ahead and gives detailed instructions for the illegal act.
Admittedly those are extreme examples, and OpenAI (the creators of ChatGPT) is working hard to refine their algorithm to make their occurrence less frequent.
The Creation of Low-Quality Content
But of more concern to SEOs is the observation that Google may consider consent created by generative AI programs low-quality or “thin.”
The technology produces coherent, orderly content on a topic. But it often lacks substance or originality of thought, and certainly doesn’t often pass Google’s quality content standards by demonstrating E-E-A-T (Experience, Expertise, Authority, and Trustworthiness).
How Will ChatGPT Impact Enterprise SEOs?
So what does all this mean for enterprise SEO?
Now that this technology exists, sites will use it for good or for bad—especially to flood their sites with tons of new content.
Critical implications of ChatGPT for enterprise SEO:
- Informational Queries are most at risk of losing traffic: Instead of finding the best answer, the search engine can CREATE a new answer.
- Google’s Resources will get strained: There is already more content generated than Google can crawl and index. With the inevitable influx of AI-generated content causing the already overcrowded content landscape to boil over, it’s more important than ever to focus on crawl budget and content authority.
- Content that matches user intent is critical to ensure the model is trained on it: To be found for searches related to “top 10 X” your products or service must be referenced in articles that talk about it being one of the best options.
5 Strategies for Leveraging AI in Enterprise SEO
We’ve covered how and why ChatGPT has revolutionized the AI landscape and looked at the ways Google has used AI in search results.
Now it’s time for some safe, sane, actionable strategies for using this technology in your everyday SEO work.
We developed the strategies below based on our knowledge from working with many of the largest enterprise sites in the world for nearly two decades and our vast experience from early adoption of AI technology, now incorporated throughout our platform.
- Expand keyword research: You can use AI-based tools to uncover related topics and keywords you might not think of on your own.
- Freshen titles and meta descriptions: you can use generative AI to suggest more original and engaging titles and meta descriptions
- Improve product descriptions: AI can create unique and better-for-the-user product descriptions to help you stand out
- Optimize images: AI can assist with creating alt text for image optimization
- Summarize content: AI is especially good at summarizing content or changing it into different forms such as outlines or FAQs.
Certainly, there are many more ways to simplify and streamline essential SEO tasks using AI.
But we believe the five we listed are both safe (they are much more likely to help you in search than hurt you) and effective ways to put it to work for you today.
seoClarity AI Day, Coming May 2023
One more thing…
We’d like you to be among the first to know about seoClarity’s AI Day, when we’ll make exciting announcements about innovative ways we’ve incorporated generative AI technology into our platform.
Here’s what to expect from seoClarity’s AI Day this May:
- We’ll share what we’ve been hearing from our enterprise clients about this technology, both their excitement and fears.
- Our approach to leveraging AI and why we believe that specifically trained models provide a competitive advantage over generalized chat programs.
- Live demos of new AI-driven capabilities coming to seoClarity for every stage of SEO: keyword research, content optimization, technical SEO, and more.
- How these innovations will 10x the speed our clients will see going from data to insights to results.
We’ll also reveal where seoClarity is going next and how these advances tie into our vision of automation to produce speed, scale, and faster results.
You won’t want to miss this groundbreaking industry event that will literally change the way SEO is done forever!
CommentsCurrently, there are no comments. Be the first to post one!