AI adaption in

digital marketing

An investigation on marketers’ expectations from AI, and

the applicable knowledge on search engine marketing.

Main Subject area: Informatics

Author: George Iskef
Supervisor: Martin Lindh
JÖNKÖPING 13/12/2021
This final thesis has been carried out at the School of Engineering at Jönköping
University within informatics. The authors are responsible for the presented opinions,
conclusions, and results.

Examiner: Ulf Seigerroth

Supervisor: Martin Lindh
Scope: 15 hp (first-cycle education)
Date: 2021-12-13

i
Abstract
The following research conceptualizes the applicability of AI in search engine
marketing. Following a rigorous investigation on professionals’ opinions, the
investigation regards their perceptual understanding of the applicable AI practices and
the required level of interrelated knowledge to technical applicability for a more
successful marketing strategy.
Purpose – Digitalization has changed many industrial sectors towards becoming
technologically dependent. These technological advancements are somewhat visible by
professionals and applicable in their industry. The thesis examines AI’s applicability in
search engine marketing and the understanding of professionals in its handling. Based
on research that identifies marketers’ misperceptions of AI to be ineffective in
marketing, and other research that finds technical knowledge in AI as connected to
successful marketing management. This investigation intends to examine marketers’
expectations from AI, and compare that to what AI professionals think that should be
expected. Thus, with the knowledge acquired through the interviews and theory review,
this research intends to summarize marketers’ expectations from AI in search engine
marketing, and the AI knowledge that should be expected from a marketer.
Method – The research methods are qualitative, applying practices such as online
interviews with open-end questions and template analysis to classify the data collection.
On the first level, this research distinguished between marketers’ expectations from AI
and current AI capabilities. In advance, focusing on Google AI in search engine
marketing, to investigate a specific search engine. Finally, establishing an expectancy
of marketers’ knowledge in AI, relational to search engine marketing.
Findings – While technical understanding in AI infrastructures may increase problem-
solving capabilities in search engine marketing, technical proficiency in AI is not
among the primary contributors for successful marketing. The research findings show
a different result from the expected outcome on knowledge requirements. And while
marketers hold high expectations from the automation of search engine marketing
however, they are certain of the irreplaceable human contribution in creating abstract
and strategic development.
Keywords – Artificial Intelligence (AI), Search engine marketing (SEM), Knowledge
management

ii
Table of content
Abstract ............................................................................................ ii
Table of content ............................................................................. iii
1 Introduction............................................................................... 1
1.1 BACKGROUND .................................................................................................. 2

1.2 PROBLEM STATEMENT ...................................................................................... 3

1.3 PURPOSE AND RESEARCH QUESTIONS ............................................................... 4

1.4 SCOPE AND LIMITATIONS .................................................................................. 5

1.5 DISPOSITION ..................................................................................................... 6

2 Method and implementation ................................................... 7

2.1 APPROACH ....................................................................................................... 7

2.2 DATA COLLECTION ........................................................................................... 7

2.3 DATA ANALYSIS ............................................................................................... 9

2.4 VALIDITY AND RELIABILITY ........................................................................... 11

2.5 ETHICAL CONSIDERATIONS ............................................................................ 11

3 Theoretical framework .......................................................... 13

3.1 AI IN MARKETING ........................................................................................... 13

3.2 AUGMENTATION AND AUTOMATION ............................................................... 14

3.3 AI AND DECISION MAKING .............................................................................. 15

3.4 ALGORITHMIC MARKETING ............................................................................ 16

3.5 AI IN SEARCH ENGINE MARKETING ................................................................ 16

4 Results ...................................................................................... 19
4.1 COLLECTED DATA .......................................................................................... 19

4.1.1 Template analysis of the interview with marketers ............................... 19

4.1.2 Template analysis of the interview with AI professors ......................... 21

iii
 4.2 DATA ANALYSIS ............................................................................................. 22

4.2.1 Perception of AI in marketing................................................................ 22

4.2.2 Marketers’ role in SEM ......................................................................... 24
4.2.3 Google AI in SEM ................................................................................. 24
4.2.4 SEM and future development ................................................................ 26

5 Discussion ................................................................................ 27
5.1 RESULT DISCUSSION ....................................................................................... 27

5.2 METHOD DISCUSSION ..................................................................................... 29

6 Conclusions and further research ......................................... 31

6.1 CONCLUSIONS ................................................................................................ 31

6.1.1 Practical implications ............................................................................. 32

6.1.2 Scientific implication ............................................................................. 33
6.2 FURTHER RESEARCH ....................................................................................... 34

References ...................................................................................... 35
Appendix A .................................................................................... 40
Marketers’ Interview questions............................................................................ 40
AI experts’ interview questions ........................................................................... 45

Appendix B..................................................................................... 49
Marketers’ Interview questions............................................................................ 49
AI experts’ interview questions ........................................................................... 50

iv
1 Introduction
Technological advancements continuously change and impact social and industrial
development (Pathan, 2018). As Papadopoulos & Martin (2011) concluded, a few
decades ago, small to medium enterprises (SMEs) were facing difficulties in analyzing
market selection and segmentation due to high costs. Today, digitalization enables
SMEs to research and explore new markets cost-effectively (Papadopoulos & Martin,
2011; Dethine et al., 2020;). While from a marketing perspective, automation has
developed new advertising spaces, entry opportunities and made workflow more
efficient (Rusell & Norvig, 2016).

However, the marketing industry is constantly developing and adapting to innovative

processes. For example, in a commercial investigation by PwC (2017) on AI, the report
estimated 14% higher GDP by 2030 on a global scale due to investments in AI
development. Moreover, Bughin et al. (2018) estimated AI to have created $13 trillion
output by 2030, boosting global GDP by approximately 1.2 percent annually. In similar
research by Fagella (2019), marketing is believed to one of the most prosperous areas
for improvement. Also, according to Fagella (2019), the application of AI in marketing
is found to have the highest profit potentials in the following areas: search, customer
segmentation and targeting, recommendation engines, programmatic advertising, and
market forecasting.

A PwC report by Rayn (2019) finds that AI capability to act based on emerging
variables is limited. The report suggests that AI needs to undergo appropriate data
training. The report also suggests that AI is not capable of autonomously classifying
such emerging variables into new categories and acting on that information without
human input. For example, machine learning is trained based on predetermined
variables to make predictions and classify data set labels appropriately. Data sets
specific to business problems and in line with enterprise standards. The premise of the
Dartmouth conference was that every aspect of learning and intelligence could be so
precisely described that machines are made to simulate it (Epstein, 2015). According to
Marinchak et al. (2018), AI alternates the development of marketing processes. The
author states that marketing strategies and processes are increasingly becoming AI-
augmented and automated processes. There is some controversy regarding scholarly

1
opinions on the future development of AI in marketing. While some believe that
automation is the solution, others think that AI should be augmented and operated by a
human in its bottom line (Jarrahi, 2018; Epstein, 2015).

Overall, AI is of high impact on marketing, and it will facilitate the development of

business opportunities (Kolbjørnsrud, Amico & Thomas, 2016). The imperative is clear
that businesses must integrate AI into their marketing strategy if they expect to keep up
with the competition (Pradeep et al. 2019). This thesis examines the applicable practices
of AI in search engine marketing, and more specifically Google SEM. This decision
has been made based on the popularity of Google with clients and businesses.
Considering a constantly evolving industry and the relevance of AI, the focus is to
investigate effective knowledge adaptation by marketers.

1.1 Background

This chapter distinguishes the characteristics of AI and explains the application of AI

in marketing. Kelvin, Lui & Karmiol (2018) define AI as software that mimics human
behavior by learning and applying data patterns. Moreover, Marinchak et al. (2018,
p.22) summarize that “how, what and to whom to sell, as well as- what, how, and from
whom to buy, is increasingly an all-digital, AI-augmented, and automated process”.
The most used AI framework to solve these questions is machine learning (Kelvin Lui
& Jeff Karmiol, 2018). And while different AI frameworks have been developed to
cope with enterprise modules and systems of operations. Overall, AI is evaluated for its
computational ability to solve problems, continually learn new patterns, and improve
system infrastructure (Rusell & Norvig, 2016).

Moving from the definition of AI into its applicable practices in marketing, research on
international market selection and segmentation notes that the digitalization area is a
fundamental trigger that changed marketing (Papadopoulos & Martin, 2011; Dethine et
al., 2020). Due to technological adoption and new channels that better enable
communication with consumers, firms are adapting their online strategies to reduce
cost, while deriving sales (Faroudi et al. 2017). For example, market research is reduced
to a numeric analysis through AI automation, leading to cost reduction.

In all regards, AI is not simply used for data collection and management, but rather for
its ability to accurately personalize data and create a more interpersonal relationship

2
with customers that highly amplifies marketing position and strategy (Google AI, 2020;
Maghsudi, 2021, Gentsch, 2018;). For example, businesses can use algorithmic
personalization for their strategy to differentiate from competitors and target a separate
niche (Gentsch, 2018). Either way, successful marketing is defined as being capable of
reaching consumers effectively, persuading them to connect, and in advance convert
into sales (Pradeep et al., 2019). Also, marketing should be cost-efficient and not waste
money on practices that do not generate financially positive value (Pradeep et al., 2019).
The utility of AI can be considered successful in its marketing implementation based
on the above scope. But since AI has very broad applicability in the field of marketing.
For this thesis, the investigated theory considers the applicable properties that AI has
on search engine marketing or in short SEM. SEM is a strategy used to increase the
visibility of a website on search engines.

Finally, a distinction should be made regarding the applicable utility of AI in marketing.

Due to constant technological development, there is a shift in the responsibilities of
marketing managers (Wedel & Kannan, 2016; Whitler et al., 2017). And while AI
automation reduces the workforce of marketers, many practitioners and scholars
envision AI augmentation rather than AI automation (Jarrahi, 2018; Epstein, 2015). The
main distinguishing factor is that AI augmentation enhances the operators’ capacity
rather than operating individually, while AI automation means that machines take
independent decisions and act based on machine learning to bring results (Jarrahi,
2018). Based on this, the research investigates Google for the way that AI is being
utilized, and interviews marketing specialists and AI specialists on their perception of
AI in search engine marketing. Google AI is used as a core example because of its
popularity, and availability of tools (Heitzman, 2017; Google AI, 2021). To that end,
the research examines Google AI for its automation capacity in search engine marketing
and then reflects on marketers’ expectations of its capabilities. This will increase the
understanding of AI in search engines and facilitate potential growth towards the most
suitable growth area of knowledge.

1.2 Problem statement

Marketers find themselves using AI software at an increasing rate, and due to constant
development, there is a shift in the responsibilities of marketing managers (Wedel &
Kannan, 2016; Whitler et al., 2017). In many situations, professionals are not required

3
to recognize or understand AI, but rather learn to apply its capabilities (Gentsch, 2018;
Pathan, 2018;). However, marketers face challenges in deploying cost-effective
campaigns (Pathan, 2018). Misinterpretations of the usage and advantages provided by
AI tools can cause financial damage through ineffective marketing (Pathan, 2018;
Tadely, 2014;). For example, Tadelis et al. (2014) showed that misperceptions of SEM
capabilities are financially harmful.

Even though AI tools are enhanced with user-friendly interfaces that allow for anyone
to use them, Gentsch (2018) notes a problem in limited imagination towards AI usage.
The user misplaced confidence in AI capability to utilize information and autonomously
bring results. Such misinterpretations arise from the inability to distinguish between the
two following factors. “First, AI software is not capable of carrying such actions.
Second, the software is augmented AI and requires human input to classify such
parameters and specifications (Athey. et al., 2020)”. However, such boundaries and
expectations between automation and augmentation are not clear. In similar research
by Yalif (2019) the findings also show that marketers misplace confidence in AI
capabilities to solve problems or replace a process. An investigation by Wedel &
Kannan (2016), concludes that marketing managers are increasingly required to
accumulate knowledge in both marketing analysis as well as the technical
implementations of AI. This means that marketers that understand AI infrastructures
and the applicable practices of AI in the utilized marketing tool, are more likely to carry
out more successful campaigns. This is because they familiarize themselves with AI’s
technical limitations. The research identifies that in some situations, marketers’
expectations do not meet AI capabilities. And based on an increasing requirement to
understand both disciplines of marketing and AI, the research examines them as
interrelated knowledge in carrying successful marketing campaigns.

1.3 Purpose and research questions

Based on the research background and the problem statement, AI must develop within
marketing through automation and augmentation. However, such distinctions are not
always clear, and the limitations for automation capacity may be ignored in marketing
campaigns. Furthermore, according to Gentsch’s (2018), limited knowledge may result
in limited imagination towards the utility of AI. This thesis seeks to investigate the
current boundaries in AI automation for marketing platforms, and further distinguish

4
knowledge requirements on AI for the most effective implementation of marketing
strategies. To that end, the research is subject to the following questions:

• What expectations for automation do marketers hold from AI based marketing

platforms?
• What knowledge in AI should be required by marketers?

1.4 Scope and limitations

This thesis explores marketers’ expectations from AI automation in search engine
marketing and in advance the utility of AI knowledge in search engine marketing. While
the research explores the currently existing capacity for automation, its scope does not
report any technical implementations that AI has on search engines. For example, this
thesis will not explain how machine learning works in search engines. Finally,
regarding the development of this research, the following limitations are worth
mentioning:

• First, the interviews were conducted online. This is something that underthrows
human interaction.
• Second, this was the first time that the author conducted an interview, and
experience had to be acquired during the process.
• Third, this thesis interviewed three participants from each target group. This is
a relatively small number of participants, and in future research, it is
recommended to sample data from a larger audience.
• Fourth, even though all participating marketers are Google certified, they do not
share a similar background and experience. In future conducts, group
participants should have the same or similar specialization within the field, for
higher accuracy.

5
1.5 Disposition
So far, this chapter collaborates on the relevance of AI to marketing, and further
distinguishes a knowledge gap in marketers’ knowledge on AI. The thesis claims that
marketers overestimate the current state of automation and that their lack of knowledge
in AI is a limiting factor to their success. In the following chapters, the thesis will be
clarifying the methodological approach to data collection and research analysis.
Further, the theoretical framework presents the applicable capabilities of AI in
marketing. In advance, the thesis presents a template analysis of the interview questions
and data collection. Finally, the data collection is analyzed and discussed with
consideration to the research purpose and theoretical framework to conclude.

6
2 Method and implementation
The following chapter describes the qualitative nature of this thesis and presents the
methodological design. Also, the research elaborates on data collection, quality,
analysis technique, limitation, and ethical consideration. The data collection relies on
recoded structured interviews, and in advance, the conducts use a word analysis.

2.1 Approach
According to Sanders et al. (2016), qualitative research can provide insights into a
problem and help explore the topic in depth. Considering that the research questions
explore a phenomenon directly linked to marketers’ experience, and its linkage to AI
knowledge, the methods chosen for this thesis are qualitative.

In advance, this study follows an inductive process to research investigations, and it is

conceptualized as a path to discovery (Patel & Davison, 2011). In further analysis, the
process of discovery is based on data collection and empiricism (Patel & Davison,
2011). Based on the way that this thesis has been developed and established, the process
described above is highly reflective of the author’s actions. For example, research
investigation was originally developed with a broader approach to AI in marketing, and
in continuation, the research purpose has been narrowed down to identify marketers’
expectations from AI and knowledge requirements. According to Saunders and
Thornhill (2016), the qualitative approach is valuable in seeking insights into tacit
knowledge gathered from real-life situations and experiences. During the process of
data investigation, this thesis has been constantly changing based on empiricism. Thus,
the research questions and perspectives have been changed while adjusting to these new
findings (Creswell & Creswell, 2018).

2.2 Data collection

As mentioned in the research approach, this thesis develops inductively. In advance,
the research topic has been viewed from an exploratory standpoint. Exploratory
research is generally considered to be inductive and qualitative (Stebbins, 2001). This
was necessary due to the limited knowledge of AI and its implication in SEM. When
the author's background does not implicate relevant knowledge to the field under
investigation, it is required by the author to deeply investigate the topic (Breyman &
Bell, 2005). This will build understanding before moving further with research

7
conducts, and it is of utmost importance since it affects a researchers’ interpretation and
the direction that the research takes (Breyman & Bell, 2005; Gadamer, referred in Gikje
et al., 2007).

The research investigates AI applicability in marketing by interviewing two different

types of professionals in a qualitative manner. The first group of professionals
specializes in AI, while the second group of professionals specializes in search engine
marketing. The thesis takes this approach due to the nature of the topic that investigates
the knowledge utility of one discipline incorporated into the other. To that end, the
research interviews three professors in AI to find out what type of knowledge can
improve a marketers’ capabilities in SEM. The participating group of professors is
considered for their knowledge of AI, the existing automation, and their ability to
provide insights on AI limitations that marketers should realize. This will facilitate
understanding and help in distinguishing the existing automation characteristics. These
professors have been chosen based on their professional experience and knowledge in
the field of AI. While on the other hand marketers’ group is interviewed to find out in
what way they utilize AI and their opinion on the knowledge requirements for the best
practice of SEM. All marketers in the interview are chosen based on their SEM
certifications and market experience. According to Google and based on the market
consensus, this certification implies that they should be able to utilize Google platforms
effectively without any professional knowledge in AI.

The data is collected through structured interviews. According to Sanders et al. (2016)
the investigator can use structured interview to maintain control and guide and the focus
of the participants towards a predetermined direction. According to Bogner et al.
(2018), face-to-face interviews provide rich context and are always the best alternative
in interview methods. However, in restricted situations such as geographical location
and health disparity, the best alternative interview methods are videoconference
software (Bogner et al., 2018; Bryman & Bell, 2011). In this thesis paper, the interviews
took place on Zoom and were recorded due to health disparity, and later transcribed.
The interview questions are found in the appendix, transcribed to word format, and
further analyzed in discussion (chapter 5). One of the advantages of interviewing a
person is the capacity to retrieve rich information, including emotional state or beliefs.
The transcripts also contain brief reporting of such emotions according to Bogner et al.

8
(2018). Finally, six people have been interviewed in total. The research sample reports
data on three AI professors and three marketing specialists.

The participants have been contacted a month in advance, and the interviews took place
at a time of their convenience. According to Sanders et al. (2016), research conduct
should consider ethical aspects. Thus, participants have been reassured of
confidentiality and shown appropriate appreciation towards their contribution. All
interviews are conducted individually online and recorded on the Zoom platform.
According to Patel & Davison (2011), face-to-face interaction and recordings are the
best approach for qualitative interviews. This approach can also allow for changes and
adjustments to the interview questions. The researcher can change his approach based
on the empiricism gained after each interview. For example, the researcher’s
communication with interview participants, or the research questions that were not
appropriately formulated, can be modified, or removed.

2.3 Data analysis

This thesis follows the process of thematic framework analysis. The main reasons for
using thematic framework analysis are that it is ideal for novice researchers that are
new to qualitative analysis, and it allows for flexibility when targeting large data sets
(Virginia Braun & Victoria Clarke, 2006). According to Srivastava, A. & Thomson, S.
B. (2009), framework analysis should follow five steps:

1. Familiarization with the transcript of the data collection. In this step, the
researcher should understand, make observations, and start detecting repetitive
patterns.
2. Identify a thematic framework from the data set. These themes may emerge
from previously mentioned issues, however, at this stage, the research can allow
the data to dictate the themes.
3. Indexing refers to the process of identifying portions of the data set that
correspond to the themes.
4. Charting the pieces of data into contextual values of categories and sub-
categories.
5. Mapping and interoperations are the final stage of analysis that builds on
conceptualizing the information that will guide the research.

9
In advance, a method of presenting framework analysis is template analysis. The thesis
uses template analysis to codify interview results. Template analysis is a form of
framework analysis that emphasizes the use of hierarchical coding and balances high
levels of structural analysis that is adapted to the needs of the research (Joanna Brooks
et al., 2015). Also, according to Sanders et al. (2016), template analysis is formulated
based on data structures, available before codifying all interview results. This
implicates the thematic categorization. For example, the research questions can be
considered a structural frame for categorization. The research questions guide the
interview and have been formulated before codifying the interview results. Thus, the
thematical categorization of this template is adapted by the research questions. The
themes are chosen based on the key values of each question. In advance, the
subcategories are a result of repetitive patterns in data collection that emphasize specific
outcomes. The data collection for each question is analyzed based on its relevance to
the theme. The investigator’s job is to highlight and present the most important key
points that are consistent between participants. Moreover, after highlighting the key
contributions to each theme, the categories and subcategories are generated based on
the hierarchical value of importance. In the final step, the investigator’s job is to
iteratively examine the structural construct of this template and improve on it until it is
accurate and corresponds with the research results.

Through this analysis approach, there can be classification and hierarchical order to
things (Sanders et al., 2016). This facilitates credibility towards qualitative analysis.
Also, concerning data analysis, the qualitative nature of this research implies an
investigation that analyzes participants’ data in repetitive ways to establish correlating
patterns (Creswell & Creswell, 2018). This process balances a high degree of structure
in textual analysis and provides the flexibility to adapt the thematology to the needs of
a particular study Jonna Brooks et al. (2014). The templates have been finalized in “4.1”
by following the five steps described above.

Finally, this thesis utilizes one of the knowledge management principles to distinguish
the most credible pieces of information from the data set. The main distinction that has
been made from knowledge management is between personalization and codification.
According to Hislop (2018), personalization is a strategic approach to sharing
knowledge between peers through communicating their experience and actively
transmitting tacit knowledge. While codification defines knowledge that is easily

10
reusable and can be documented (Hislop, 2018). The main contribution of that is to
indicates that aside from the contributor knowledge that is codifiable, the importance
of empiricism gained during the interviews is emphasized by the author’s critical
assessment.

2.4 Validity and reliability

This thesis considers a set of criteria presented by Creswell & Poth (2013). First, the
reach is credible when the results are accurate on the participants meaning. Second, the
authenticity of the research is guaranteed when there is consideration to all participants
and the arguments that have been raised. Third, there is a critical appraisal of all aspects
of research. Finally, the research integrity is found to be based on the investigator/s self-
criticism. These principles have been addressed through the following actions.

• The investigator has used the same terminology that the participants used in
their statements without altering the meaning.
• This research analysis has considered all participants in the process of data
codification.
• This research maintains a clear path of argumentation by utilizing relevant
information that is constantly presented throughout the research.
• The author of this thesis has analyzed the validity of this paper through iterative
proofreading and shown consideration to the criticism of other scholars.

2.5 Ethical Considerations

This thesis has shown high consideration for its ethical obligation towards the
participants. The thesis follows the principles of ethical conduct presented by Sanders
et al. (2016). According to sanders et al. (2016), ethical concerns emerge in the process
of designing and planning the research and are different for each research. Also, while
there cannot be a specific “recipe” for ethical conduct, the research conducts should be
aspired to ethical principles and standards. Some of these standards include being
objective, truthful, promoting accuracy, being respectful, and carefully considering his
actions as to cause no consequential harm to the participants. Finally, the investigator
should ensure confidentiality by respecting the privacy of the participators, informing
them of the voluntary nature of participation, and providing informed consent. Finally,
the ethical considerations for university research and thesis conducts should consider
the ethics of the university itself. According to Sanders et al. (2016), the research should

11
be guided by the university’s code of ethics. In this case, the research methods have
been discussed with the supervisor. To ensure that these principles were applied, this
thesis took the following steps:

• The participants were contacted a month in advance in a very discrete and

polite email.
• The participants could choose the date of their convenience, thus, making
them comfortable to participate.
• Furthermore, the participants have been given a consent contract that
guarantees their privacy and gives them freedom on being mentioned or
excluded from the research. The agreement has been signed by the investigator
as well as the participants.
• During the interviews, the participants were welcomed and in advance, the
investigator kept his professionalism and showed objectivity by keeping the
discussion relevant to the research questions.
• The participants were informed of their right to withdraw at any given
moment.
• The investigator did not alter the results in any way and kept the data findings
accurate and consistent with the opinions of the participants.
• The investigator showed gratitude for their participation and ensured them that
he will take the necessary actions to protect their data.

12
3 Theoretical framework
This chapter presents relevant theory that has been reviewed to answer the research
questions. Considering that the research is focused on the applicable implementations
of AI in Google’s SEM, the theoretical framework outlines a multiplication of
contributor theories. First, this chapter outlines the relevance of AI to marketing, and in
advance AI in search engine marketing. Second, the chapter distinguishes the currently
accepted model of AI implementation and decision-making. Third, the chapter explains
what algorithms are for AI and how algorithms are implemented in marketing. Fourth,
the chapter describes AI maturity to be able of building an understanding of the current
AI capabilities for marketing. Finally, the research considers the most advanced Google
platform for search engine marketing and explains its automation capabilities. This
process will facilitate an understanding of the current capabilities, limitations, and ways
of utilizing AI for search engine marketing. It will also allow for the understanding of
the current state of Google AI and later comparison with research findings.

3.1 AI in marketing

With an increasing number of scholarly articles and books about AI applicability in

marketing (Russell & Norvig, 2014; Jarrahi, 2018; Gentsch, 2018; Pathan, 2018;), it is
only natural to assume relevance between the future of marketing development and AI.
As it was distinguished in the background statement, there are two scholarly predictions
on the development process of AI. Some believe AI to develop as an automation tool
authorized to act, while others view AI development to be an augmented process in
which the decision-making process is human (Jarrahi, 2018; Epstein, 2015; Gentsch,
2018; Pathan, 2018;).

Considering that marketing aims to generate revenue for a brand (Moorehead, 2020),
the research investigates whether marketers’ expectations from AI are met there. The
success rate is evaluated based on consumer ad exposure and its financial
counterfactuals (Varian, 2015). This study is concerned with AI as a rational agent, and
the ability to do the “right thing” based on what it knows (Russell & Norvig, 2014). As
mentioned in this research purpose it is important to develop an understanding of AI
capabilities within marketing. This will facilitate an understanding of AI implications
in marketing. Marketing is broad and AI applicable to a wide range of products, so in a
data proactive investigation, the purpose is to determine AI influence on SEM.

13
According to Pathan et al. (2018), writing on AI techniques in marketing, they note a
constant problematization of marketing budget towards an effective solution. Similarly,
this research demonstrated such examples by Steven Tadelis et al. (2014), where they
examine search engine marketing in utilizing counterfactual data. Their investigation
examines search engine marketing for its financial competence and finds that there is
confusion in applying search engine marketing effective practices.

The investigation regards AI's relationship to marketing by outsourcing research

material that is applicable practice. According to Russell & Norvig (2014), marketers
strive to understand, predict and manipulate the world. According to Russell and Norvig
(2014), AI research can be developed based on criteria evaluating intelligence through
thinking or with behavior, and if one wants to model humans or work from an ideal
standard. In this situation, AI is examined as a relational agent capable of acting towards
achieving the best possible outcome. Since this research is concerned with AI usage in
marketing to generate financially positive returns, the modeling is based on an ideal
standard and evaluated for its ability to think rather than behave.

3.2 Augmentation and automation

The general distinction for AI platforms is between augmented and automated AI. what
distinguishes one from the other is the extent to which the machine can take over a
particular assignment. For example, an assignment that can be completely carried out
by a machine is considered automated AI. While on the other hand the machines that
cannot fulfill the task correctly without human input are considered augmented (Jarrahi,
2018).

According to Susan C. et al. (2020), on the allocation of decision authority to humans

and AI, the consensus often indicates AI with automation that completely replaces
humans in decision making. However, as it is also shown in the research this is not the
case at all. Their research notes that while AI can make predictions and minimize the
margins for error, on the other hand, it cannot be allowed to make decisions that involve
delicate and subjective decisions. They also raise the question of whether AI
development is decreasing human effort and productivity. Something that aligns with
the problem stamen of this thesis on the inability to distinguish the limitations that AI
has in marketing. And while the research acknowledges that there is a phenomenon that

14
they call “AI as a replacement” for the task, the research also recognizes that the degree
of reliability should be extremely high for “AI as a replacement”. Thus, while most
processes can be partly automated, authority and decision-making remain human
processes. Their research concludes that any important decision that has even a small
chance for a marginal error should be taken by humans, thus, making it an augmented
process. In similar research on AI for decision making by Yanqing Duan et al. (2019),
the conclusion has been more extreme on supporting a general consensus that in all
situations AI should be developed to augment rather than automate.

3.3 AI and decision making

Syam and Sharma (2018) argue that decision-making is a fundamental driver in the
fourth industrial revolution. In the last decade, decision-making in marketing would be
taken by a human, however, in the second decade of the 21st century, AI has become
the evaluator and started taking decisions on business practices and market performance
(Syam & Sharma, 2018). This is an important distinction in an investigation that
involves the influence of AI on search engine marketing. And can be also considered
as part of the developmental decision-making within organizations.

To define the value of AI in marketing, this research considers decision-making from

human to machine. According to Kahneman (2003 & 2009), scholarly research on
human decision-making agrees on distinguishing two types of cognitive decision
making. First, fast and effortless decision-making is often influenced by emotions and
habits. This type of decision-making is considered behavioral and intuitive. Second,
decisions are based on a slower, effortful, and rule-based environment. Such decisions
are monitored consciously and deliberately controlled. And while this classification
defines human decision-making, Kahneman & Klein (2009) recognize machine
learning to be more capable in the following environments. First, environments with
valid data require repetitive attention to detail. Second, in environments with low
validity, that humans struggle to validate consistencies. In conclusion, Kahneman &
Klein (2009) find that even though algorithms are limited in their capability to cope
with new information that has not been previously evaluated, it always surpasses
humans within a repetitive analysis.

15
3.4 Algorithmic marketing
Oxford dictionary defines an algorithm as a process or set of rules to be followed in
calculations or other problem-solving operations. On the other hand, the algorithmic
enterprise is defined as business reliance on the competitive advantage generated
through mathematically defined algorithms (Oxford dictionary, 2020). While
algorithmic marketing is defined as the usage of a rule-based system to accordingly
adjust real-time targeting and pricing (Oxford dictionary, 2020). All above statements
share a common vision on the applicable terminology that defines algorithms towards
a process of automating based on rule-based solutions. Thus, in AI development it is
the algorithm that determines the value of information processing. As opposed to the
assumption that big data holds value in AI marketing, one must first consider that
algorithmic manipulation determines the result. Big data is widely available, but the
functional aspect of how this information is traded relates to its algorithmic
manipulation. This immediately implies that the distinction between AI maturity level
has little dependence on big data and more so on the functional set of rules applied on
the software tool (Gemtsch, 2018). Finally, business intelligence relies on multiple
algorithms to execute its business plan (Gemtsch, 2018; Ramesh Sharda et al., 2014).
For example, data marts are the subset of data analyzed for a specific department, or
subject area (Ramesh Sharda et al., 2014). Considering this literature collection and
based on the understanding that I built reading through scholarly articles, algorithms
are a defining set of calculations that are used to evaluate the inserted data. Business
uses different algorithms to set new rules and analytical criteria.

3.5 AI in Search engine marketing

The research has so far described search engine marketing as the process of increasing
the visibility of a website on the results of a search engine. This can be achieved in
multiple ways and through different techniques. The most common distinction in search
engine marketing is between organic traffic and paid traffic. While the most utilized
strategies are search engine optimization, paid submission, search engine advertising,
and shopping ads. First, the purpose of search engine optimization is to boost the
ranking of a website, thus, increasing its leads and conversion. According to Google
academy, this can be achieved through multiple ways, some of which are the relevance
of content, presenting something that is naturally useful for the audience, and applying
correct web infrastructure to the website. Second, paid submission is a method used to

16
increase the referring directories that lead to a website. This can increase citation flow,
traffic, and organic ranking. Third, search engine advertising is a strategy to make
campaigns that are paid for to bring traction. Search engine ads usually appear above
the organic results and can bring faster traction in easier ways. The website links are
labeled as ads and are paid-per-click. Finally, shopping ads are also a paid per click
strategy, but as opposed to search ads, shopping ads are image-based ads that focus on
products, price, and reviews.

While there is a distinction between organic and paid traffic, the process of ranking a
website for organic traffic is not easy and many of the strategies that marketers follow
cost money. For example, paid submission is one of the strategies that marketers follow
to rank a website by increasing its presence and authority. However, the intention of
this research is not to distinguish between SEM strategies, but rather investigate the
way that marketers perceive AI in SEM and how knowledge in AI may affect their
marketing effectiveness. For that reason, the research questions have been formulated
to address the effect that AI has on SEM rather than the way that AI is utilized in a
specific SEM strategy. Thus, it is very important to talk about the way that AI affects
SEM. But since the scope in which AI applies to Google’s SEM is very broad, this can
be explored by focusing on marketeers’ strategies to stay competitive in SEM. First, it
is important to emphasize that Google AI is constantly updated. Therefore, Google
recommends that marketers read all core updates when they are released. This affects
the ranking of all websites and when marketers are keeping up with such updates it can
be a very powerful competitive advantage. Second, according to Google, core updates
are developed with consideration to the client that searches for something rather than
how marketers can benefit from their update. This immediately means that marketers
are required to generate content that has value to the user. Third, according to Google,
AI algorithms are developed to investigate and understand web infrastructures, user
journey, user satisfaction, and content. This allows the search engine to punish cheating
tactics and make the experienced user centered. For example, SMIT algorithm was
released by Google to investigate long text and detect “transition pathways” in the topic.
Meaning that the algorithm can tell if the topic is naturally shifting to provide
information and make a conclusion.

As mentioned above, Google AI is focused on a user-centered approach. One of the

machine learning techniques, predictive learning is frequently used to predict customer

17
insights and personalize content creation to targeted groups (CMO Survey 2019).
Predictive learning is a form of supervised machine learning technique reliant on
historic data collection to make decisions based on client preference (Kelleher et al,
2015, pp. 1-2). This means that the results are also affected by browser history, which
is used to customize the results. However, predictive learning extends to organizational
decisions and not only targeted advertising. For example, it can be used for document
classification, Propensity modeling, measuring market risk, evaluating product pricing,
and measuring risk (Gentsch, 2018; Kelleher et al., 2015;). Propensity is an even
narrower investigation of predictive behavior that determines customer behavior based
on independent and confounding variables affecting behavioral sequence (Kelleher et
al., 2015). For example, propensity supports the customer evaluation of the likelihood
of product purchase. This is achieved by utilizing browser history and in many cases
cross channeling data between a browser and social networks. This is a strategy that
can be used to take advantage of the user-centered approach that SEM is developed for.

18
4 Results

As mentioned in research methods, this thesis has gathered information from six
participants in total. The participants can be separated into two groups after their
profession. Three participants in each group. In advance, the results are presented in
two separate segments. Furthermore, the interview questions have been distinguished
and grouped for their relational contribution to the research purpose in the following
subsections.

4.1 Collected data

This chapter provides an overview of the data collection after a framework analysis.
The following is a template analysis of the two participatory groups. As mentioned in
data analysis, the template analysis is an extension of framework analysis. The analysis
is a result of the following steps. First, the author should get familiarize with the results
(see Appendix A). Second, the author should identify the thematic framework. In this
case, the themes are based on the interview questions. As it was explained in data
analysis, the investigator should highlight the key points that guide the research. Thus,
the main focus of the question generates the theme (see Appendix B). Third, identifying
and indexing the data values that correspond with these themes. This means that the
participants answers. Fourth, charting the data into categories and subcategories to
indicate hierarchical value and importance. In advance, repetitively accessing the
template for the most accurate classification. Finally, start mapping and making
interoperations to guide the research. This is achieved by making interpretations and
conceptualizing the information in a more descriptive way that guides the results. This
is presented in the next subsection as data analysis and contains a more collaborative
description of the data results with interpretations.

4.1.1 Template analysis of the interview with marketers

1. Marketers’ perspectives on AI and SEM.

1.1. Background in search engine marketing.
1.1.1. Marketing specialization.
1.1.1.1. Degree in marketing.
1.1.2. Experience with SEM.
1.1.2.1. SEM certification.
1.1.2.2. Work experience.

19
 1.2. The general sense of AI applicability in marketing.
1.2.1. Good perceptual understanding of AI.
1.2.1.1. Understanding AI applicability to SEM.
1.3. The effect of AI on his/her responsibilities as marketeers.
1.3.1. Provides people with he ability to focus on strategic roles.
1.4. Knowledge for the best utilization of SEM capabilities.
1.4.1. Primary.
1.4.1.1. Constantly learning possible implications.
1.4.1.2. Knowledge on analysis and strategizing.
1.4.2. Secondary.
1.4.2.1. Knowledge relevant to AI.
2. Marketers’ knowledge.
2.1. Capabilities of search engines.
2.1.1. AI automations in SEM.
2.1.1.1. High levels of automation.
2.1.1.1.1. SEM automations are widely unknown.
2.1.2. AI augmentation for SEM.
2.1.2.1. Applicable processes.
2.1.2.1.1. Data segmentation and labeling.
2.1.2.1.2. Data management.
2.1.2.1.2.1. Capability to minimize bias.
2.2. Future Improvements in SEM.
2.2.1. Automate processes that require repetition.
2.2.2. Create interactive shopping ads.
2.2.3. Adapt to cooky-less systems.
3. AI augmentation & Automation.
3.1. AI relationship to firms and marketers.
3.1.1. Pros.
3.1.1.1. AI is highly beneficial to marketers.
3.1.2. Cons.
3.1.2.1. Companies are afraid to lose control in decision making.
3.1.2.2. Consumers are not appropriately informed.
3.2. Marketers’ role in SEM.

20
 3.2.1. Strategic role.
3.2.1.1. Creative solutions in SEM.
3.2.1.1.1. AI is limited in creative solutions.

4.1.2 Template analysis of the interview with AI professors

1. AI professors’ perspective on AI in marketing.

1.1. Profession in AI.
1.2. Background knowledge in marketing.
1.2.1. High level of understanding.
1.3. AI augmentation and automation.
1.3.1. Automation.
1.3.1.1. Far from marketing automation.
1.3.1.2. Constant automation will keep developing.
1.3.2. Augmentation.
1.3.2.1. The status of marketing today.
1.4. Barriers in AI development.
1.4.1. AI cannot understand emotional intelligence.
1.4.2. AI does not have the capacity for creativity.
1.4.3. AI works well within a framework.
1.4.3.1. Does not cope well with unknown variables.
1.5. Important knowledge for future marketers.
1.5.1. Knowing the boundaries of AI systems to minimize bias.
1.5.1.1. System possibilities.
1.5.1.2. System limitations.
1.6. Importance of AI knowledge in marketing.
1.6.1. Marketers should not be required to understand AI.
2. Machine learning and the status of AI in marketing.
2.1. Decision making by AI in SEM.
2.1.1. AI is unable to make strategic changes.
2.2. AI impact on marketing companies and consumers.
2.2.1. Companies.
2.2.1.1. AI as a competitive advantage.
2.2.1.1.1. AI will keep developing on existing products.
2.2.2. Consumers.

21
 2.2.2.1. Consumers should be more aware of data collection.
2.2.2.1.1. Neural networks can manipulate populations.
2.3. Marketing benefits from AI and SEM.
2.3.1. Exposure to consumers.
2.3.2. Accurate predictions by cross channeling data.
2.4. AI applicability to SEM.
2.4.1. AI personalizes the experience based on historic data.
2.4.2. AI makes customer predictions based on historic data.
2.5. Improvements in SEM knowledge.
2.5.1. Understanding system behavior.
2.5.1.1. Input & output.
2.5.2. Capacity to realize business opportunities.
2.5.2.1. Know the capabilities offered in SEM.
2.5.2.1.1. Minimize bias.
2.6. Suggested improvements in SEM.
2.6.1. Marketers should be able to collect and utilize a broader set of
information.
2.6.1.1. Make it possible to extract more data.
2.6.2. Search engines can assure marketers.
2.6.2.1. Motivate automated suggestions and predictions.

4.2 Data analysis

This chapter briefly summarizes the interview results. As mentioned in research
methods, the qualitative nature of this research implies an investigation that analyses
participants’ data in repetitive ways to establish correlating patterns (Creswell &
Creswell, 2018). The headings used are based on such similarities in the participants'
answers. In advance, the results presented close the gap of biased information by using
the same terminology used by the participants. Finally, the thesis has followed the
necessary steps to identify the appropriate data values for this research, and in this step
the thesis will start mapping and inter

4.2.1 Perception of AI in marketing

The first part of the interview regards the participants' profession and background. Also,
the participants were asked to describe AI within marketing and provide opinions and

22
perspectives on the matter. The interview questions were formulated in a manner that
approaches AI placement within marketing from an automated or augmented
perspective.

AI professors: The data collection shows a small fluidity concerning their specialty
determinants within AI and their knowledge in marketing. However, all participants
have similar answers about their perceptual understanding of AI's relationship to
marketing. They all believe the current state of AI to be far from completely automating
marketing strategies. The entire sample seems to agree on AI barriers in future
development for marketing to be the fundamental definition of intelligence. The
findings show that in general there is skepticism on how AI can understand human
creativity, emotions and how such things impact people’s behavioral sequence. Also,
about marketers’ knowledge in AI, this group believes that the systems are developed
in such ways that marketers are not required to know the underlying infrastructure. And
while this may help them, and facilitate a better understanding, they agree that
marketers need to know the true boundaries of the utilized tool. This implies that they
are expected to know the tool for its strength as well as for its weaknesses. This way
they can utilize such systems without any biased results. AI specialists hold low
expectations in general from marketers' understanding of technical and technological
infrastructures. However, they believe marketers to have a good understanding of
search engine marketing. While they believe that increasing a marketer’s knowledge in
AI infrastructure may be positive, the most important is for a marketer to increase
knowledge in the behavior of the utilized system, and the effect of input and output.

Marketers: Marketers showed a generally good understanding of AI and its placement

in marketing. Without knowing the exact distinction between automated and augmented
AI platforms, marketers were capable of picturing one of the two. However, the context
in which AI was described by marketers does not show distinction and clear boundaries
of automation and augmentation. Marketers described AI whether through automation
or augmentation. For example, one of the participants described AI as a machine that
takes decisions and actions by itself. While another participant described AI as the
machine that is constantly developing towards human intelligence.

On the other hand, marketers have utterly controversial ideas regarding developing
knowledge in AI infostructure. Some participants said that knowledge in technical

23
infrastructure and web infostructure can increase problem-solving capacity, while the
most experienced of the participants also claims that mathematical perception can be
utilized in analytics and advanced applied to search engine marketing. However, even
though they believe that such skills can be positively utilized, they support the idea that
they are not a must in successful search engine marketing.

4.2.2 Marketers’ role in SEM

The interview asks questions on the utility of AI in marketing and the responsibilities
that marketers should have. It also regards the importance of creativity and social
understanding within SEM. The participatory groups also answered voluntarily how
creativity may be applied in AI development and in advance to marketing. Both
participating groups find AI incapable of decision making and creative problem solving.

AI professors: The response received by AI professionals indicates that the

development of AI is a long difficult process that will take time. One of the participants
quotes “we cannot develop systems that go into uncharted territories and solve open-
ended problems”. AI professors agree on the AI barrier in future development for
marketing to be the fundamental definition of intelligence. The findings show that in
general there is skepticism on how AI can understand human creativity, emotions and
how such things impact people’s behavioral sequence.

Marketers: Their response towards creativity and social understanding in SEM, shows
that creativity is a big part of developing strategic thinking towards consumers and
understanding the most suitable target audience. On average marketers believe that
technical understanding and creative thinking coexist and are equally useful in
successful SEM. They go further to describe the development of AI as dependent on
creative and abstract thinking to further utilize human intelligence. Finally, marketers
provide an insightful response towards GA360 automation development. In an
interesting turn of events, marketing participants stated “Highly automated tools in
premium google platforms are taking over the process. However, marketing managers
are afraid to utilize such tools out of fear of losing control”.

4.2.3 Google AI in SEM

The two participatory groups were asked to express their opinion on AI and the many
utilities that it provides to search engine marketing. Considering their experience and

24
knowledge of applicable practices, the participants described AI capabilities in SEM
and the way it is currently employed by Google or utilized by the company that hires
them.

AI professors: Two out of three of the participants started to understand marketing

very well however, only one participant believed himself to deeply understand Google
AI in SEM. However, all participants stated that almost all machine learning works in
similar ways and follows similar rules. The following statements have been repeated by
all participants, in this group regarding Google AI in SEM, thus, making a pattern.

According to this group of participants, Google AI is highly applicable to historical data

through machine learning. Search engines provide unique and personalized content to
each individual client based on browser history. They make predictions based on a
highly heuristically built framework from historical data. Also, Google AI utilizes data
collection from Facebook, Instagram, and other social platforms in combination with
browser history. However, in relation to the current state of capabilities that AI
possesses, AI specialists seem to believe machine learning with a focus on automating
a business marketing strategy may exist on an experimental level or as a tailored
solution to organizations. There are no systems available that can comprehend and
develop complex system strategies. The systems cannot teach themselves to become
smarter and solve problems in new ways. No open-ended questions that AI can solve
for itself. In its current state, artificial intelligence is incapable of generating new
strategies without human input.

Marketers: On the other hand, marketers identify a huge leap of technological

development that is currently ignored. In the interview with marketing specialists, one
of the participants having considerable experience and knowledge with SEM said
“Google provides highly autonomous tools in its high-end platforms. However, even
big marketing conglomerates are afraid to utilize such tools out of fear of losing control
to the system”. The participant also identified other AI tools that are in place to enhance
automation and accuracy in relation to search engine marketing. For example, while
360 is using AI, it does not provide AI insights. In the company he is employed they
use Kenshoo to get insight on cross-Channel traffic and how target groups spend their
time on each platform. This specific solution is made to compare and combine SEM
with social platforms or any other popular advertising space. However, the participants

25
had a distinct opinion on AI automation and its effect on marketing. While some see
that AI is gradually automating most positions required by marketing companies in
SEM and gradually replacing humans, others seem to believe that the human aspect is
irreplicable and Ai should be more focused on bigger problems, and more sleight
forward problem-solving.

4.2.4 SEM and future development

Finally, the interview asked both participatory groups to provide an opinion on the
future of AI in marketing. The results show similarities from both participatory groups
regarding the future of AI development for marketers as well as consumers. Both groups
believe that consumers are being manipulated to some extent and since they agree to
share data, it is not a company’s mistake. However, both groups also suggest that
consumers should be more informed on what they get themselves into. While in relation
to a few unique suggestions by each participatory group, the following categories
decorative some examples given by each group.

AI professors: In further developing SEM AI specialists believe that search engines

need to make it possible for marketers to extract more knowledge. For example, search
engines can provide more than a prediction, they can provide a reason on why the target
audience is a good match for this campaign. Thus, explaining the system's decision-
making and its confidence in its prediction. This is an ongoing Google research and can
be also associated with the demand shown by marketers’ participation and response on
Kenshoo.

Marketers: In relation to SEM, marketing participants suggest that AI automates

keyword combinations towards target audiences. Also, in relation to a recent evaluation
of system infrastructure, someone suggested that AI should be looked upon as the
solution for cooky-less systems. Considering that companies such as Facebook and
others will soon remove cookies, this is an area of ongoing development. However,
there were also more creative suggestions such as developing dynamic shopping ads on
SEM.

26
5 Discussion
This chapter compares literature collection and interview analysis. In the previous
chapter, the analysis has been performed according to qualitative methods. This
research will distinguish between marketers’ expectations from AI capabilities and the
expected knowledge for appropriately utilizing AI in SEM.

5.1 Result discussion

The interview results have supported a deeper understanding of the contribution that AI
has in marketing and the knowledge requirements for the best utilization of AI tools.
The interview results show high relevance with the theoretical framework in the
following areas.

First, the definition of AI as it was quoted during the Dartmouth conference, “AI can
develop on every aspect of learning and intelligence that can be so precisely described
that machine can be made to simulate it” (Epstein, 2015). Both participatory groups
showed similar understanding and make the same indication.

Second, they all show an understanding of AI progression within marketing. The results
show that marketers completely understand this constant development in their role and
position. They display an understanding of keeping track of emerging technologies and
tools that can enhance their work. This is completely aligned with the theoretical
framework. As mentioned in the theoretical framework, there is a constant shift in
marketers’ responsibilities due to technological development and an increase in
automation (Wedel & Kannan, 2016; Whitler et al, 2017, Jarrahi, 2018).

Third, the professors in AI share the same understanding when it comes to the
development of AI in marketing. They all indicated that AI development is a long-
lasting process with developers constantly adding parameters and features to make it
act more humanly. For example, developers that keep adding human features to the
system, rather than the system being developed in a way that it can think humanly for
itself. This is in complete alignment with the theoretical framework on automation and
augmentation. As mentioned in the theoretical framework, according to Susan C. et al.

27
(2020), AI cannot be allowed to make decisions that involve delicate and subjective
decisions.

However, the research results also show differences regarding AI knowledge

requirements. According to Wedel & Kannan (2016), concludes that marketing
managers are increasingly required to accumulate knowledge in both marketing
analysis as well as technical and technological implementation. While marketers agree
on such benefits, they believe it to be secondary knowledge and not necessarily required
in SEM, and AI professors believe that it does not contribute at all to successfully
utilizing AI in SEM. And while knowledge of AI is not perceived as an important
facilitator for the creative utility of SEM. The participants emphasize the importance of
knowing the full extent of capabilities as well as the limitations of AI in SEM and being
capable of creatively utilizing SEM

Also, the participants believe that AI is a tool that can replicate rational intelligence,
but not human thinking in any way. This does not align with the theory suggested by
Russell & Norvig (2014), where AI intelligence can be separated into thinking humanly,
thinking rationally, acting humanly & acting rationally. Interview analysis shows that
the participants do not believe AI to be capable of thinking and strategizing humanly.
This is where they find themselves confident that AI cannot replace them. Based on the
above observations the research questions can be addressed as follows:

1. What expectations for automation do marketers hold from AI-based

marketing platforms?

Research findings show that marketers should stay constantly updated with emerging
trends and new tools that can further enhance their work. In relation to marketers’
expectations from AI, the results show that marketers expect a constantly increasing
automation to gradually take over SEM processes. They identify a variety of SEM tools
that can automate their current responsibilities but are feared because they lack
transparency in the data processing. Surprisingly, this has been also mentioned as a
possible improvement by the group of AI professors. They suggest that AI tools that
are highly automated should provide marketers with a brief explanation for the actions
taken. This should explain the variables considered to take such autonomous decisions.

28
In relation to SEM, marketers showed worry and in some situations incapability to
understand the type of control that the company maintains over its campaigns when AI
automation is utilized to its utmost. They specifically identify GA360 and say that it is
increasing its number of automation tools at an accelerating rate, and the company
hesitates to utilize its capabilities out of fear of losing control. Thus, marketers
recognize high levels of automation and possibilities in the area however, they are also
certain of AI incapacity to understand human intelligence.

2. What knowledge in AI should be required by marketers?

The results show that technical understanding in AI infostructure and algorithmic

understanding is not necessarily required. The participants consider technical
knowledge to be secondary. They believe that utilizing a tool does not necessarily
implicate the tools’ construct. Both participatory groups seem to believe that AI tools
are developed to be used without any AI knowledge or understanding. While the thesis
introduction presents that knowledge in AI infostructure can increase the chance of
rational decision-making (Wedel & Kannan, 2016), this was not proven. However,
according to both participatory groups, such knowledge could be partially useful.
Finally, research results show consistency between marketers and AI professors
towards completely understanding the boundaries of the utilized tool.

Accordingly, campaign strategy is built on experience and understanding of society and

consumers. According to marketing participants, this is built based on experience in the
field as well as staying informed on new tools, AI capabilities, and strategies.

5.2 Method discussion

This thesis has chosen qualitative research methods and conducted structured
interviews. The questions that have been formulated are open-end questions and do not
necessarily require a short and specific answer. This provides the opportunity for richer
information sharing however, it makes the distinction and codification of knowledge
harder and possibly more vulnerable for biased interpretations. For that reason,
measures have been taken to minimize such errors (e.g., template analysis). While in
terms of methodological rigor, this thesis follows through with the described
methodologies. The research questions have been answered by considering repetitive
findings from the theoretical framework, interview analysis, and comparison between

29
the two. In future research, it is recommended that the data sample will be collected
from a larger number of participants. Thus, increasing the validity and accuracy of
prediction. Concerning the validity of this research, the codification of information has
been kept objective. The word analysis from the video recordings has been documented
using the same terminology and maintaining the same value meaning.

30
6 Conclusions and further research

The previous chapter discusses the result of this research in comparison to previous
theories implicated to the topic. While in this chapter the research presents the
conclusion and suggestions for future research.

6.1 Conclusions

The purpose of this thesis has been to investigate marketers’ knowledge of AI and
advance the understanding of the most suitable areas of knowledge learning to best
utilize SEM practices. This was originally hypothesized to be knowledgeable in the
technical applicability of AI, but it has been disproven.

Many of the findings are in line with the theoretical framework, such as the advantages
of a marketers’ analytical mindset (Wedel & Kannan, 2016). Also, the premise of a
long-lasting automation development towards super-intelligent enterprises by Gentsch
(2018), has been confirmed. As the interview with AI professors shows that we are far
from calling Google AI automated or autonomous. This development of AI in
marketing is a long-lasting development that will possibly take more than 20 years.

Marketers have high expectations from AI to keep developing automation parameters,

and they expect a constantly increasing automation to gradually take over most of the
SEM processes. However, marketers also showed worry and in some situations
incapability to understand the type of control that the company maintains over its
campaigns when AI automation is utilized to its utmost. However, marketers are also
certain of AI's incapacity to understand human intelligence. Something that is also
agreed upon by AI specialists. While they seem to believe that automation is
increasingly making their life easier, and it will keep solving problems, they also worry
about the loss of control in their campaigns.

The expectation that AI knowledge may greatly impact marketers’ performance is not
met. The research is conducted under the assumption that building an understanding of
the underlying infrastructure of AI can greatly impact marketers’ ability to utilize AI
tools to their maximum potential. However, this has been disapproved by both

31
participatory groups. Instead, the required knowledge in AI for the best SEM practices
is perceived as knowing how the system will behave by understanding the capabilities
of the utilized tool. For example, knowing how to minimize bias and the methods of
utilizing AI tools rather than understanding the underlying structure of such tools.

6.1.1 Practical implications

This thesis contains the empiricism gained by a systematic review of the literature and
data sampling from interviews with professionals in AI and SEM. While the applicable
benefit of this thesis is to enhance knowledge and empirical understanding of the
disciplines. Based on the interview discussions and comparison to the theoretical
framework, future marketers can benefit from the empiricism presented. This thesis has
shown findings regarding the current automation capabilities of Google and how they
are being utilized in the field. Also, the interview contained questions on current
problems and future potential growth in AI for SEM. The following implications are
specific to this research and can further facilitate future growth for the industry.

1) The findings have shown that marketing conglomerates are afraid to utilize the full
automation capabilities provided by Google AI. In correlation to that, AI specialists
have suggested that Google's platform needs to communicate clearer explanations on
the actions taken automatically by AI.

2) Marketers have provided insights on problems in search engine marketing that

emerge in their work and are relatively simple to solve. Some of the examples presented
in data collection are interactive shopping ads, cookie-less systems, and increasing
automation in a specific part of the process that can be automated.

3) Marketers have provided a lot of empirical information on how to utilize data

analysis for SEM. A very good example of the empiricism that was presented is cross-
channeling information between social media marketing and search engine marketing.
This can be achieved by utilizing the AI tool that was suggested named "Kenshoo".

4) In consideration of the research purpose, marketers’ expectation opinions on the

current state of Google AI have been documented, analyzed, and discussed. Moreover,
this thesis has investigated the hypothesis that knowledge in AI contributes to
successful marketing. The results have disapproved this hypothesis and answered that

32
the most suitable type of knowledge in SEM is not knowledge on AI’s technical
applicability, but rather a complete understanding of the utilized.

5) The empiricism presented can highly benefit emerging marketers to develop their
perceptual understanding. This thesis has presented empiricism from revised literature
and analysis on professional input. This means that marketers can broaden their
perceptual scope of the implicated AI contributions to SEM.

6.1.2 Scientific implication

The results of this research have demonstrated the opportunities that Google AI
provides to SEM. In advance, the results clearly indicate that there is no requirement
for knowledge in AI’s technical infrastructures as opposed to the research by Wedel &
Kannan (2016). But rather than the possible improvements in marketers’ knowledge
should be an analytical mindset and strategic thinking. Also, regarding the possible
improvements in Google AI, this research findings have shown that shopping ads
should be more interactive, as well as other contributor ideas such as the increase of
available data streams for the marketers.

Considering that there is a limited number of publications on the relation of AI

knowledge to a marketer's capabilities in search engine marketing, the results of this
research have explored the possibility of combining AI knowledge with SEM. This
thesis forms a hypothesis, explores the hypothesis, and further clarifies the emerging
concepts (Stebbins, 2001).

Also, based on Sanders (2016) using template analysis has facilitated a better
understanding and hierarchical structure to the contextual values presented during the
interviews. According to Sanders et al. (2016), using widely accepted templates to
classify information will result in a clearer understanding and data presentation. Finally,
following such methods can further facilitate knowledge transfer from an experienced
marketer (Hislop, 2018). This facilitates growth and understanding of the potential in
AI knowledge to further enhance the experience of future search engine marketers. This
will help in developing a framework of thinking and understanding knowledge work
characteristics in SEM.

33
6.2 Further research
This research investigates limited creativity in utilizing AI tools because of limited
knowledge. However, future research could consider investigating what defines
creativity is in SEM, rather than the creative limitations due to lack of knowledge in
AI.

A different approach could consider comparing the value of creative thinking and
objective analysis for SEM, through quantitative research. For example, while one of
the participants finds creativity important, another participant emphasized
mathematical perception instead.

34
References

Adam Heitzman (2017). How Google Came To Dominate Search And What The
Future Holds. Forbes.

Alex Dogariu, Nicolas Maltry, Mercedes-Benz Consulting (2018) AI in Marketing,

Sales and Service: How Marketers Without a Data Science Degree Can Use AI, Big
Data and Bots. Springer International Publishing AG. Chapter 5 as cited by Gentsch,
P. (2018).

Al-Sakib Khan Pathan (2018) Technological advancements and innovations are often
detrimental for concerned technology companies. International Journal of Computers
and Applications, 40:4, 189-191.

and Elites. In: Flick, U. 2018. The SAGE handbook of qualitative data collection. 1st
ed. Los Angeles: SAGE, 652-668.

Benjamin Dethine, Manon Enjolras, Davy Monticolo (2020) Digitalization and

SMEs’ Export Management: Impacts on Resources and Capabilities. Technology
innovation management review.

Bogner, A., Littig, B., & Menz, W. (2018). Generating Qualitative Data with Experts

Bryman, A. (2008). Social research methods. 3rd ed. Oxford: Oxford University
Press.

Bryman, A., & Bell, E. (2005). Företagsekonomiska forskningsmetoder. 1st edition.

Malmö: Liber.

Bryman, A., & Bell, E. (2011). Business research methods. 3rd ed. Oxford: Oxford
Univ. Press.

Bryman, A., & Bell, E. (2017). Företagsekonomiska forskningsmetoder. 3rd ed.

Malmö, Liber.sources and Capabilities. Technology innovation management review.

Bughin, J., Seong, J., Manyika, J., Chui, M. and Joshi, R. (2018). Notes from the AI
frontier: Modeling the impact of AI on the world economy. McKinsey.

35
Creswell, J., & Poth, C. (2013). Qualitative inquiry and research design: Choosing
among five approaches (Fourth ed.). Los Angeles: SAGE Publications.

Epstein, S. (2015). Wanted: Collaborative intelligence. Artificial Intelligence. 36-45.

Fagella, D. (2019, 31 January). Machine Learning Marketing – Expert Consensus of 51

Executives and Startups. Emerj. https://emerj.com/ai-market-research/machine-
learning-marketing.

Foroudi, P.; Gupta, S.; Nazarian, A.; & Duda, M. (2017). Digital technology and
marketing management capability: achieving growth in SMEs. Qualitative Market
Research, Vol. 20 No. 2, pp. 230-246.

Gentsch, P. (2018). AI in Marketing, Sales and Service: How Marketers Without a

Data Science Degree Can Use AI, Big Data and Bots. Springer International
Publishing AG.

Gilje, N., Grimen, H., and Andersson, S. (2007). Samhällsvetenskapernas

förutsättningar. 3rd ed. Göteborg: Daidalos.

Google AI (2021). Give customers more of what they love.

Guy Yalif (2019) AI Mythbusters: Six misconseptions about AI for marketers. Forbes
Technology Council.

Hal R. Varian (2006) The Economics of Internet Search. University of California at

Barkley pp.177-191.

Hal R. Varian (2015) Causal inference in economics and marketing. Edited by

Richard M. Shiffrin, Indiana University, Bloomington. Approved May 25, 2016.

Hanssens, D. & Pauwels, K. (2016). Demonstrating the Value of Marketing. Journal of

Marketing. 173-190.

Jarrahi, M.H. (2018). Artificial intelligence and the future of work: Human-AI
symbiosis in organizational decision making. Business Horizons, 61(4), 577–586.

36
Joanna Brooks, Serena McCluskey, Emma Turley & Nigel King (2015). The utility of
template analysis in qualitative psychology research. Pages 202-222.

John W. Creswell, J. Davids & Creswell, (2018). Research design : qualitative,

quantitative, and mixed methods approaches (5th ed.). Los Angeles: Sage Publications.

Juan Alcácer, John Cantwell and Lucia Piscitello (2016) Internationalization in the
information age: A new era for places, firms, and international business networks?
Journal of International Business Studies Vol. 47, 499–512.

Kahneman, D. (2003). A perspective on judgment and choice - Mapping bounded

rationality. American Psychologist, 58(9), 697–720.

Kahneman, D., and Klein, G. (2009). Conditions for Intuitive Expertise: A Failure to
Disagree. American Psychologist, 64(6).

Kaya Ismail, (2020). 6 Open-source AI frameworks you should know about. Published
at CMS Wire.

Kelleher, J., MacNamee, B., & D'Arcy, A. (2015). Fundamentals of machine learning
for predictive data analytics: Algorithms, worked examples, and case studies.
Cambridge, Mass.: The MIT Press.

Kelvin Lui & Jeff Karmiol (2018). AI Infrastructure Reference Architecture. IBM
Systems.

Kolbjørnsrud, V., Amico, R. & Thomas, J., R. (2016). The promise of artificial
intelligence - Redefining management in the workforce of the future. Accenture
Institute for High Performance.

Kukla, A. (2013). Social constructivism and the philosophy of science. Routledge.

Liz Moorehead (2020) What is the purpose of marketing; An update for 2021. Published
on Impact+ learning community.

37
Marinchak, C.M., Forrest, E., & Hoanca, B. (2018). Artificial intelligence: Redefining
marketing management and the customer experience. International Journal of
Entrepreneurship and Innovation, 8(2), 14–24.

Martin Rayn (2019) Responsible AI and Predictions for 2019. PWC reporting (2019).

Melewar, T.C. & Foroudi, Pantea & Gupta, Suraksha & Kitchen, Philip & Foroudi,
Mohammad. (2017). Integrating identity, strategy and communications for trust,
loyalty and commitment. European Journal of Marketing. 51. 572-604. 10.1108/EJM-
08-2015-0616.

Nicolas Papadopoulos & Oscar Martin (2011) International market selection and
segmentation: perspectives and challenges. International Marketing Review Vol. 28
No. 2, pp. 132-149.

Nonaka, I. and Takeuchi, H. (1995) The Knowledge-Creating Company: How Japanese

Companies Create the Dynamics of Innovation. Oxford University Press, New York.

O’Brien, F. (2019). Artificial Intelligence and the Future of Web Design. Designing
Websites. Viewed 17 January 2020:

Patel, R., & Davidson, B. (2011). Forskningsmetodikens grunder: Att planera,

genomföra och rapportera en undersökning. 4th ed., Lund: Studentlitteratur.

Pradeep, A. K., Appel, A., & Sthanunathan, S. (2018). Ai for marketing and product
innovation : Powerful new tools for predicting trends, connecting with customers, and
closing sales. ProQuest Ebook Central.

Ramesh Sharda, Dursun Delen & Efraim Tuban (2014) Business intelligence and
analytics. Tenth Edition.

Rich, E., Knight, K., & Nair, S. B. (2009). Artifcial Intelligence (3rd ed.). New York:
Tata McGraw-Hill.

Russell, S. J. & Norvig, P. (2016). Artificial intelligence: a modern approach (3rd

ed.). England, Essex: Pearson Education Limited.

38
Russell, S.J., & Norvig, P (2014). Artificial Intelligence: A Modern Approach. 3rd,
Pearson New International ed. Harlow: Pearson Education Limited.

Saunders M., Philip Lewis, Adrian Thornhill (2016). Research Methods for Business
Students. Published by Pearson (2016).

Saunders, M., Lewis, P., & Thornhill, A. (2012). Research methods for business
students. 6th ed. Harlow, New York: Pearson.

Setareh Maghsudi, Andrew Lan, Jie Xu, and Mihaela van der Schaar (2021).
Personalized Education in the AI Era: What to Expect Next?. Published by Cornell
university.

Simpson, J. A., Weiner, E. S. C., & Oxford University Press. (1989). The Oxford
English Dictionary. Oxford: Clarendon Press.

Srivastava, A. & Thomson, S. B (2009). Framework Analysis: A Qualitative

Methodology for Applied Policy Research. JOAAG, Vol. 4. No. 2

Stebbins, Robert. (2001). Exploratory research in the social sciences: what is

exploration?. Exploratory Research in the Social Sciences. 2-18.

Susan C. Athey, Kevin A. Bryan, and Joshua S. Gans (2020) The Allocation of Decision
Authority to Human and Artificial Intelligence. AEA Papers and Proceedings, 110: 80-
84.

Syam, N. & Sharma, A. (2018). Waiting for a sales renaissance in the fourth industrial
revolution: Machine learning and artificial intelligence in sales research and practice.
Industrial Marketing Management.

Thomas Blake, Chris Nosko & Steven Tadelis (2014), Consumer Heterogeneity and
Paid Search Effectiveness: A Large Scale Field Experiment. University of Chicago
and eBay Research Labs.

Vegard Kolbjørnsrud, Richard Amico, and Robert J. Thomas (2016) How artificial
intelligence will refine management. Harvard business review.

39
Virginia Braun & Victoria Clarke (2006) Using thematic analysis in
psychology, Qualitative Research in Psychology, 3:2, 77-

Wedel, M. & Kannan, P. K. (2016). Marketing Analytics for Data-Rich Environments.

Journal of Marketing, 80(6), 97–121.

Whitler, K.A. & Morgan, N. (2017). Why CMOs Never Last. Harvard Business Review

Whitler, K.A., Boyd, D.E. & Morgan, N.A. (2017). The criticality of CMO-CIO
alignment. Business Horizons, 13–324.

Yanqing Duan, John S. Edwards, Yogesh K Dwivedi, (2019) Artificial intelligence for
decision making in the era of Big Data – evolution, challenges and research agenda.
International Journal of Information Management, Volume 48, 2019, Pages 63-71.

Yulia Gavrilova (2020) Artificial intelligence vs machine learning vs Deep learning:

essentials. Published at Serokel blog.

Appendix A
Marketers’ Interview questions

Explain AI augmentation & Automation:

Section 1: Audience introduction, demographics & Perspectives.
1. Tell me about your background and what you do in your work, as well as how
it relates to marketing?
1. Digital marketing. Masters with a specialty in B2B digital marketing.
2. Journalism, International communication, corporate communication
community manager social media analyst, SEM. Main focus on paid
google ads and social media analysis, currently employed by a big
company to manage Volkswagen marketing campaign.
3. Business administration, sales, and marketing internships, and
currently employed by nestle as a marketing manager for Amazon.
2. Do have any experience with working on search engine marketing?

40
 1. Experience in start-ups and high-end companies in SEM & SEO
2. Experience working in four different companies specializing in data
analysis and online marketing. The last two years working in a very big
company as SEO & SEM specialist
3. Very little practical experience, but lots of theoretical knowledge and
certifications.

3. Where or how did you acquire your knowledge of search engine marketing?
1. Google courses and certifications as well as internship experience.
2. Google ads & search certifications, display, YouTube, Amazon,
acquired almost all google certificates
3. Google ads fundamentals, Google search ads, Amazon advertising,
Hub spot email marketing.
4. What is the level of knowledge you have on SEM?
1. 5-6
2. 7-8
3. Very good theoretical knowledge but 1/10 in practice.
5. In your opinion, how well do marketers understand web infostructures and /or
other facilitating AI platforms & tools? (On average...1-10)
1. 6-7
2. Gerard rated his knowledge on web infostructure 6, while knowledge
on AI platforms 7-8 (In particular he quotes: “I am not an engineer and
don’t know how algorithms work, but I do know how to use them.). an
average of 7
3. An average of 6
6. What does AI mean to you in a general sense?
1. Automation contributor, and a good assistant in social development. AI
is a replacement for human intelligence in the areas that it can
substitute for human intelligence.
2. Code and algorithm for the machine to think by itself and provide
results (no requirement for human input and manipulation).
3. A way to bring together technology and people, technology
approaching the people rather than people approaching technology.

41
 7. What does AI mean for your tasks as a marketing manager? Follow up: How
has it affected your organization?
1. Based on the limited experience and knowledge_ She suggested that it
will be widely used for personalized marketing and automation.
_Timesaver in various areas such as customer service etc. however,
customers seem to prefer human communication.
2. AI supports data analysis in a much higher capacity than humans. As a
result, it provides people with the capacity to focus on more strategic
roles. (In relation to the research gap he mentions regular clients that
may be excluded from the marketing campaign in such strategies.)
3. Has a generally positive impact on the financial and strategic sector.
Power BI is used to keep track and classify specific segments.
8. What skills do you think that marketing managers should develop for the best
practice of SEM? (Technological understanding, mathematical perception,
understanding the AI tool for its strengths, weaknesses, and capabilities)
1. Familiarize themselves with relevant emerging tools, general emerging
trends, and finally having programming skills can increase their
problem-solving ability. From personal experience Having
programming knowledge may save time and energy. Thus, highly
appreciated in the industry. Finally, technical understanding may
facilitate more rational decision-making and attention to detail.
2. Analytical, the patient, detailed, and constantly learning. From the
provided examples, mathematical perception for SEM.
3. Analytical, multitasking capacity, versatile. From the provided
examples, technical understanding

Section 2: return on investment in SEM & future improvement:

1. Search engines capable of recognizing regular clients to some webpage and in

advance excluding them from search engine marketing. Why would this be
required? How is this applied in search engine marketing?
1. She suspects that the first step will be automated in the future. Couldn’t
provide an intuitive answer on why excluding some population of the

42
 audience may be a good idea. Gusset that such exclusion is achieved
by adjusting keywords on advertising.
2. No overlapping (marketing campaigns are not targeting the same
audience or segments of the same audience), in attracting new users/
clients you may want to separate/ exclude customer list (such
customers are found to be irregular and not matching with your
campaign). Customers that you don’t need or don’t want to impact, and
no overlapping. Exclude audience that already purchased.
3. Exclude customers that already purchased.

2. What possible improvement in knowledge search engine marketing would you

suggest?
1. Familiarize yourself with as many SEO & SEM tools that come into
the market Learning some of these tools and having the ability to
evaluate the differences between them. Such tools can do half the job.
It is important to find the silver lining in between what this tool can do
and how marketers’ knowledge is best applied.
2. Shopping ads should be improved in dynamic and creative ways. Start
developing a strategic plan for Cooky_less environments. Implied Giff
3. It should be forbidden to buy a competitor’s keyword and not
necessary for brands to pay for loyalty. Giff or other interactive
elements in shopping ads. Automation in updating information for all
site domains on a global scale.

Section 3: Explain AI augmentation & Automation:

1. How do you think the relationship between firms and consumers will evolve
because of AI, and how will it affect your role as a marketing manager?
a. Automation will keep accelerating an augmented process. It may be
seen as a threat for consumers and an opportunity for marketers.
b. Consumers are not completely aware of the impact that their actions
have, and how data is handled. This is also reflected in brands. Brands
should consider their interaction with a user and what reputation they

43
 build around their data privacy etc. The intention of industrial
development and big corporations such as Google, Facebook etc. is to
enhance AI capabilities. However, this is dangerous for the business
because they lose control. The loss of control is feared by marketing
companies. (“automation means less control for me”)
c. Companies should be informing their consumers on how AI is
affecting data manipulation. From a marketing prospect, increasing AI
requires additional structural development. In its current state, there are
issues with the system infostructure and is not being utilized
inappropriate ways. It is being used in areas that are better fulfilled by
humans.
2. What decisions do you make today that could already now be automated
and/or supported with AI? Follow up: How do you think it will look in the
coming years?
a. Analysis still requires human input in the overall analysis
measurements and parameters. This may be completely eradicated.
b. Automating strategic steps and/or measurable variables that are not
there yet
c. Automating search engine marketing by optimizing keywords based on
the best possible combination for the target audience. (A new arising
question is how this audience is measured;). There are already some
platforms doing such things however, they are paid for (e.g. sea360).
There are lots of things that can be done at the moment but they are
paid for.
3. How do you make use of your creativity in your role as a marketing manager
(in SEM)?
a. Implement strategies, arrange customer journeys, and personalise the
experience based on user behavior requires creative and abstract
thinking.
b. Being strategic is part of being creative.
c. Implementing strategies and designs, ways to develop and adjust the
content, and in general enhancing the work process and reputation of
the company.

44
 4. How do you make use of skills such as persuasion, social understanding, and
empathy in your role as a marketing manager? Follow up: do you think these
skills will be more important to possess as a marketing manager?
a. Such skills provide the ability to understand your customers, frame and
segment them into appropriate categories. It is an analogical strategy
for customer segments. Such skills may be considered more important
because of their foundational ability to generate new applicable ideas
for the client’s interest. Such skills may be considered a foundation in
AI development...
b. These skills are important to understand in internal company
interactions. While in the general sense of marketing the overall
purpose is to sell, regardless of how this is achieved. 50-50.
c. 50-50

AI experts’ interview questions

Section 1: Audience introduction, demographics & Perspectives.

1. Tell me about your background and what you do in your work, as well as how
it relates to AI?
1. AI reasoning and computational complexity (Investigates the relation
between AI and intelligence.)
2. Adding confidence to AI prediction. (How the existence of well-
celebrated probabilities should be used. Risk-taking and measures.
Probability etc.)
3. Specialist in machine learning. Research, theory, and practical
implementation of machine learning in private sector.
2. How well do you understand marketing, Especially the relationship between
AI and marketing? (1-10)
1. 4
2. 7
3. 9-10
3. Do you think the future of AI in marketing is augmented or automated?
1. Partial automation at this point. Far from complete automation.

45
 2. Augmentation in the short term. Automating marketing strategies is far
in the future.
3. Far from complete automation. Possibly no completely automated AI
for the next 20 years
4. What is the biggest beerier in future AI development for marketing purposes?
1. Emotions and feelings. Getting AI to understand how emotions impact
behavioral sequence and its effect on creativity.
2. Skepticism regarding the trust of an automated system to do the work
as well as we can (human creativity within the system. The system
works well within a framework, but less so in new arising variables).
Understanding how to structure knowledge in a way that is optimal for
the marketer (fool the system to work for you).
3. AI can be used for very specific tasks (We need to define the problem
and build solutions for ourselves). We can not define intelligence in a
general manner within AI. It is difficult to design systems that
understand the human brain and work in the same patterns. Thus, we
can not develop systems that go into uncharted territories and solve
open-ended problems.
5. Given this development, what skills do you think will be most important for
future marketing managers? (Technological understanding, mathematical
perception, understanding algorithmic applicability)
1. Marketers should understand the problems and weaknesses of the
system in place. Getting the system to a point that does not suggest
biased results.
2. Understand the capabilities and how/when to use these capabilities to
deliver value. Understanding how to define problems that are suitable
to use AI for.
3. Know the boundaries of AI systems. Know when to use and not use a
particular solution.
6. How important is it for marketers to have AI technological understanding and
algorithmic applicability in search engine marketing or machine learning in
general? (1-10)
1. More knowledge is always better.

46
 2. 3-4
3. When it comes to knowledge in possibilities and risk assessment, it
should be as much knowledge as possible (close to 10). When it comes
to algorithms and coding it is close to 0.

Section 2: Machine learning and status of AI in marketing:

1. Can machine learning in its current state label new data and autonomously
generate algorithms suited to business strategies? (Is this a Google AI capacity
or any other AI)
a. On an experimental level, a few machine learning tools are being
developed to generate algorithms applicable to business strategy.
b. It can exist as a tailored solution for a company.
c. There are no systems available that can comprehend and develop
complex system strategies. The systems can not teach themselves to
become smarter and solve problems in new ways. No open-ended
questions that AI can solve for itself.
d.
2. What do you think AI will mean for marketing companies, consumers, and
society in the long term?
a. Companies must need to incorporate AI in their strategies. Thus,
utilizing its advantages and keeping up with the competition (game
theory). While for consumers, it is very important that they understand
how ads connect to data collection.
b. Neural networks can be used to manipulate consumer populations.
c. Marketing companies will keep building new products and services
based on currently existing products and services (Real-time data).
3. In your opinion, what type of marketing opportunity do search engines provide
(e.g. branding, sales, etc.)?
a. Search engines provide the ability for small companies to get exposure
through applying smart strategies. By utilizing knowledge, small
companies can increase exposure and compete with bigger companies
that have the capital to buy space based on expensive keywords.

47
 b. Exposure. The goal is to be seen before your competition.
c. Cross running different data collection and information streams.
4. How do you think that AI applies in such search engines? Can you provide
some examples?
a. Google AI is highly appliable to historical data through machine
learning. Search engines provide unique and personalized content to
each client based on browser history.
b. AI is applicable through machine learning. Machine learning is
generally what applies to almost all modern AI systems.
c. Google AI is highly appliable to historical data through machine
learning. They make predictions based on a highly heuristically built
framework from historical data. Also, Google AI utilizes data
collection from Facebook, Instagram, and other social platforms.
5. In your opinion, how well do marketers understand web infostructures and /or
other facilitating AI platforms & tools? (On average...)
a. Poor understanding of AI and what it can be used for. Improving their
understanding of AI, can improve their marketing strategy and
utilization of such tools.
b. An expectation of 6 in general & 8 on search engines.
c. 6-7
6. What possible improvement in knowledge search engine marketing would you
suggest?
a. Marketers should know the behavior of the system. How this affect the
input and output of the system.
b. Should learn to identify business opportunities using AI.
c. Marketers should realize the boundaries of the utilized tool. This
means both its limitations and capabilities.
7. What should be improved in search engines?
a. Search engines should not hold history information on clients.
b. Make it easier for the marketers to find solutions and gain more
information based on the collected data.
c. Make it possible for marketers to extract more knowledge. Search
engines can provide more than a prediction, they can provide a reason

48
 on why the client/ target audience is a good match for this campaign.
(why is the system making these decisions and how confident is the
system in its prediction)(this is ongoing research on how to do that in a
good way)

Appendix B
Marketers’ Interview questions

To answer the research questions, this research has interviewed two participatory
groups. The focus has been slightly different between the two groups. The interviews
examine marketers for their expectations from Google AI, and what knowledge should
be considered for the best utilization of Google AI. For example, what knowledge they
find to be advantageous based on their empirical knowledge in SEM. The research
questions for this target group have been divided into three sections:

1. The first sections of the interview examined the individuals for their marketing
specialization, experience, work contribution, and relevance to AI. Also, this
part asks the participants about their opinion on AI contribution to marketing
and what they find to be useful knowledge in utilizing Google AI.
a. The participants have had a much different background from one
another. They have all received a Google certification for SEM, but only
one of the participants is highly experienced due to his position.
b. All participants seem to believe that marketers have a good
understanding of web infrastructure and what AI can do with an average
of 7/10. This was later shown in their answers that are very close to the
collected literature.
c. Regarding AI’s usefulness to marketing, they all find AI to be a tool for
high levels of data analysis and automation. While they believe that
marketers’ job is a strategic role. The participants seem to agree that
marketers should possess analytical thinking and constantly learning
new possibilities to improve on their work. They find technical
understanding as an addition that could facilitate problem-solving in
some situations but is not necessarily required.

49
 2. The second section examines marketers’ knowledge and contribution to return
on investment. Also, this section explores possible suggestions for Google AI
improvement in SEM.
a. On audience exclusion and return on investment, the participants seem
to understand the methods for improving their conversion rate. They
know that the requirement is to exclude reoccurring costs and approach
the assignment from an analytical point of view.
b. Regarding the possible improvements for Google AI, they suggest some
interesting ideas as well as news on the future development of Google
AI. For example, interactive and dynamic shopping adds, automation for
updating information for the same organization on a global scale. Or in
terms of planned changes in Google AI future, one of the participants
mentioned the development of cooky-less systems.
3. The third section of the research examines marketers’ perspectives on the
augmentation and automation of AI. The intention has been to understand their
perspective on the current capabilities of AI in the market. Also, this section has
investigated the current capabilities of automation, and the way that automation
affects the job of a marketer.
a. Regarding AI development, marketers show consideration on different
levels. They all believe that there is a benefit for the marketers, but
tremendous downsides for consumers. But they also show consideration
to the job of a marketer. They say that by implementing high levels of
automation, marketing companies are afraid to lose control and
consciousness of the decision-making.
b. In terms of automation, they find that there are a lot of processes that are
automated but not widely available. While the processing of information
by marketers is seen as creative through its strategic applicability. By
removing such processes and adapting automation, this part of the
process is highly removed.

AI experts’ interview questions

The interviews have examined AI experts for their knowledge on AI, and what is the
contribution of AI in search engine marketing. The research questions for this target
group have been divided into two sections:

50
1. The first section of the interview examined the individuals for their AI
specialization, experience, work contribution, and relevance to AI. This section
is a demographic determinant to separate the knowledge of participants to
marketing and applicability that AI may have in SEM. Finally, this section
investigates the required knowledge of marketers in AI, examining the opinion
of AI experts on the knowledge areas that could help a marketer better utilize
Google AI.
a. The AI professors that have been interviewed had different specialties.
Their marketing understanding averages 7/10. They believe AI to be far
from automated and will remain augmented for at least the next 20 years.
b. Regarding the current barriers in AI development for marketing, they all
recognize that it is difficult to design systems that are truly intelligent.
Ai cannot understand the complexity of humans, human brain and
cannot solve problems in uncharted territories.
c. Based on the development of AI, the participants suggest that marketers
should understand the problems and weaknesses of the system at pace.
This way marketers can get the system to a point that does not suggest
biased results. Finally, while acquiring knowledge is always good, the
participants believe that technical understanding in AI or algorithms is
not necessarily required.
2. The second section examines machine learning and the status of AI in
marketing. This implies the current capabilities and possibly experimental steps
towards automation. Their perspective and evaluation of the best contributions
that AI has on search engine marketing.
a. Even though the current systems that exist can develop complex systems
and strategies //But//, the systems cannot teach themselves to become
smarter.
b. Regarding the influence of AI on marketers and consumers, the
participants believe that companies will keep improving on the currently
existing systems to keep up with the competition. While consumers need
to be better informed about the effect that AI has on their life.

51
c. The participants here believe that the biggest competitive advantage
provided by SEM is exposure. By applying smart and creative strategies
to the information stream, small companies can compete.
d. In terms of technical applicability, Google AI is highly applicable
through machine learning and the evaluation of historical data.
e. The participants believe that marketers’ knowledge of web
infrastructures and search engines averages 7/10.
f. Regarding the improvements that AI experts are expecting from
marketers and search engines. Marketers are expected to know the
behavior of the utilized system, and the effect that data input can have
on the business opportunities. While in terms of SEM improvements,
they believe that marketers should be able to utilize a broader range of
information and, thus, knowing having more control over the results.

52