Confused about how to write with AI search in mind? Here are some tips:
There’s so much conflicting information about how to make your writing appear in AI search.
Google recently released its latest guidelines on optimising content: and it says a lot of the advice people give is borderline useless.
This caused a lot of argument in the search engine optimisation (SEO) world. Although many agreed, others said it contradicts what Bing (run by Microsoft) recommends and that it doesn’t take into account what might work for non-Google AI search tools.
As someone who’s been working with, and teaching people about, online content for over two decades now, I have my own take on Google’s recommendations and how they affect you. So let’s go through them now:
Don't use the terms GEO or AEO
Google says acronyms such as GEO (Generative Engine Optimisation) and AEO (Answer Engine Optimisation) are redundant and that SEO is enough to cover everything.
To me, this makes sense. SEO has always evolved to encompass new developments and so it doesn’t make sense to suddenly invent new acronyms for it. And yet, like it or not, the term GEO is now used so much (including by Bing) that it’s also hard to just ignore it.
My take as a professional editor: most acronyms are obnoxious anyway, so this disagreement simply gives me yet another good reason to primarily use the phrase “AI search” instead of GEO or AEO.
Write like a human without stressing about keywords
SEO has traditionally been about keywords. You have broad keywords, such as “thriller novels” as well as far more specific long-tail keywords (such as “Thriller novels set in small Australian country towns filled with bikies”).
Yet AI’s large language models (LLMs) don’t parse your query by matching keywords. Instead, they try to work out the meaning of your query by looking at all the other words you typed to understand your query as a whole (the technique behind this is called “self-attention” – you can find a good explanation of how self attention works here.
This means you don’t need to stuff your site with as many different variations of the same term as possible: instead, you simply write what you mean.
However, this isn’t as new as it seems. Back in 2018 – way before ChatGPT was even launched – Google began using a neural matching AI technique to understand the meaning behind queries rather than just relying on specific keywords. People who say the days of keyword matching are over are way behind the times.
My take? As an editor, I’ve always believed you should write like a human being, for human beings. I've had multiple clients tell me they hired an SEO consultant in the past only to find that while their search traffic increased, their sales (and other conversions) fell: which isn't surprising. After all, when you make the user experience worse (such as by keyword stuffing or making a page five times longer than it ought to be), it's not going to work out well.
Content chunking is not crucial
Google says content chunking, which is one of the most common suggestions for optimising for AI search, isn’t necessary: which goes against what everyone else has said about AI search up until now.
Content chunking refers to the practice of breaking your content up, such as by using short paragraphs, and making sure the key information is high on the page. For example, if your web page talks about how to fix a leaking tap, don't start with a history of plumbing or by waffling on about how frustrating a leaking tap is: instead, just put the answer right up the top.
But Google says its AI search can understand multiple topics on a page and that its AI pulls information from its core search engine.
My take? Large chunks of text aren't easy to read and, unless you're a brilliant writer working on something creative, you should always put key information up the top (this is called frontloading) so you don't waste readers' time. In other words, even if chunking isn’t necessary for AI search, you should still break content up (such as through meaningful subheads and paragraph breaks).
Don’t stress over LLMs.txt files
LLMs.txt files are a relatively new addition to the SEO world. They are files you upload to your website’s root directory, with the idea being they’re meant to tell AI search tools what’s most important on your site.
However, there's a lot of disagreement about whether or not they work. While some claim AI search tools do use LLMs.txt, others say at this stage that’s not the case (at least not officially) and that it’s still just a proposed standard, not a working one.
Google’s latest guidelines firmly agree with the latter group, with Google saying LLMs.txt is not important.
My take is that even if LLMs.txt doesn’t do much (or anything) for now, it also doesn’t do any harm either. Furthermore, LLMs.txt files are easy to create if you have a basic understanding of code (I made one for my website in less than 10 minutes): so if you can, it’s worth doing. And if you can’t: don’t lose sleep over it.
Don’t stress over structured data either
Then there’s structured data.
This is code that’s added to your site that can give extra information to Google and other search tools.
For example, imagine you have a recipe blog. In that case, you could add structured data to each recipe that provides details such as how long the meal takes to cook, what rating out of five your users have given it, and so forth. This information can then appear on Google’s search results page in the form of rich results: so when people are searching for recipes, they’ll see the star ratings, time to cook and so forth.
Sounds good, right? Well, it is … and it isn’t. Sometimes Google will show the data in its results: sometimes it won’t. It depends on the type of product or service you’re offering, who you are and so forth.
Furthermore, Google says it’s not required for AI search, but that it “helps with being eligible for rich results on Google Search”.
My take: have a look at how your competitors appear in search. If they have rich results appearing and you don’t, you should definitely add them. If your competitors don’t – and that’ll be the case with many organisations – then it’s not make or break. However, as with LLMs.txt, it won’t do any harm: so if it’s easy enough for you to add then you should. Just don’t expect it to magically transform your SEO.
Make your writing helpful, reliable and targeted at people, not machines
I’ll finish this article by saying my favourite recommendation from Google’s latest guidelines is to: “Focus on developing unique, expert-led content that provides value beyond common knowledge.”
Which, as any good writer or editor will tell you, has always been the basis for good non-fiction writing.
However, this is now even more important than before because AI search won’t cite you if you’re stating general knowledge. It might, however, cite you if you have something unique to say.
And, if you think about it, this is what we should have been doing all along.
