TL;DR
Google AI Mode is out in the wild and available to test in the US. We've been doing just that, to better understand how things might change for us, as digital marketers, in the next three to six months.
There's no denying it's a massive evolution in how search works. Sure, Google has changed quite a lot over the past 25 years. But the fundamental concept of search hasn't.
You enter a search query, you hit the search button, you get served a list of results. Rinse. Repeat.
Algorithms have become more sophisticated. SERPs have gained (and lost) features.
Voice search has been talked about (see what I did there…) but has never fully taken off. But the fundamentals have remained.
However, with the launch of AI Mode, there's the promise of search becoming more conversational, more personalised and with simple one-and-done queries being replaced by prompt-based sessions.
In its current form, it's nowhere near what it'll eventually become - Google has said as much, with the launch of agentic shopping experiences (where users can buy straight from search, no website session necessary) mooted for the future. And in truth, from our initial experiences with it, it’s essentially an extension of what we’ve been seeing in AI Overviews.
For now, all we can do is speculate on how it might shape the search landscape.
There have been lots of interesting viewpoints from across the industry - from the panicked cries of 'content is dead, SEO is dead, long live content, long live SEO!' to the excitement of the challenge ahead.
We fall in the latter camp. We don't believe SEO or content will die. And no, not because it makes us feel warm, fuzzy and less concerned about the adaptation for a new era of search.
It’s more because we genuinely believe the fundamentals of good content and SEO will remain just as important as ever. When you strip SEO back to its core, it’s about structuring and signposting information so it can be ordered and ranked effectively.
Just as well structured, formatted and marked up content provides clear signposting for traditional search algorithms, these same principles are likely to be valuable for LLMs (large language models) in AI search.
This new mode of search fundamentally changes how information is aggregated and presented, placing an unprecedented emphasis on modular, high-quality content.
It promises to be less about relying on keyword mentions, but rather semantic relevance and coverage of topics.
We can best understand this concept through the analogy of toy building bricks.
One of the features of AI Mode that has grabbed my attention the most is its potential to conduct multiple searches at once. ‘Fanning out’ is what Google has termed it – even going as far to patent ‘Thematic Search’.
It captured my imagination and got me thinking about the similarities between someone (theoretically, at this point) using AI Mode and someone building using toy building bricks.
Let's break each of the elements into a cast of characters or roles:
He knows what he wants to build with the toy building bricks; he's just not quite sure which bricks he needs or how to go about it.
This is similar to the Searcher. That’s you. That’s me. We have our questions or a desire to gather information about a specific topic.
The Boy has an idea of the final model he wants to see built. He's got this in the instruction booklet. He could be super clear, providing a very structured brief. Or he could be much looser - giving greater freedom for interpretation.
Just like the Prompt the Searcher enters into Google AI Mode.
You need toy building bricks to build a model.
Sure, they come in lots of different shapes, sizes and colours. However, what makes them so amazing is that they're modular - you can use them in any number of combinations and formations.
For example, some of the same building bricks that could build a queen's castle may also be used to construct a model of an F1 car. They're just assembled in different ways.
When broken down into modules or components, web content is extremely similar. You might have:
Just like toy building bricks, each of these content modules could function independently. Or they could be assembled in a multitude of combinations to make something completely different.
The key takeaway here for SEOs and content writers is this:
The Master Builder’s role is to interpret the instructions provided by The Boy. He's got the knowledge and experience to go out and source the bricks needed to build the model.
He'll sift through multiple bricks, making judgement calls on whether they're the right fit or of the right quality.
For example, some of the bricks might have been chewed by the dog, gone through the vacuum cleaner or have been snapped in half. Using his judgement, he'll look for the best quality bricks available from multiple sources.
The AI Agent's unique ability, akin to a Master Builder who can simultaneously browse thousands of brick bins across an entire store, is to 'fan out' - conducting multiple, interconnected searches across the web at once.
Once the Master Builder has finished assembling the bricks to build the model, he returns it to The Boy in its completed form.
However, the Master Builder's work might not be done there. The Boy may decide he wants him to tweak certain elements. He may want to tweak the shape or colour of the bricks used. Or he may be perfectly happy with what he's got.
The Final Model represents the AI's initial synthesis: the answer to the Searcher's prompt. But just as The Boy might want to swap out a green brick for a blue one, the AIgenerated response is often a starting point, not a static endpoint.
This interactivity signals a future where content isn't just consumed, but actively shaped and refined by the AI based on continuous user input. It could make the depth and versatility of your content 'bricks' even more critical for subsequent queries and deeper explorations within a single session.
…is less likely to be selected as a useful ‘brick’ by the AI Agent.
It’s no longer a case of looking for that one perfect article. It’s efficiently identifying the most relevant and highest-quality content ‘bricks’ from disparate sources to construct the best response.
Ultimately, we believe that the biggest adaptation will be in the way content is approached. Not necessarily in the value it provides its target audience – if anything it’ll be about providing more.
It all comes back to signposting - making understanding what’s on the page easier.
For example, LLMs and AI crawlers often don't execute JavaScript extensively. So, if you’ve got key content hidden or only rendered client-side (without server-side rendering (SSR) or static site generation (SSG)), you could miss out.
SSR and SSG help to make sure content is rendered in the HTML before the crawler sees it. If it’s not there, the content may not be accessible or indexed properly – unlike with traditional search engine crawlers, which are better equipped to handle JavaScript.
Schema and Structured Data are likely to be important in signposting to AI agents and LLMs - helping them to get a clear and quick grasp of what it is that’s featured on the page. Who knows… maybe there’ll be new forms of Structured Data markup as (if) Search GPT, Perplexity and other AI search engines establish a foothold.
Content that is well structured, has good depth, solid contextual internal linking (to deeper level guides which explore topics in greater detail) and offers this in a format that’s simple and concise – this is what we think could win the day.
Sound familiar? That’s because these are the core principles of great content - full stop!
This is where we believe businesses are best positioned to bring originality to content on topics that have seemingly been ‘done to death.’ By drawing on the voice of the experts within an organisation, businesses can regularly bring greater originality, as well as expertise and authority to their content.
Keep an eye out for more of our insights on the implications of AI in search and beyond. Want to stay ahead of the curve and future-proof your brand? We’d love to talk - get in touch with us today!