Whether you’ve become a ChatGPT guru, Bing Search addict or Midjourney dabbler, most of us have had at least a cursory brush with artificial intelligence. And if not directly, all of us feel its effects. Most of all, regardless of our personal beliefs about whether it brings with it an overall good or evil, AI is most definitely here to stay. In today’s episode, we’re going to explore Artificial Intelligence, particularly Generative AI (which means an AI tool that is capable of creating NEW content), in the context of Search Engine Optimisation. We’ll look at how it impacts the way we create content and the way we optimise. We’ll also cover how search engines themselves are changing in how THEY leverage AI to help users find what they need, including release of chat-based AI search tools.
Should we leverage AI in our organic search optimisation activity, and if so…how?
Firstly, let’s set the scene here. Google has explicitly stated that AI assisted content is not inherently against its content guidelines. Google Analyst Gary Illyes describes it as “like any other tool you can use…and misuse.” He advocates using it liberally and innovatively as long as it creates something high quality and helpful.
But does that mean you should use ChatGPT or another tool to wholesale generate content for your ecommerce site. Absolutely NOT.
In fact, in the Google search team’s own experiments with Google’s Gemini AI, using it to generate answers on technical SEO questions, Illyes has raised serious concerns at the inaccuracy of answers supplied. Meaning that it’s critical that AI generated content is fact checked because it could be outdated or simply untrue.
I’ve personally have had more than one conversation with ChatGPT where, unless I explicitly asked to do so, it has simply made up quotes or pieces of writing and attribute them to known persons when it is in fact it’s own generated content and was never produced by the person in question.
But it’s not just about accuracy. It’s about so much more than that.
Back in December 2022, Google added an extra E to its EAT acronym for best practice in the creation of content for websites. So now the acronym stands for Experience, Expertise, Authoritativeness, Trustworthiness. Those are all the things that Google rewards content for if it meets these standards. And outside of the automated parts of its algorithm, this acronym forms an important part of the Search Quality Rater Guidelines which are the guidelines used by actual humans that Google pays to help assess its search results.
The question is…what do these things MEAN in practice?
Dr. Marie Haynes of MHC says this:
“Today I think of E-E-A-T like a template that uses the signals available to Google to create an idea of which content has the qualities of authenticity a searcher would expect for each particular query.”
A search for a medical query will be identified as needing a response from an entity known for their medical expertise.
Now if we apply that to the ecommerce space: a search for a handbag query will be identified as needing a response from an entity known for its expertise in handbags. So if your website sells handbags, and you know the product inside out…guess what…that’s you!
When content is rated low in terms of its meeting of EEAT, its usually for one of these reasons:
* The content creator lacks adequate experience, like a product review written by someone who has never used that product
* The content creator lacks adequate expertise, like a how to article written by someone who has never actually done the thing they’re trying to explain.
* The website or content creator is not an authoritative or trustworthy source for the topic of the page, so like if you sell surfboards and you start an article trying to talk about and rank for mobile phones
* The page or website is not trustworthy for its purpose, so things like a shopping website that appears to have very little customer service or about information
Now lets assume that you’re writing about the products you sell and you’ve used those products extensively and therefore, you meet all those criteria without a worry. The very next question then becomes…how do you SHOWCASE your expertise, how do you bring that expertise to life by creating content that is authoritative and trustworthy?
Probably the biggest way you can do this, is by creating what I like to call an ORIGINALITY PIE. Now that doesn’t mean you can’t use AI to help you brainstorm ideas, or provide feedback on your work, or help find an angle for the use of keywords that you know your ideal customer uses.
Originality doesn’t just mean that you put it in a copy checker and it comes back as being 100% written by a human. You can still have a very generic, very un-unique piece of content written by a human as much as you can by a robot. Whether users can consciously tell the difference is not necessarily the point. Google will prioritise content that—consciously or unconsciously consumers are drawn to, as well as content rated highly by its human search moderators.
And we know that, short of having a direct interface with our brains and siphoning out what we know somehow, it can only work with knowledge that it is trained on. So your unique knowledge is unlikely to be in its coffers. And some experts think that AI company demand for data to train their next generation AI models will outstrip supply in as little as two years.
So with all this in mind, when it comes to writing, these are the core aspects that I encourage both our clients at Foresight, my team at Foresight and our customers at Foresight Academy to prioritise when they write content for an ecommerce site:
The first is Specificity: Think of this as the base of the pie—the crust.
One of the things you’ll notice when you work more with AI generated content, is that its specificity is very low. That is, it’ll spit out a whole bunch of nice sounding words but those words actually say very LITTLE. It’s all fluff. By adding specificity to your writing, you’re creating a clear demarcation. Even if these are things you help feed to help AI edit your work if you need to.
For instance, the exact dimensions of a product, or how many and what types of things could fit into the types of bags you’ve added in this collection or category page. Specificity works best when you have first hand experience with a product, or, you’re interviewing someone who has first hand experience.
The second is Sensory: this layer of the originality pie is the filling.
These are the things that help bring your product to life in a physical way for the customer. They’re what help make it leap of the digital page so to speak. Sight is often quite clear in an ecommerce context through imagery, but what about the taste, the touch, the smell or the sound? Giving consumers a more visceral and immersive experience through language can deepen their connection to your products, and continues to signal the uniqueness and originality of your content to Google.
The third is Voice: this layer of the originality pie is the top covering.
Your writing voice is a really interesting one because this is another area that AI can run roughshod over your originality. If you’ve ever encountered ChatGPT spitting out content that constantly has words like “Elevate” or “Game Changer” or “Skyrocket” you’ll know what I mean.
Now there are ways of briefing tools like ChatGPT to give you, and doing so very often produces a much better score in AI checker tools, BUT you have to actually HAVE a unique author voice to feed the AI in the first place in order for this to work. You have to develop your own tone of voice.
Tone of voice is the term branding agencies use to describe the application of this type of voice at a brand level vs just to an individual person. In fiction writing it’s called the Author’s voice because its attributed to an individual. However you choose to describe it in your business, it’s a combination of not just the types of words you use, but the other things you might weave in to the description. For instance, we have a brand we work with who sells kids lunchboxes, and her natural voice blends not just the product information, but its imbued with her experiences as a mother, and peppered with humour. It’s colouring in your content with the very basis of your life experience. That is something that no one can take away from you, no other brand can replicate
The fourth is the customer. They are the spoon - not a layer at all, but the very implement with which the pie is consumed.
This refers to your is your deep understanding of what makes your customer tick. Your customers want you to speak to them, to their needs. And you know all of this! You know it from your conversations with them via messenger or email in a customer service context and from their reviews as well.
So when you next write a piece of content for your site, I want you to really start thinking about all these aspects of what makes up originality.
* Specificity
* Sensory
* Voice
* Customer
And if you can apply all of that in an accurate way that is factually correct, and you’re writing only on things you have the authority to write about—which is the area of the things you already sell…then you will automatically meet Google’s guidelines and put yourself in the best position to rank.
The final question is…now knowing all this, are there any key areas in which we can use generative AI for website content production? While this is still ultimately an area where the boundaries are still being tested, my perspective is that if you’re doing all those other things, then there is absolutely no concerns with:
* Using generative AI to brainstorm ideas or to more deeply research areas that you need. Just don’t take its writing wholesale.
* Using generative AI to help you suggest natural locations that you might use secondary keywords that you’re wanting to incorporate into a page
* Using generative AI to assess or provide feedback and edits. But again, ensuring
* And in some cases, if you have lots of good, strong examples of your tone of voice, using it to help apply a consistent tone of voice to the writing.
* And you should absolutely consider the use of AI supported tools that help score your content writing as it relates to its ability to rank in the top ten for the keywords you’re focussed on.
Regardless, if you are going to use AI in any kind of generative context, you should always make sure that it is hitting over 95% score in an AI copy checker. Something like Copyscape or Originality AI.
Alright that’s content creation for SEO in the context of generative AI covered. Now let’s zoom out for a moment and look at how search engines themselves have started to use AI in the context of giving consumers more options for how they search.
When someone searches for a keyword…there is SO much we don’t know about what they really mean. Let’s say someone typed in mens shorts; are they a man, or a woman looking to buy shorts for her partner? Do they have preferences around style or brand? Is it a purchase for themselves, or as a gift. None of this is known with a traditional search for a top level keyword.
Imagine that same person conversing with a chat-bot based search engine and telling it exactly what they need, and getting the results that best match what you’re looking for. In theory this is what Bing Chat and other rising AI chat based search engines like Perplexity are trying to achieve. In practice, they’re still getting there and of course search engines like Google who have built a billion dollar empire on people searching in a particular way are naturally having to be far more careful when rolling out changes that could impact behaviour to such an extent that it decreases their ad revenues.
Now I had a go at testing out an ecommerce search using different generative AI tools.Starting with Perplexity AI, I told it:
I'm looking to find some shorts for my brother as a birthday present. He's about 6 foot and prefers longer shorts - somewhere just past the knee. He's in his late thirties so something appropriate but also stylish. Linen fabric would be great.
The results were accurate in terms of what I was looking for, but because I hadn’t specified my location, all I got were American department stores like Walmart. When I asked it to provide Australian options, it gave me stores, but unfortunately no imagery this time. From an ecommerce perspective, understanding that one of the most powerful parts of how people shop and buy comes down to product images will be crucial for these tools to consider in their development.
I then used Bing Chat—also known as Copilot for Microsoft Edge browser. It had the same problem; good results, but zero images, and all UK based stores. After that I tried Chat GPT, which summarised proposed shorts styles well but didn’t suggest brands. When I added a follow up question to provide links to Australian websites and images, it gave collection page links but no images. It will be interesting to see how Open AI’s beta of SearchGPT pans out once it rolls out to its customers because it could position them as a much more direct competitor to Google search.
Google Gemini’s results immediately localised to Australia without any input. But it had no images or links so appears to be actively discouraging people from using it this way. And finally, Brave browser’s Leo AI returned a puny piece of text, no links and no images.
From top to bottom I’d rank Perplexity at the top, followed by Chat GPT, Gemini, Bing Chat, and then Brave's Leo. I think it’s safe to say that from an ecommerce perspective, there is still a long way to go in terms of the way chat based AI is used to search and find products to purchase online.
That being said, with the rate at which artificial intelligence is evolving, it’s difficult to estimate how fast things could change. The only thing that is going to slow down the speed of developments and sophistication in consumers using these tools in an ecommerce context is MONEY. If Google can figure out a way to deliver some version of this without cannibalising its own revenue or its own brand equity in how websites are currently ranked and listed…then this is coming sooner rather than later. The capability is there now, it’s just about the will and the strategy for these companies now.
So here’s what I would suggest you to stay across the development of chat based search. It’s as simple as using it every so often to get a sense of how it’s improving. You can do three key things:
1. First, download Microsoft Edge to get access to Copilot (or download Bing Chat on your phone) and / or sign up to ChatGPT
2. And second, keep Google Gemini, as well as Perplexity as a newer search engine market entrant in your browser favourites.
Original content fuelled by your first hand product experience and knowledge is where you should focus your energy first. Then couple that with strategic use of AI tools. Remember to eat your Originality Pie for breakfast lunch and tea: Be specific, sensorial, channel the customer and bring your own unique voice to the table.
At the same time, keep an eye on how search engines are using generative AI in practice with consumers.
Now you know more about how different aspects of generative AI for content and search impact SEO, what are you going to do differently? Foresight Academy has a complete SEO blueprint for ecommerce brands using Shopify. If that’s you, head to foresightacademy.co/shopify-seo-blueprint to learn more about the course that gives you step by step guidance on getting your store and products ranking where customers can find them.
Until next time, I’m Michelle Bourke, and this is Foresight Academy.