HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

A Practical Guide for

Marketers Who Want to Use
GenAI
Understanding the risks, rewards, and how to get started.
by Oguz A. Acar
Published on HBR.org / December 11, 2023 / Reprint H07XTE

NikolaVukojevic/Getty Images

A recent McKinsey survey revealed that only 10%–14% of companies

consistently deploy generative AI in their marketing and sales
initiatives. Why, in a function that seemingly stands to gain the most
from generative AI, is adoption so limited? And how can marketers
bridge the gap and fully harness this potential? As marketers try to
navigate this transformative moment, it seems many are struggling with
where to begin.

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 1

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

There are four areas where generative AI holds the greatest potential for
marketers, and four areas where it holds the greatest potential for risk.
Here’s what leaders need to know about those areas — and a framework
for how to get started.

The Four Cs of Marketing-AI Opportunity

In my view, there are four key areas where generative AI holds the
greatest potential: Customization, Creativity, Connectivity and Cost of
Cognition.

Customization
Consumers crave authentic experiences that resonate with their unique
preferences and journeys. And generative AI offers an avenue to deliver
this at an unprecedented scale.

Consider Carvana, a digital platform for selling used cars. The company
generated 1.3 million unique AI-generated videos, which it called
“joyrides.” Everything in the video from the visuals to soundtrack was
bespoke. Carvana achieved this just by drawing on some basic customer
data — like the car’s model, color, year, purchase date, and location —
and enriching it with cultural events from that time and place. And the
actual production? It likely took less than half a workday, thanks to their
cloud infrastructure’s ability to process 300,000 videos an hour.

Other brands are experimenting with generative AI to bridge language

barriers and cater to different markets. Spotify, for example, is piloting
AI-driven voice translation for select popular podcasts in which the
podcaster’s original voice is retained.

Generative AI also holds great potential to improve customer

interactions by helping customer service employees to engage in
more customized conversations. A recent study with 5,179 customer
support agents provides a glimpse of this potential; deploying an AI-

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 2

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

based conversational assistant tool to augment agents has significantly

improved the number of individual customer inquiries resolved while
also increasing overall customer satisfaction.

But this is just the beginning. Recent developments in multi-modal

AI technology will further expand the capabilities of AI-driven
conversational chatbots in terms of generating tailored solutions to
customer problems. Imagine a customer effortlessly submitting a photo
of a faulty product for reimbursement and the AI chatbot swiftly
providing guidance on the matter.

Moving forward, a promising frontier is the chatbot personas tailored

to specific customer segments. To get a sense of this potential, just
look at the success of Character.ai — the most popular application after
ChatGPT — which allows users to engage with AI-generated characters
be they fictional or historical creations. Indeed, Meta is launching AI-
driven persona chatbots across their flagship platforms with plans to
enable brands to craft their own versions.

Creativity
A second avenue for harnessing generative AI in marketing is creativity.
While the debates around whether machines can be creative continue, it
is clear that AI can generate outputs that are deemed creative.

Take the findings of a recent study for example, which showed that
ChatGPT4 surpassed the creativity of elite university students in new
product ideation; the majority of the standout ideas were generated
by AI. In fact, such output has even earned creative accolades; an AI-
generated image was the winner of Sony World Photography Awards in
the creative photo category

But the true value of GenAI in creativity is about its potential to

augment human creators. A recent experiment with 453 professionals

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 3

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

who regularly generate written output as part of their job, including

marketers, showed that using ChatGPT raised output quality and
originality by 18%. Likewise, another study showed that having access
to generative AI increased a writer’s creativity significantly, particularly
among less creative ones with improvements of up to 26%.

These gains were achieved with a simple training. Now with new tools
designed specifically for marketing, such as Jasper, it is easier to achieve
creativity gains at a scale for various content creation tasks. Some global
companies like Unilever are even building their own generative AI tools
to write advertising content.

Marketers can also completely reimagine what is creatively possible

thanks to generative AI. A case in point is Coca-Cola’s “Masterpiece” ad
campaign where paintings from famous historical figures and emerging
artists were brought to life with generative AI. The company even
launched an NFT collection based on digital art from the ad, fetching
over $500,000 in just 72 hours.

Connectivity
Generative AI offers new avenues for brands to connect deeply with
customers, facilitate consumer-to-consumer interactions and empower
consumers to have a more active role in brand narratives.

A case in point is the Jen AI campaign of Virgin Voyages where an

AI version of Jennifer Lopez invites people on board. As part of the
campaign consumers can create their own custom invitations using the
Jen AI tool and share with each other, leading to an engagement rate
that is 150% greater than previous campaigns.

A real, but relatively unused, potential of generative AI is democratizing

participation to marketing processes as it reduces technical barriers for
consumers. That is, every consumer can now be a designer, a storyteller,

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 4

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

and an influencer. Using text-to-image models, for example, consumers

can now generate a visual without having any expertise in graphic
design, opening new doors for wider co-creation opportunities with
consumers.

A great example of this is Coca-Cola’s “Create Real Magic” initiative.

Consumers were invited to take part in an image-generation contest
using a bespoke tool enabling consumers to create image from prose.
Winning designs are set to shine on global advertising stages like New
York’s Times Square.

But the benefits of such initiatives are not confined to winners or

even direct participants. As my paper with Stefano Puntoni suggests,
the benefits are wide-ranging, not only for those who are directly
engaged, but also for the broader market by vicariously strengthening
relationships and enhancing purchase intentions.

Cost of Cognition
One of the most transformative impacts of generative AI is its potential
to significantly reduce the cost of intelligence. Marketers can achieve
significantly more, at a fraction of the time and cost.

This transformative impact spans across many cognitive tasks. From

crafting a company blog post and analyzing qualitative consumer
feedback to designing a new advertisement and generating code for a
new web application, generative AI’s potential influence is vast.

Consider the results of a recent field experiment with Boston Consulting

Group (BCG) consultants. Across a range of realistic consultancy tasks
— such as brainstorming new product ideas, market segmentation,
drafting press releases — consultants working with AI showed
remarkable efficiency. On average, they completed 12.2% more tasks and
executed them 25.1% faster. Even more impressive, the quality of their

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 5

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

output significantly exceeded that of their peers in the control group,

being rated over 40% higher.

Some of the advertisers are already starting to use the tool to leverage
cost benefits in advertising. A case in point is WPP, the world’s largest ad
agency, whose CEO noted that the savings through the use of generative
AI in advertising can bring 10 to 20 times savings.

The Four Cs of Marketing-AI Risk

While generative AI presents unprecedented opportunities, it is also

riddled with risks. Four of those are particularly pressing and pertinent
for marketers, which I refer to as the 4Cs of Marketing AI Risks:
Confabulation, Consumer Reactance, Copyright, and Cybersecurity.

Confabulation
Generative AI is known to confabulate — or produce content that is
not accurate — and does so in a very compelling way. It can also
generate content that is biased or contextually unsuitable. Such outputs
can cause misinformed marketing decisions or, worse, tarnish a brand’s
image in the eyes of its consumers.

While these are genuine concerns — particularly in customer-facing

and strategic marketing tasks — there are ways to mitigate these risks.
For example, fine-tuning a LLM with task-specific data or incorporating
in-context learning can enhance appropriateness of AI-outputs and
reduce risk of confabulation. An even more effective approach is to
maintain human oversight over AI-generated content. This not only
allows catching and rectifying confabulations but also aligning outputs
with brand values and objectives.

Consumer Reactance
Generative AI technologies can lead to consumer reactance, especially
when used for marketing outputs that are directly experienced by

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 6

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

consumers — think customer service chatbots, promotional materials,

or AI-created products. This concern is more pronounced in domains
traditionally reliant on human qualities, such as communication, social
interactions, or anything that requires a nuanced grasp of individuality.

In such domains, emphasizing the human aspects — by for example

being marketed as human-like or human-made — could be an effective
countermeasure. When automation takes the lead, being transparent
about the rationale is also key. After all, consumers are less likely to
push back when they believe the driving force behind an AI-powered
solution is about elevating the user experience, not just cutting costs.

Copyright
What does generative AI mean for creative ownership? This is not just
a philosophical question; creative output generated with AI entails
complex copyright risks, which could expose a brand to consequential
legal issues. Yet, mitigating this risk is also possible.

Marketers must remain alert on legal skirmishes that leading AI

developers face. Currently, while content solely produced by AI might
come under copyright protection, when human creativity is infused
with AI-generated outputs, protection might well be possible.

A risk mitigation strategy is seeking out partners with proactive stances

in addressing copyright concerns. For instance, Getty Images developed
a text-to-image tool trained exclusively on its own creative library
and offers customers standard royalty-free license as well as uncapped
indemnification and perpetual, global, nonexclusive usage rights. Other
major model developers such as Google and Open AI also recently
introduced new indemnification policies.

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 7

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

Cybersecurity
Generative AI brings novel security challenges. Not only does it equip
cybercriminals with powerful tools to launch sophisticated attacks, but
also exposes brands to “prompt injection attacks” — where people
can trick chatbots into revealing sensitive data or generating deceptive
content. From marketing standpoint, the threats manifest not just as
tangible financial burdens — though that too is formidable, averaging
a global cost of $4.45 million — but more critically, lasting damage to
brand reputation and trust.

Yet, the recent data from IBM paints a puzzling picture. Managers, in
theory, grasp the gravity of cybersecurity in the generative AI era. But
the real-world application of this understanding remains sparse. The
solution seems clear-cut — while admittedly easier said than done —
stay updated with cybersecurity protocols to safeguard against evolving
threats. Whether it is teaming up with established cybersecurity firms or
building in-house expertise, the moment calls for proactive managerial
action.

The Path Forward

The risks around generative AI are real. But this does not necessarily
justify inaction, nor does it require a blanket approach to generative
AI where marketers either blindly take risks or avoid AI altogether. In
fact, a promising way to navigate this is a nuanced approach for which
marketers systematically identify areas to experiment with generative
AI and areas to wait-and-see.

I recommend a four-step framework — the DARE (Decompose, Analyze,

Realize, Evaluate) Framework — to move forward with generative AI
transformation in marketing.

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 8

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

1) Decompose Roles.
Begin by reframing marketing roles not as uniform entities but as a
collection of tasks. Take the role of a content marketing specialist; it is
a blend of tasks such as content creation, SEO optimization, audience
research, calendar management, performance analytics, collaboration
with media team, and much more.

2) Analyze Tasks.
For every individual task, juxtapose the potential opportunities against
the intrinsic risks. Assign a score, on a scale of 1 to 10, gauging both
the promise and the perils. Then go deeper: detail out what these
opportunities and risks look like. For instance, deploying generative
AI for social media content creation could scale creativity while saving
cognitive overhead, but could be particularly vulnerable to pitfalls like
confabulation or consumer reactance.

3) Realize Transformation Priorities.

Chart these tasks on a 2×2 matrix, contrasting opportunity versus risk.
Your matrix will guide the transformation strategy.

High priority tasks (High Opportunity, Low Risk Quadrant): These

tasks are ripe for generative AI-driven transformation, promising ample
reward with minimal risk. Immediate experimentation should be in
order.

Moderate priority tasks (High Opportunity, High Risk Quadrant): These

tasks are a double-edged sword. Loads of promise, but plenty of
pitfalls too. As such, they require proactive managerial attention. Before
plunging in, explore risk-mitigation avenues.

Low priority tasks (Low Opportunity, Low Risk Quadrant): These are
your wait-and-see candidates. They should be on the back burner of
your transformation strategy. Engage if/when resources permit.

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 9

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

Non-priority tasks (Low Opportunity, High Risk Quadrant): Proceed

with business as usual here. The potential benefits of AI do not justify
dealing with associated challenges.

4) Evaluate Iteratively.
Continually revisit and recalibrate your AI transformation roadmap.

The AI landscape is ever-evolving. New innovations, like recent strides

in multi-modal AI, continually redefine what’s possible. And with
players like Google proposing solutions to challenges like copyright
issues, the risk landscape is equally dynamic. As such, this is not a
one-off exercise.

The Human Side of Transformation

In the whirlwind of strategic transformation, it is all too easy to overlook

the human side of change. Yet this remains as critical for the success of
transformation efforts.

Anticipate a certain level of apprehension from employees. After all, we

humans are intrinsically wired to be wary of change and innovation,
particularly when it casts shadows on job security. Undiluted,
transparent communication is therefore essential. Make your team
realize that this is not about replacing jobs but reshuffling tasks.
Generative AI, in most instances, is set to augment, not substitute. It
is (and should be) less about technology ousting the human touch and
more about reallocating cognitive resources. Even when AI steps in to
tackle specific tasks, it does not mean axing entire job descriptions.

And to truly tap into the potential of generative AI, you need a team
with the right skill set. Skills like problem formulation, exploration,
experimentation and critical evaluation. Whether it is setting the right
tasks for AI or ensuring the outputs align with your vision, it is not

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 10

HBR / Digital Article / A Practical Guide for Marketers Who Want to Use GenAI

possible to successfully transform marketing without equipping your

team for the next chapter.

All in all, the potential of generative AI seems boundless, so too

are the concerns surrounding its implementation. Many managers,
confronted with the challenges and ambiguities, are tempted to remain
observers. Indeed, a recent survey by BCG shows that the majority
of the executives actively discourage the use of generative AI in their
organizations.

But avoiding the generative AI wave might just be as risky as riding it

carelessly.

It is worth remembering this is not necessarily a binary choice of fully

embracing or rejecting this technology. There is a thoughtful path in
the middle, a strategic way forward that does not require blind leaps of
faith.

This article was originally published online on December 11, 2023.

Oguz A. Acar is a Chair in Marketing at King’s Business School,

King’s College London.

Copyright © 2023 Harvard Business School Publishing. All rights reserved. 11

Copyright 2023 Harvard Business Publishing. All Rights Reserved. Additional restrictions
may apply including the use of this content as assigned course material. Please consult your
institution's librarian about any restrictions that might apply under the license with your
institution. For more information and teaching resources from Harvard Business Publishing
including Harvard Business School Cases, eLearning products, and business simulations
please visit hbsp.harvard.edu.