The Intermediary – March 2026 - Flipbook - Page 84
T E C H N O L O GY
Opinion
The search engine is no
or over 25 years, there
was one unwrien
rule on the internet:
produce great content,
rank well on Google,
receive traffic. The value
exchange was clear. Website owners
understood it. Marketers built entire
strategies around it.
That rule has broken. Customers
are now fully immersed in chatbot
experiences. They ask a question, they
get an answer, inside the platform,
without ever clicking through to a
website. The traffic doesn’t arrive.
The session ends. Somewhere in that
invisible exchange, a buying decision
gets shaped.
This isn’t a gradual change. It’s a
structural one. McKinsey research
published in October 2025 found that
half of consumers now intentionally
seek out artificial intelligene (AI)powered search engines, with a
majority saying it’s the top digital
source they use to make buying
decisions.
By 2028, $750bn in US revenue is
projected to flow through AI-powered
search. For any business that relies on
trust and visibility, which is virtually
every business, the implications are
significant.
F
Biggest shift in 20 years
I’ve been in organic growth and
SEO for over two decades. I oversaw
the shi from desktop to mobile. I
watched the first wave of voice search
arrive in 2018 and largely fizzle. I’ve
seen algorithm updates wipe out
publishers overnight and watched new
players build entirely new audiences
from scratch.
This is different. By a considerable
distance. In Q4 2022, ChatGPT
launched GPT 3.5. Days before that, I
was on a podcast saying I’d never use
AI for content creation. I was wrong.
Within weeks of using it, I understood
that the shi was real, and that the
question was no longer if things were
changing, but whether you set up now
or tried to catch up later.
82
The Intermediary | March 2026
ChatGPT now has more than 900
million weekly active users, with 5%
on paying subscriptions. But make no
mistake, Google is still the dominant
force in search and has accelerated
hard into AI with AIOs, AI Mode and
Gemini all forming part of its toolkit.
Alphabet’s acquisition of DeepMind in
2014 was a long-term bet that has paid
dividends.
The reason they weren’t first to
market with a consumer AI product is
straightforward: when your existing
search business generates over $100bn
a year in revenue, you have limited
incentive to cannibalise it.
AI search, or generative engine
optimisation (GEO), is an entirely
new behaviour layered on top of
that landscape. And it operates on
completely different rules.
What is GEO?
GEO is not a replacement for SEO.
That needs to be said plainly, because a
lot of content out there says otherwise,
and it’s wrong.
SEO fundamentals, technical
health, content quality, domain
authority, structured data, carry
directly across into AI search
performance.
If your content ranks well in
Google, there’s a reasonable chance
it becomes part of the pool AI models
draw from. That foundation maers
and shouldn’t be abandoned.
But here’s where the similarity
ends: ranking well in traditional
search does not guarantee visibility
in AI search. The overlap between
highly-ranked pages and AIcited pages is smaller than most
marketers expect.
Obsero’s own research, analysing
almost 200,000 citations across
ChatGPT responses in the UK, found
that niche and specialist publishers
account for 25% of all citations – the
same share as brand sites.
Traditional and mainstream media,
despite dominating most PR plans,
contribute only 15%. Reddit, despite
all the noise, sits below 2%.
ANDY FRANCOS
is founder of Obsero
The reason is context. When
ChatGPT or Google Gemini answers
a prompt, it runs something called
query fan-out: it breaks the prompt
into multiple sub-queries, pulls from
sources it considers authoritative for
that specific topic, and synthesises
an answer.
‘Authoritative’ here doesn’t just
mean highly-ranked, it means
contextually relevant. A specialist
trade publication covering your exact
category can outperform a Tier 1
national media site on a specific query,
simply because it’s a closer match for
what the model is trying to answer.
What GEO adds is a layer on top of
SEO: understanding how your brand
is represented within AI-generated
answers, not just whether your pages
appear in traditional results.
McKinsey’s analysis found that in
major categories – credit cards, hotels,
electronics, apparel – top brands are
absent from some AI-powered search
answers entirely.
Traditional brand strength is no
guarantee of AI visibility. Those are
new questions. And most businesses
don’t yet have answers to any of them.
The measurement
problem
The difficulty with AI search is that
it’s almost entirely invisible. Most AI
platforms have no native analytics.
There’s no dashboard showing
you how oen your brand is cited,
what prompts trigger mentions,
