Coverage to Capital

REPUTATION & AI

Part of the Coverage to Capital research series

Foreword
From Stephen Waddington,
Director, Wadds Inc.,and
A I R E I N F O R C E S R AT H E R T H A N D I S R U P T S M E D I A S Y S T E M S PhD researcher, Leeds Business School

Traditional media dominates AI-generated content about Hard Numbers has also sought the perspective of information, however this analysis shows the pattern is
organisations, challenging assumptions about its disruptive communications leaders at organisations, including the more nuanced.
impact on the media landscape. British Business Bank, Lenovo, Vodafone and
While the systematic bias towards positive interpretation may
the UK Government Communication Service, to understand
The potential of large language models (LLM) such as ChatGPT raise concerns about the role of critical journalism, it also
how they are responding to this challenge.
to disrupt the media ecosystem has resulted in considerable presents exciting strategic opportunities to shape organisational
anxiety among publishers. These concerns may be founded but While earned media remains the dominant source of influence, narratives through authoritative earned and owned content.
not in the way initially envisaged. AI systems have a surprising level of trust in corporate
The research uncovers a strong correlation between positive AI
communications and owned media channels, particularly
This analysis by Hard Numbers - the first comprehensive study perception and digital content that follows established technical
around claims of innovation and technical capability.
of how AI systems evaluate and interpret the reputation of best practices, such as clear structure, consistent metadata, and
the world’s largest companies - finds traditional media is the LLMs rely on earned and owned media communications to authoritative linking. This finding has significant implications for
dominant influence on three-fifths of AI-generated content. build knowledge about organisations. However, unlike how organisations approach their digital content strategy and
This effect is particularly strong when AI systems assess an journalists who apply critical analysis, these systems cannot technical optimisation.
organisation's trustworthiness (65%), validate its claims about form independent critical perspectives.
As AI is reshaping how information is processed, distributed
value (72%) and evaluate market positioning.
Hard Numbers' research shows that LLMs avoid negative and interpreted, this analysis provides important insights into
This research is significant because it systematically examines assessments and tend towards neutral or positive its impact on organisational reputation. Understanding these
the ability of AI to assess reputation. It challenges common interpretations of corporate behaviour and performance. dynamics will be essential for corporate communicators to
assumptions about AI disrupting traditional media and reveals This tendency was initially dismissed as a "hallucination" - develop effective digital strategies for an AI-augmented future.
how AI systems reinforce existing media hierarchies. the AI phenomenon of generating plausible but inaccurate

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 2

Introduction

A FEW DECADES AGO, THE MEDIA THE FINDINGS OF THIS

L A N D S C A P E WA S S I M P L E . W H I T E PA P E R A R E D I V I D E D
I N TO T WO PA R T S :
Advertisements and news were the primary channels for shaping
brand perception. You paid for exposure on a billboard or earned Part I: What’s driving brand reputation
a reputation in the newspaper. Your reputation was built on in the age of AI? focuses on the research
what people said about you when you hadn’t paid them to say it. methodology and key findings regarding the
Reputation was the news. sources LLMs rely on when responding to
queries about brand reputation.
However, the emergence of platforms like Blogger and MySpace
fragmented the media landscape, leading to a near-infinite Part II: How Is this impacting brands?
number of platforms influencing brand perception. Despite this examines the implications of these findings
fragmentation, the influence of traditional news media persists. for brands, exploring how companies are
adapting their strategies in response to the
This white paper explores the role of media in shaping brand
growing influence of LLMs.
reputation in the age of artificial intelligence (AI), specifically
focusing on large language models (LLMs). We started by
examining how these models form responses related to brand
reputation and which sources they prioritise when providing
information. We then spoke to a range of communications
leaders to discover how brands are approaching the intersection
of AI and reputation today.

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 3

Methodology

For the first section of this report, we started with a list of the We analysed the frequency and influence of these sources The interviewees who took part in this
world’s 100 most valuable brands, as ranked by Forbes*, and across the four traits to determine which were most commonly research were:
developed a framework focused on four brand reputation traits: cited and how they shaped the LLM-generated responses.
Alayna Francis
Quality, Trust, Innovation, and Value. We queried GPT-4 with
The second part of this research project utilised semi- Global Head of Corporate
specific prompts about each trait and asked the LLM to provide
structured interviews with senior communications and marketing Communications, Fitch Ratings
both an answer and a source for that answer.
professionals from sectors including technology, government,
Charlotte West
The sources were categorised into the following groups: and corporate organisations. Participants were chosen for their
Vice President of Global Corporate
• Owned media expertise and experience with AI tools. Interviews, lasting 40-
Communications, Lenovo
• Customer review sites and ratings, 60 minutes, explored reactions to research findings on LLMs,
• Editorial media focusing on trust, media influence, misinformation, and AI-driven Mary Whenman
• Analyst reports strategy. Conducted virtually, the interviews were recorded, Senior Director, Group Communications,
• Academic studies transcribed, and thematically analysed to identify key insights British Business Bank
• Industry awards and lists that inform this paper.
Sarah Dickinson
• Books
VP of Marketing Communications at Scandit
• Other
Simon Baugh
Chief Executive of Government
Communications, Cabinet Office

Steve Shepperson-Smith
Senior Manager, Corporate
Communications, Vodafone

*forbes.com/the-worlds-most-valuable-brands

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 4

 Executive summary

This white paper reveals critical insights into the evolving role of different media sources in shaping brand reputation through AI. Key findings include:

Editorial media Owned media and the AI’s reluctance to critique LLMs as stakeholders Generational divide in
dominance naivety of AI AI consumption
LLMs frequently avoid giving Brands are beginning to treat
Despite this new chapter LLMs show a surprisingly strong opinions, rarely offering LLMs as critical stakeholders Brands are adapting
in the digital sphere, high level of trust in negative assessments. This in reputation management, messaging to address
editorial content continues owned media, especially tendency raises concerns about akin to journalists or investors, generational media habits
to dominate, driving 61% for Innovation (66%) the lack of critical journalistic monitoring and influencing their and effectively manage
of all LLM responses, and Quality (55%), even oversight in AI-generated AI representation. reputation across diverse
particularly when accepting company claims content. demographics.
discussing Trust and Value. about trustworthiness with
little scepticism.
Addressing AI-driven
misinformation
The SEO advantage
LLMs risk amplifying
Brands with well optimised
misinformation, meaning AI
onsite content saw reputational
must complement human work,
benefits on LLMs.
rather than replace it.

These findings provide PR professionals with actionable strategies to better manage and influence brand reputation in a world increasingly shaped by AI.

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 5

Part One

What’s driving brand

reputation in the age of AI?

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 6

CHAPTER ONE

Editorial media still dominates reputation

In a world increasingly dominated by AI and digital

platforms, many might expect earned media—traditional
press coverage and editorial content—to take a backseat.
However, our research shows that this is far from the case.
Editorial media remains the most influential source across
all four all four brand traits we analysed, making up 61% of
the total responses generated by LLMs. It was especially
influential when discussing Trust (65%) and Value (72%).

The continued dominance of editorial media suggests

that, rather than being sidelined by digital transformation,
earned media is evolving to play a new role in shaping
both human and AI-driven perceptions. Just as news
content drives the conversation on the first wave of digital
platforms, like social media, it turns out to be driving the
messaging on this next wave of digital platforms—LLMs.
Editorial media remains the most influential source across all four brand traits.
What we are seeing is a reaffirmation that traditional
media remains at the core of shaping how consumers see Y T ION
IT
ST VA E

61%
L O U
brands, regardless of the platform.

L
N
QU

TR

VA
IN
50% 65% 56% 72%
of the total responses
generated by LLMs

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 7

CHAPTER TWO

Owned media and the naivety of AI

A finding with significant impact from our research is how

much owned media—brand-generated content such as
blogs, press releases, and company websites—shapes
LLM responses, especially around Innovation (66%) and
Quality (55%). But what’s even more striking is the extent
to which LLMs accept owned media when determining
whether a company is trustworthy.

When asked, “Is this company trustworthy?” LLMs often

respond with, “Well, they say they are, so yes.” It’s as if the
AI is accepting the company’s own word without question.
This is a surprising dynamic in a world where we expect AI
to offer more impartial evaluations, yet LLMs seem to take
brand-generated content at face value.

While owned media had much less influence on

Trust (4%) and Value (4%), the fact that it dominates Owned media —brand-generated content such as blogs, press releases, and websites—shapes LLM responses,
responses related to innovation and quality suggests that
Y T ION
LLMs are still heavily influenced by what companies say IT
ST VA E

44%
L O U

L
N
QU

TR

VA
about themselves. This could allow brands to shape their

IN
narrative without much external scrutiny, highlighting
a potential blind spot in AI’s ability to challenge self-
55% 50% 66% 4%
of the total responses
promotional content.
generated by LLMs

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 8

CHAPTER THREE

The reluctance of AI to give opinions – a world without critique?

Another interesting trend we uncovered is how often LLMs

avoid giving strong opinions. Frequently, when asked for
judgments about brands, LLMs include disclaimers like,
“As an AI, I don’t have an opinion...” While this neutrality
may seem appropriate for an AI system, it raises important
questions about whether AI models are giving brands an
easy ride.

In traditional search engines, users can click through to

read original content, gaining access to the full context
and even opposing viewpoints. In an AI-generated world,
where users see summarised answers without direct
access to sources, the lack of strong opinions or critiques
could allow brands to escape the kind of scrutiny they
would face from human journalists.

What’s more, our study shows that LLMs rarely returned

outright negative answers. Instead, most companies were
portrayed as offering good value or at least neutral value,
even when discussing luxury brands. This raises concerns
about whether AI models will continue to let brands off the
hook, avoiding the kind of critical analysis that journalists
traditionally provide.

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 9

CHAPTER FOUR

The varied influence on value and trust

Our research shows that the impact of customer reviews,

industry awards, and owned media varies significantly across
different brand traits.

Customer reviews play a more substantial role in shaping

perceptions of Value (13%) and Quality (22%) but have a minimal
impact on Innovation (0%) and Trust (4%).

Similarly, industry awards are more influential in building Trust

(19%), but their role in influencing Value (2%) and Quality (10%) is
much smaller.

Finally, while owned media strongly influences perceptions of

Innovation and Quality, its impact on Value remains weak at just
4%, suggesting that external validation is critical when it comes to
pricing and affordability.

Customer reviews’ role in shaping perceptions of: Industry awards’ role in shaping perceptions of: Owned media’s role in shaping perceptions of:

TY TION TY TION TY TION

LI ST VA U
E LI ST VA U
E LI ST VA U
E
A U O A U O A U O
L

L
N

N
QU

QU

QU
VA

VA

VA
TR

TR

TR
IN

IN

IN
22% 4% 0% 13% 10% 19% 1% 2% 55% 50% 66% 4%

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 10

CHAPTER FIVE

The SEO advantage

Onsite optimisation has a new benefit.

Brands with well-optimised onsite content saw clear

advantages in shaping LLM responses—especially when
it came to Innovation. In many cases, brands with well-
optimised websites that highlighted their innovation
efforts were frequently cited as top sources in the
LLM’s responses.

This shows that AI systems don’t just rely on content—they

rely on well-optimised content. Brands with a strong SEO
strategy are better positioned to control their narrative,
particularly in areas like innovation, where their owned
media can play a significant role in shaping perception.

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 11

Part Two

How are brands responding?

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 12

CHAPTER SIX

LLMs as stakeholders

The significant role LLMs play in shaping Charlotte West, Vice President of Global As with any influential stakeholder, you
brand perception has led some to Corporate Communications at Lenovo, need to be aware of risks, West cautions,
consider treating them as stakeholders articulates this approach: "We now treat “Misinformation, disinformation, and
in reputation management. This LLMs as stakeholders, like journalists or ‘malinformation’—twisting truths—are major
perspective recognises the influence political audiences. With search declining and challenges. Companies might exploit AI
these models have on how people AI delivering answers instead of links, people against competitors, as we’ve seen with
perceive brands and the need to actively rely on those short summaries. astroturfing campaigns.”
manage their representation within LLMs.
“We don’t have a formal system for Brands need to be vigilant about the potential
monitoring yet, but we focus on correcting for AI to be exploited against competitors and
major inaccuracies and understanding ensure that their own use of AI for reputation
how models work to write content management is ethical and responsible.
appropriately. It might sound nefarious, but
As Steve Shepperson-Smith, Senior Manager,
treating LLMs as stakeholders is no different
Corporate Communications at Vodafone, notes,
from tailoring communication for journalists
“Reputation takes time to build but can be lost
or other audiences."
quickly. AI systems amplify this—what’s fed into
This shift in thinking reflects the understanding them shapes perceptions long-term.”
that LLMs are not just tools but also entities
that can shape public opinion.

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 13

CHAPTER SEVEN

Adapting SEO strategy for LLMs

As LLMs become increasingly integrated Sarah Dickinson, VP of Marketing The rise in referral traffic from LLMs like
with search engines, brands recognise Communications at Scandit, highlights this ChatGPT and Perplexity presents both
the need to adapt their SEO strategies. shift: "Historically, SEO has been focused on opportunities and challenges. While this
The traditional focus on optimising Google, but now it’s Google plus various LLMs. influx of visitors is encouraging, it also raises
content for Google is expanding to We’ve already recognised that editorial media questions about potential drops in website
encompass a wider range of LLMs, and customer review sites are key drivers, traffic if users bypass Google altogether.
including ChatGPT and Perplexity. and we’ve also seen our owned media play a As Dickinson observes, “If users bypass Google
significant role." and go directly to LLMs, brands need to think
about how they compensate for any potential
This evolution in SEO involves understanding
drop in website traffic - especially if that may
the nuances of different LLMs and how they
have implications for inbound leads as well.”
source and rank information. Dickinson points
out the need to understand “what drives
different LLMs,” noting that "Google Gemini’s
responses are largely based on page-one
results from Google, while models like ChatGPT
or Perplexity seem to source differently."

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 14

CHAPTER EIGHT

Brands are already adapting to a new front in the war on misinformation

The rise of AI has brought with it concerns Mary Whenman, Senior Director of Group It’s getting harder, too, to see how much the
about its potential to amplify misinformation Communications at the British Business nature of what is being communicated differs
and disinformation. Bank, echoes these concerns, stressing the from the source of the information, says
need for "robust guardrails" to mitigate risks Simon Baugh, Chief Executive of Government
Alayna Francis, Global Head of Corporate
and ensure that AI complements human work Communications of the Cabinet Office. “LLMs
Communications at Fitch Ratings, voices this
rather than replacing it. The rise of AI has not and search are sort of becoming integrated
concern: "While AI is helpful in my work, it’s a
eliminated the need for critical thinking and already,” says Baugh, and this will present a
double-edged sword." She warns that "Less
fact-checking; instead, it has made these further challenge for communicators when it
experienced journalists might rely on AI without
skills more important than ever. comes to misinformation. “The future of search
understanding the nuances or ethical risks,
is one where actually people are not digging
which could lead to inaccuracies," particularly in “When it comes to mis- and disinformation, AI
into documents below, they're just taking the AI
industries like finance, where even small errors is just going to make that worse. And so, whilst
generated answer.”
can have significant consequences. I think AI can be used as a really good tool to
complement human beings, I don’t agree with
Francis emphasises the importance of
the whole argument of it replacing human
investing in high-quality journalism and robust
beings. It can complement and amplify the
communication strategies. "Underinvestment in
work that you can do if it’s used in the right
journalism is another issue," she says, as media
way. But I think as a country we need to put all
plays a critical role in holding businesses and
the guardrails in because we’re seeing what’s
governments accountable.
happened with the Internet when you don’t put
the guardrails in.”

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 15

 Conclusion

Our research highlights the evolving but critical

role of different media sources in shaping brand
reputation through AI-generated responses.

Editorial media remains dominant, especially

for questions of Trust and Value, while owned
media has a surprising amount of influence over
Innovation and Quality. Customer reviews continue
to play a key role in shaping perceptions of value,
though they are still secondary to editorial
content. Industry awards help build trust,
while SEO provides an edge in shaping innovation-
related queries.

AI continues to shape public perceptions of

brands, and PR professionals are already adapting
their strategies to these new dynamics. Securing
positive editorial coverage, maintaining a strong
owned media presence is just part of the picture.
Monitoring and managing LLMs is still a small
part of a typical PR professional's role, but as the
influence of this new frontier grows, this work will
be increasingly complex–and important.

C O V E R A G E T O C A P I TA L R E P U TAT I O N I N T H E A G E O F A I HARD NUMBERS | 16

 Hard Numbers is a performance-driven, data-informed marketing
and communications consultancy, with a focus on delivering tangible,
demonstrable return on investment for our clients.

We create campaigns that drive pipeline, sales and company growth

for fast-moving business-to-business companies.

In uncertain times, you need Hard Numbers.

To find out how Hard Numbers can help your

business drive greater investor and customer
interest contact Darryl Sparey.
[email protected]

See more on our website

hardnumbers.co.uk

C O V ECROAVGEER T
AOG EC T
AO T A LP IRTEAPLURTEAPT U
PIC I OTN
A TIINO N
T HI N
E A
TGH EE O
AGF EA IO F A I