A client asked me recently how I had optimised my copy for AI search. Given the current sound and fury about AI in the SEO industry, I’m surprised it took so long. But that question shows it’s time to share my thoughts about AI search and copywriting.
For everyone who doesn’t want to read the entire article, the bottom line is:
I prioritise traditional SEO over AI SEO.
It’s not laziness, or lack of knowledge.
It’s my assessment of the market and what will work for my clients.
So, what’s driving my assessment?
- Google and traditional search still dominate
- AI overviews appear in about half of search results.
- Getting into the AI overview is more about your off-page SEO than your on-page SEO.
- The AI overview overlaps with traditional search, but offers wider opportunities.
- AI overviews barely appear for local search.
- The recommended copywriting tweaks for AI Search are all about ‘extractable’ content. Which you should have been doing anyway.
- ‘Semantic relevance’ vs ‘keyword density’ is a furphy.
One point before I dig into the detail behind those statements: I’m focused on AI search and copywriting for my client base. That is, small-to-medium service businesses, often business-to-business, and often serving a ‘local’ market. (ie a city, region or metropolitan area).
If that’s not you, some parts of the analysis may not apply.
1. Google and traditional search still dominate
Here are some top-level numbers for activity:
- ChatGPT handles around 2.5 billion prompts per day
Source: OpenAI via Axios - Google processes over 5 trillion searches per year. A tad under 14 billion searches per day
Source: Google
Google wins hands-down. Even before you consider the number of ChatGPT prompts which are not search-related – for example, creating copy or images.
What about traffic to websites?
Ahrefs have a neat little site which reports on traffic coming from search engines and from AI assistants. Here’s a chart from February 2026:
Add up the traffic from all the AI assistants – ChatGPT, Perplexity, Gemini, Copilot, Claude and Mistral – and you get 3.3 million visits. Coincidentally, about the same as the traffic from DuckDuckGo.
Meanwhile Google is delivering 385.5 million visits to the same sites.
3.3 million visits vs 385.5 million.
Does that feel like a good reason to switch how you do everything?
2. AI overviews only appear in about half of search results
Once upon a time, the top result on a Google search page was a ‘snippet’. It featured content from one website, and owning the snippet was a major goal for SEO. (For those with deep pockets!)
Now, snippets are rare. Instead, approximately half the time, we have the ‘AI Overview’. This works quite differently:
- It has content from multiple websites, with links for those who care to follow them. (Me. I’ve always liked to know my sources.)
- The AI overview is generated from scratch each time someone searches. (Snippets weren’t.)
The websites cited are constantly changing. 40-60% churn in AI overview mentions is common.
What does that mean?
- It’s ‘easier’ to get into the AI overview, but less advantageous because you don’t ‘own’ the overview exclusively.
- It’s also far more likely that you’ll get rotated out.
Most importantly, aiming for the AI overview is only relevant half the time (or less, depending on your business).
3. The AI overview overlaps with traditional search, but offers wider opportunities
Bright-Edge have been tracking overlap between the AI overview and traditional search results. Their data makes for interesting reading.
The AI overviews are created by AI running searches. Overlap is not surprising. But AI doesn’t search the same way humans do. (What would be the point of AI search if it did?) It has more capacity and speed. It doesn’t get bored. AI searches more deeply and more thoroughly in a very short time, to surface the most relevant pages.
- 54% of sources in the AI overview rank in traditional search results. (That is, they appear in the top 100 search results.) This percentage is growing for all industries except e-commerce.
- Only 16.7% of sources rank in the top 10 results. (The old ‘first page of Google’.)
In traditional SEO, you need to be in the top three results to drive significant traffic. To get actual clicks to your website, which might turn into leads or enquiries. That’s a winner-takes-all model. Most people lose.
But the content which makes it to the AI overview is far more diverse.
Over 37% of AI overview sources rank in traditional SEO, but not highly enough to generate traffic.
That’s a massive opportunity for businesses like my clients.
Businesses which don’t have the budget of the corporates and multinationals to score the top three results. Quality content, optimised for traditional SEO, can be ‘buried on page 2 of Google‘ and still make it to the AI overview.
Why wouldn’t you optimise for traditional SEO?
4. AI overviews barely appear for local search
‘Local search results’ are the pages you get from searches like ‘conveyancer near me’ or ‘Hills District physio’ or ‘family lawyer sydney’. Usually, they include the classic ‘Google Map Pack’:
Many (if not most) of my clients serve customers within a specific geographic area. (Your physiotherapist or florist needs to be nearby. Your lawyer or accountant could, in theory, be on the other side of the country, but most people want someone they can meet in person if required.)
These clients are much better off working on local SEO so they can show up in the map pack. An active and positive Google Business Profile, plus website copywriting laced with geographic references, will help them. Worrying about AI Search won’t.
5. Getting into the AI overview is more about your off-page SEO than your on-page SEO
Just to make sure we’re all on the same page before we look at this:
- On-page SEO is what happens on your website. You have control. Copywriting, Keywords. Meta data and alt text (in the code, not visible to the average user). Site speed.
- Off-page SEO is what happens external to your site, and is far harder to control. It’s about who links to you and how much authority they give you by doing so. (A link from the government or a major university is more impressive than one from your local butcher.)
AI Search – appearing in the AI overview or in search results within an AI platform like ChatGPT – is skewed towards off-page factors. The AI algorithms focus on what other people say about you, rather than what you say about yourself. Reputation is everything. Reviews, citations, links from respected journals, institutions and so on.
The copywriting on your website is self-promotion. Copywriting is simply less important to AI search algorithms than it is to traditional search algorithms. If you want to appear in AI search, consider investing in an off-page SEO strategy. Or PR.
6. Copywriting recommendation for AI Search are about ‘extractable’ content.
Which you should be doing already anyway.
AI wants to quote you in a response it compiles. It loves easily extractable content, including
- bulleted lists
- FAQs with 1-2 sentence answers
- definitions of terms
- numbered processes
- comparison tables
So do humans. So does traditional SEO.
Once in my life, without trying, I scored an SEO snippet. (I was pretty smug – I beat out Hubspot and other multinationals for a good while there!) I did it quite by accident – I used numbered headings to help human readers break the content into digestible chunks – and Google liked them too.
I use bulleted lists, FAQs, comparison tables and more for the same reason – so human visitors can digest the content easily.
If your copywriter isn’t doing this already, you have the wrong copywriter.
The main copywriting ‘tweak’ I can think of based on the recommendations is this: – consider a page with a glossary of industry jargon. Then you’ll get cited when people are looking for definitions.
But before you do that, ask yourself: Are the people looking up industry jargon the same ones who are going to enquire and become customers? More visitors are only worth it if they’re the right visitors.
7. ‘Semantic relevance’ vs ‘keyword density’ is a furphy
The AI Search brigade often raise the issue of ‘semantic relevance’ rather than ‘keyword density’.
They’re living in the past.
Yes, traditional SEO was built on keywords. Yes, keywords matter. But so does semantic relevance. Google has been using semantic analysis since at least the Hummingbird update in 2013. That’s 12 years ago, folks!
We’re talking evolution here, not revolution.
Evolution is much more common, but revolution is scarier. And in revolutions, people who’ve got something risk losing it. All those businesses which have paid a fortune for SEO already? If they think they have to rip it all up and start again, they’ll pay more than they would for a ‘tweak’. Guess who’s hyping up the ‘AI Search Revolution’?
Summing Up
This seems like a good time to review a few fundamentals:
- Traditional search operates on algorithms. So does AI. They are both evolving.
Not that different. - AI search finds its information using search algorithms. It may find, scan and process far faster than you do, but it searches the same data set, with largely similar tools.
Not that different. - Relevant, clear, helpful content is more popular. For AI search. For traditional search. With humans.
Not that different.
And, looping back to my opening sentence, let me finish by misquoting Shakespeare.
The AI Search ‘revolution’ is a tale
Told by a hungry service provider, full of sound and fury,
Signifying nothing.
Me? I’m going to continue writing relevant, original and clear content for my clients and for myself. Because that’s what their end customers want, so that’s how I deliver value.