Hey, it’s Amit here.
Over the past week or so, I’ve seen a lot of people in SEO, PR, and “GEO” circles reacting to a new study from Profound that looked at 27 million AI citations across ChatGPT, Gemini, AI Overviews, Perplexity, Copilot, and Claude.
The headline stat that got everyone talking:
Only 2.6% of earned media citations in AI search came from Tier 1 publications.
The remaining 97.4% came from trade publications, niche blogs, specialist review sites, regional outlets, and vertical media.
A snapshot from the study
That number has triggered predictable reactions:
some people cheering (and sharing their own anecdotal evidence to back it up)
some pushing back
some arguing methodology
some getting defensive about PR as a discipline
I’m less interested in picking sides - and more interested in what this actually means for how brands are spending money right now.
Because the uncomfortable part of this conversation isn’t about whether Tier 1 media matters.
It’s about how much we’re weighting it, and what job we expect it to do.
What the data is really saying (and what it isn’t)
Let’s get this out of the way first.
This study does not say:
Tier 1 media is useless
authority doesn’t matter
PR is obsolete
niche sites magically beat large brands in all cases
What it does show very clearly is this:
AI discovery is overwhelmingly long-tail.
When AI systems are answering:
comparison questions
BOFU prompts
“is X worth it?”
“best alternative to Y”
implementation details
edge cases
They are not primarily sourcing:
front-page news
prestige narratives
brand storytelling
They’re sourcing:
specific answers
repeat coverage
narrow topical expertise
boring but relevant content that exists in many places
That’s not a flaw in the study.
That’s how people are using AI.
One of the better debates sparked by this study centred on authority.
And the most accurate framing I’ve seen is this:
Authority acts as a multiplier, not a substitute.
When a high-authority site:
fills a genuine content gap
answers the question clearly
goes beyond surface-level coverage
…it compounds.
But when it doesn’t?
AI systems don’t “reward the effort.”
They route around it.
That’s why you can simultaneously observe:
big brands ranking with mediocre content
and long-tail sites winning citation volume at scale
Both things can be true at once.
This also explains why, in regulated or institutional spaces (health, policy, government, standards-heavy B2B), AI systems often prioritise:
regulators
academic sources
official bodies
That’s not media tiering.
That’s ground-truth weighting.
Different job. Different inputs.
Where PR budgets quietly drift out of alignment
Here’s where I think the real issue sits — and this is where the conversation gets uncomfortable.
Most PR programs are not billed by tier.
Agencies don’t charge “per Forbes link” or “per Bloomberg hit.”
That’s not the problem.
The problem is how brands internally allocate effort, expectation, and perceived ROI.
In practice, many comms strategies still:
overweight Tier 1 pursuit
treat prestige as a proxy for discovery
assume authority will “trickle down” into organic visibility
That assumption used to hold better when:
humans were the primary discovery layer
circulation approximated reach
Google leaned harder on editorial authority alone
AI breaks that model.
Because AI doesn’t experience prestige.
It experiences availability of answers.
A real example from before AI took over the SERPs
This isn’t just theory.
Back in 2024 - before AI Overviews were everywhere - I worked with a client who ran a high-quality digital PR campaign through a well-known agency.
What happened:
They secured a decent number of links
The publications were high authority
The coverage was industry-relevant
The outcomes in terms of organic search traffic, were limited…
What we saw:
A short-term ranking lift (lasting a few days)
An increase in DR
Some baseline support across certain keyword groups (I assume this but hard to isolate)
What we didn’t see:
Significantly increased ranking movements
Meaningful acceleration in organic traffic
Impact proportional to the spend, especially when compared to lower-cost SEO & marketing work
When we stacked this up against:
several months of technical SEO improvements
consistent “lower-tier” link building
broader content distribution
…the PR campaign simply didn’t seem to have an outsized effect on the results.
This wasn’t a bad campaign.
It just wasn’t efficient as a primary lever for organic performance.
That experience is why I’m sceptical when I hear statements like:
“PR links are the strongest links you can get in the world”
OR
“they’ll skyrocket you to the top”
It didn’t do that before AI became dominant in the SERPs.
There’s no strong reason to believe it suddenly will now.
Why this matters in an AI-fragmented world
Traffic is fragmenting.
Clicks are less predictable.
Attribution is fuzzy.
Budgets are tighter.
In that environment, capital allocation matters more, not less.
If Tier 1 media contributes:
a small percentage of AI discovery
at a high cost per placement
with limited compounding
Then over-indexing spend there isn’t a brand-building strategy.
It’s a risk concentration.
Especially when the data - and actual practitioner experience - keeps pointing to the same thing:
AI visibility is an ecosystem game.
Brands show up in AI answers because they:
appear across many relevant surfaces
repeat their association with topics
fill content gaps consistently
exist where answers are being formed
Not because they “won” one logo.
A more realistic way to think about earned media now
This is where I land after looking at the data, the debates, and real campaigns.
What Tier 1 media is best used for
Human validation and trust
Investor and board credibility
Enterprise and late-stage buyer reassurance
Brand reputation and legitimacy (example below)
It can also provide:
baseline authority support
entity reinforcement
But it should not be treated as:
your primary discovery engine
a traffic recovery mechanism
a brute-force GEO/AEO strategy
Where AI discovery actually compounds
Trade publications
Niche and vertical blogs
Specialist reviews
Regional and industry outlets
Boring, specific, repeatable coverage
This is where:
BOFU prompts are resolved
comparisons are formed
AI systems learn associations at scale
The full picture
PR still matters.
So does link building.
So does content.
So does technical SEO.
So do other surfaces like YouTube, forums, and communities.
None of them carry the load alone anymore.
The real takeaway
AI hasn’t killed PR.
It’s exposed what PR was actually doing - and what it was never designed to do at scale.
Tier 1 media is still valuable.
It’s just one piece of a much larger puzzle.
The mistake isn’t doing PR.
The mistake is over-indexing on prestige, and expecting it to solve a discovery problem it’s structurally bad at solving on its own.
I’m curious: If you look honestly at how your brand (or clients) allocate PR and off-page spend today - how much of it is driven by assumed discovery value, versus what actually compounds visibility now?
Hit reply. I read every response.
—
Amit Raj
The Links Guy
P.S.
In fragmented systems, resilience comes from breadth, not bets.