Authored by Andrew Dunn, Practice Lead, Streets Consulting
AI is increasingly being used for search, and this is quietly and opaquely rewriting the rules of digital marketing.
But is it a brave ‘new’ world? Or are we coming full circle, with content strategy returning to fundamentals?
Today we all live in an AI submarine. It is pervading every element of our technological lives. But as this nascent world continues to emerge, new research on the impact of these groundbreaking tools is now pointing in a very familiar direction – you cannot replace the human in the mix.
In October 2025, our blog, “Writing for Robots”, looked at how AI-powered search was beginning to reshape internet marketing. Since then, the shift has accelerated.
Reliable information on these trends is difficult to find, but indications are that AI is increasingly being used as a core way to discover, research and compare information, including how potential buyers are evaluating vendors.
94% of individuals now use generative AI somewhere in their purchase process. In the B2B software category specifically, 51% of buyers now start their vendor research in an AI chatbot more often than in Google, and 71% rely on AI chatbots for software research at some point in the journey.
A Hubspot study, unsurprisingly focusing on CRM software buyers, said that 42% used AI search during vendor evaluation, and AI search activity was the strongest correlate of purchase intent identified in the study.
Before the advent of LLM searching, life was simpler for companies looking to get their wares out to their audiences:
But – things have changed dramatically.
But – if the AI Overview cites your company content or company pages, you can now win back much of that loss. Pages/content cited in the Google AI Overview get 120% more clicks than the other results on the same page.
Here’s what you get for winning a place on the first search results page now
Being on page one of Google used to all-but guarantee visibility. So does making that golden first page of Google search results lead to being cited when people search using LLMs? Short answer: No.
Ahref (the SEO provider) studied over a million AI Overviews and found the share of AI citations also ranking in Google’s top 10 Search Engine Results Page fell from 76% in mid-2025 to just 38% by early 2026. Around 31% of AI citations do not even rank in the top 100 search engine results.
In other words, being on the top ten search results page does not mean an LLM will cite your business; it is a totally separate process with very different rules.
Now, the traffic going to Google search is not showing signs of slowing down. The inclusion of Google’s automatic AI overviews may be playing a role in defending that position. So, this means that SEO remains important, but the new field of Generative Engine Optimisation (GEO) is also coming to the rescue.
In fact, a 2024 academic study suggested that GEO could boost the discoverability of content by as much as 40%.
New, encouraging data sets are emerging
Earlier I asked if the world of LLM search was a brave ‘new’ world, and the answer is: not entirely.
One study shows that AI engines lean heavily on third-party validation rather than a brand’s owned-channel (like websites) content. Between 82% and 84% of AI citations in a recent Muck Rack study originate from ‘earned media’.
This large-scale analysis of over 25 million links across 17 sectors classifies ‘earned media’ as a category spanning journalism, academic research, government data, and community sites such as Reddit and Wikipedia. But the journalism figure stands out at 27% of all AI cited sources.
Advertorial and paid content represent just 0.3% of AI citations in this study. Moreover, the research claims that this is consistent with similar figures from this time last year, indicating that the data is agnostic of new LLM model releases.
The more things change… the more they stay the same
What’s interesting to track as this new discipline of AI search deepens its reach is where it seems to be leading us.
When AI first burst into the mainstream a couple of short years ago (granted, it feels like it was 20 years ago), fears understandably grew alongside it that AI would replace writers and researchers, and that the ease of producing content would flood the market with low-quality noise. As awareness of AI hallucination grew, we worried we could no longer trust what we were reading, and in an earlier blog we discussed the need for AI users to take responsibility for what their tools produce.
For all the impacts and upheavals of AI in the workplace, the growth of GEO, AEO and associated patterns might just be bringing the communications and marketing industry back home, to a place we recognise very well.
A place where quality, accuracy, clarity and usefulness are crucial.
A place where external validation from respected media or academic outlets, partners and trade bodies represents even more of a verification, arguably than it ever has before.
A place where originality, straight-talking and doing your homework properly all count towards an LLM ‘believing you’ and your message cutting through the noise.
Independent third-party endorsement remains key with journalism now being cited heavily by AIs. Given the challenges that the media industry faces in terms of readership and revenues across the board, could their assumed antagonist, AI, actually be pushing their value back up to where it belongs?
And once again, whilst AI can help with workloads, nothing on Earth can do what is required better than a smart human.
Nature’s LLM, the one on top of your neck, is still the best tool in town and training data is no substitute for real experience and expertise in making sure that AIs know what’s what in the real world.
46-48 East Smithfield, London, E1W 1AW