State of the Art Implementation of Big Data in Strategic and Operative Marketing. Challenges and Effective Responses

Table of contents

List of abbreviations

List of figures

List of tables

1 Introduction
1.1 Research background
1.2 Purpose and research objective
1.3 Methodology and structure

2 Theoretical background of data-driven marketing
2.1 From database marketing to data-driven marketing
2.2 Big Data

3 State of the art regarding implementation of Big Data in marketing
3.1 State of the art marketing research
3.2 State of the art target marketing

4 Challenges of successful Big Data implementation
4.1 Strategic approach
4.2 Data regulations and privacy
4.3 Human resources, internal structures, and corporate culture
4.4 Technical infrastructure and centralizing data
4.5 Data quality and data volume
4.6 Acceptance and focus on new Big Data processes
4.7 Correlation-driven analysis vs. causality-driven analysis
4.8 Courage for trial and error

5 Conclusion, recommendation, and outlook

6 List of references

7 Appendix
7.1 Appendix A
7.2 Appendix B
7.3 Appendix C
7.4 Appendix D
7.5 Appendix E
7.6 Appendix F
7.7 Appendix G
7.8 Appendix H
7.9 Appendix I
7.10 Appendix J
7.11 Appendix K

Acknowledgements

I would like to express my gratitude to you Prof. Dr. Jugel. You have been supporting me to great extent with your advice, responded to my various questions remarkably fast at almost every time of the day, led multiple coaching sessions, and enabled me to choose a research assignment that I am passionate about.

My gratitude also goes towards the experts that have participated in the expert interviews and donated multiple hours of their valuable time to me after their normal working routine.

I also would like to express my deepest gratitude to my family who has not only supported me vigorously during the last months and years, but more importantly, who formed me as a person and has been the only reason that I could fully benefit from our educational system and strive for a Master’s degree.

List of abbreviations

Abbildung in dieser Leseprobe nicht enthalten

List of figures

Figure 1 Annual size of the global datasphere

Figure 2 Data generated every minute in 2020

Figure 3 The “6 Bigs” of BD

Figure 4 BD sources

Figure 5 Traditional business intelligence infrastructure

Figure 6 BDA - Process.

Figure 7 BDA – Ontology part 1.

Figure 8 BDA – Ontology part 2

Figure 9 Traditional marketing research process

Figure 10 Search for the unknown and analysis of patterns

Figure 11 More sophisticated segments through BDA

Figure 12 US programmatic digital display ad spending

Figure 13 Survey results regarding LBA’s benefits, use intentions, and location data features

Figure 14 Factors that drive the dynamic pricing of the Major League Baseball

Figure 15 Optimized product innovation processes through the implementation of BD

Figure 16 Challenges of successful BD implement

Figure 17 The value chain of BD

Figure 18 Competitive advantage through highly customized marketing mix based on BD

Figure 19 Taxonomy of BDA

Figure 20 BDA infrastructure – Hadoop in symbiosis with the traditional data warehouse

Figure 21 New and conventional data types

Figure 22 Programmatic advertising process

List of tables

Table 1 Sample of web-analytics key ratios

1 Introduction

1.1 Research background

“Data is the new gold […]” and “The world’s most valuable resource is no longer oil, but data” . Headlines like these, published by The Economist and the World Economic Forum (WEF) in the year 2017 and 2020, underline the crucial role data plays in business nowadays.

And while gold and oil belong to the non-renewable resources, data is not only endless but also increases at an exponential pace. In 2018, the global volume represented 33 Zettabytes (ZB), or 33,000,000,000,000,000,000,000 bytes.¹ It is predicted to double until 2022 and to double another time until 2025.² Sometimes it is hard for the human mind to grasp such large numbers. To illustrate these dimensions: One ZB is the equivalent of a trillion gigabytes (GB). And if one stored the entire global datasphere from 2025 on DVDs, this would result in a stack of DVDs covering the distance to the moon 23 times or circle the earth 222 times.³ However, it is not the large volume that makes data so appealing, but rather the value that it can represent. Amazon’s, Google’s and Facebook’s market value is directly related to the value of their data and how they harness it.⁴ All three companies are now amongst the most valuable companies worldwide.⁵

The fact that data would greatly impact our lives is not surprising in the year of 2020. “Big Data” (BD) and its significant influence on fields like health, security, administration, and marketing have been discussed and elaborated extensively in Germany since 2012. Especially Big-Data-Marketing offers great opportunities to companies.⁶ More and better customer data have long been marketers’ “Holy Grail” at times when it was difficult to collect them and now the dream seems to come true.⁷ The right data, if collected, extracted, and interpreted correctly, can help to understand and address the individual needs of customers more appropriately. The application opportunities are vast, they range from personalized communication and identification of the most cost-effective channels to anticipating trends and minimizing investment risks.⁸

Nowadays, not only the tech giants like Google, Facebook, and Amazon understand the importance of data, but many companies are investing significant resources into BD technologies,⁹ assuming that these enable disruptive business model innovation,¹⁰ accelerate business transformation¹¹, and drive superior performance¹². According to a survey from 2019 of senior executives in Fortune 1000 and industry-leading US companies, 91.6% of companies are increasing the pace of their BD investments. They increasingly realize that only by allocating appropriate human, physical, and organizational resources to BD, one can benefit from this new form of capital.¹³ However, even though many companies are investing in BD, the survey also revealed that only 62.2% of the respondents reported measurable results from such investments.¹⁴ Hence, it is crucial for businesses to gain more knowledge of how BD investments can turn into a competitive advantage.¹⁵ One reason for the lacking value translation of BD investments is the presence of various challenges companies are facing. These challenges comprise the technical infrastructure and feasibility of BD, the question how to harness BD effectively, the formation of new structures within the company¹⁶ as well as cultural changes.¹⁷

Additionally, there are further factors that influence the access to and the work with data immensely. Data protection issues have been prioritized in many countries around the world in the past years.¹⁸ In the European Union (EU), for instance, the General Data Protection Regulation (GDPR), which became enforceable in 2018, was consistently improved and reviewed in June 2020.¹⁹ The GDPR gives more control to the customers regarding their data which means that a customer needs to give consent before a company can use his/her data.²⁰ Moreover, the European Union resolutely controls and punishes any regulations breaches. Google, for example, was issued a record fine of 4.34 billion in 2018 for illegal practices.²¹ Events such as the National Security Agency (NSA) scandal, uncovered by Edward Snowden, as well as other data leaks taking place regularly have shaken the public’s confidence in data security.²² In 2017, a study by PricewaterhouseCoopers disclosed that 69% out of 2,000 surveyed Americans, believed that companies are vulnerable to hacks and only 25% found companies handle their sensitive personal data responsibly.²³ Once a customer lost trust and withdraws the permission to use and store his/her data, it will be difficult and expensive to regain his/her permission, similarly to regaining a lost customer.²⁴

Despite the various challenges, the huge potential of BD is already changing the way business is done today and will certainly become an even greater driving force in the future. The question is just to which extent and how quickly this process will occur, and which economic fields will be the most affected ones.

1.2 Purpose and research objective

As mentioned in the introduction, BD finds its application in various fields, e.g., the health and security sector. However, this paper focuses exclusively on BD’s impact on marketing. Also, due to the rather technical nature of BD, there are many technical differentiations and elements that come along with the subject. In this paper, only the fundamentals of such technological elements will be briefly described without going in greater depth. The purpose of this paper is rather to introduce the latest developments regarding BD and marketing and to answer the question why BD seems to become or even already is such a driving force in marketing. Here, the focus will be set on the marketing mix’ product, promotion, and price whereas place will not be dealt with. Moreover, the reasons for a slow implementation or even rejection of BD shall be analyzed, followed by a recommendation what companies need to consider before investing in BD, and how they can implement BD afterwards in an effective way. The research questions for this paper are as follows: How is the state of the art regarding the implementation of Big Data supporting and enhancing strategic as well as operative marketing activities? Which challenges are companies facing while adopting Big Data and how can they respond most effectively to them?

1.3 Methodology and structure

This thesis is divided into five chapters. After the introductory chapter one, the theoretical background of data-driven marketing will be carefully examined in chapter two. This is done by investigating how the significance of data in marketing has evolved, analyzing what marketers mean when speaking of “Big Data” and how BD is characterized, identifying the most important BD sources, as well as by pointing out which technological perquisites are necessary to use BD most efficiently.

Chapter three deals with the detailed analysis of how BD is employed to supplement, optimize, and transform traditional systems into strategic and operative state-of-the-art marketing activities. In this regard, BD’s impact on marketing research is thoroughly examined, before the influence of BD on target marketing is scrutinized. More specifically, the analysis’ focus is set on how BD helps to optimize segmentation, promotion, price, and product. With regard to promotion, the new level of personalized communication is investigated before analyzing more closely personal communication techniques such as Programmatic Advertising (PA), location-based advertising, and the new form of engagement. Regarding price, the pricing strategies, dynamic and personal pricing, as well as the new role of personalized coupons due to BD is closely elaborated. Finally, BD’s substantial impact on product, in form of an optimized product innovation and improvement as well as refined product-related customer service is outlined.

In chapter four, a meticulous investigation is conducted regarding the main challenges companies encounter when intending the implementation of BD in existing systems and processes.

Chapter five then summarizes the main findings of this thesis and gives an outlook with respect to BD’s impact on marketing as well as a recommendation how to address the main challenges and harness the potentials pointed out in chapter three.

Throughout this thesis, the thoroughly conducted literature review is supplemented by the analysis of two expert interviews. Expert 1 has great expertise in working with BD in channel management. Expert 2 works as IT-project manager and provides valuable insights from the technical point of view. Please refer to the appendix for a more detailed description of the experts and their companies.

2 Theoretical background of data-driven marketing

2.1 From database marketing to data-driven marketing

Companies can expect great advantages by data-driven marketing. According to a survey of marketing experts by the Experton-Group, the number one expectation of BD is a better insight into customer behavior, followed by more targeted advertising campaigns and a better evaluation of market potentials.²⁵ But what is data-driven marketing exactly?

In the early 1980s, electronic data processing costs declined substantially, while marketing costs increased significantly.²⁶ Technological advances in telecommunication and computing facilitated the accumulation, management, and analysis of comprehensive customer-related data and lowered the financial and time investments necessary for handling such data.²⁷ Simultaneously, marketers observed the traditional media dropping in performance and increasing in costs due to factors like the rise of the ”information society“, an increase of competition, a reduction in leisure time and the fragmentation of consumer and business-to-business markets.²⁸ Consequently, in the face of such developments, database marketing (DBM) became a very attractive alternative.²⁹ One of the first definitions of DBM was given in 1987: “DBM is an interactive approach to marketing communication, which uses addressable communications media (such as mail, telephone, and the sales force) to extend help to its target audience, to stimulate their demand, and to stay close to them by recording and keeping an electronic database memory of customer, prospect and all communication and commercial contacts, to help improve all future contacts.”³⁰ In short, the marketing activities of companies shifted towards a more customer-centered approach enabled by a better integration of customer-related data.³¹ As a consequence of this development, every single department of a company had its own database at the turn of the millennium. With the introduction of the customer relationship management (CRM), an integration of all such customer data was pursued.³²

Kotler et al. (2016) describe CRM as maybe the most important concept of modern marketing. In the beginning, CRM was interpreted as the maintenance of customer data and the management of detailed information regarding every customer with the objective to maximize customer loyalty and retention. Nowadays, CRM is understood in a broader, more comprehensive sense: The entire process of building and retaining profitable customer relationships is based on high customer benefit and customer satisfaction. It integrates all aspects of the acquisition, retention, and development of customers.³³

The shift in marketing taking place is clear: Moving away from traditional approaches towards online-oriented, dynamic, and analytical marketing decision making based on BD sets.³⁴ Marketing today is much more technological than it used to be. The old stereotype of the marketer as creative designer who mainly focuses on magnificent graphics and catchy slogans is outdated.³⁵ Know-how regarding the handling of relatively larger data sets could already be established through DBM and CRM. In the future, however, the analysis of data streams will probably be an essential step of bringing products to market which makes BD analysis an integral part of marketing.³⁶ Also expert 1 believes that the marketer of tomorrow will need strong analytical capabilities and the skill to understand data. ³⁷ Additionally, it is not only the pure characteristic of future products but also the fact that many consumers highly welcome and start to expect personalized and real-time engagements being founded on their very personal data.³⁸

The significance of data is also underlined by Kotler et al. (2016) who state that data represents an integral part of CRM. To create customer value and build close customer relationships, marketers need current and especially insightful data that they can react upon.³⁹ The ongoing digitalization, the Internet of things (IOT)⁴⁰ and in particular technological revolutions such as social media enable marketers to access more and more of such data.⁴¹ Aspects and artifacts of the daily life like profiles, postings in blogs and social media, shopping histories and health records, just to mention a few examples, generate an increasing amount of information which is referred to in a whole as: “Big Data”.⁴² To give a better understanding of the term, the next chapter serves to define BD in detail and introduce some of the most acknowledged characteristics found in literature.

2.2 Big Data

2.2.1 Definition of Big Data and its characteristics

There are several definitions of BD, and among the most cited ones is a definition given by the Federal Association of Information Technology and New Media (Bundesverband Informationswirtschaft, Telekommunikation und neue Medien, BITKOM): “Big Data refers to the analysis of large amounts of data from a variety of sources at high speed with the aim of generating economic benefits.”⁴³ It further describes that “Big Data is always present when an existing company infrastructure is no longer able to process the data volumes and data types in the necessary time.”⁴⁴ A more marketing-related definition was given by Gartner, a global consulting company: “High-volume, high-velocity and high-variety information assets that demand cost-effective, innovative forms of information processing for enhanced insight and decision making.”⁴⁵

The characteristics of BD have been compiled in various publications. Among the first was a paper by Laney (2001) who has introduced the three “Vs” to describe the characteristics of BD - “volume, variety, and velocity”.⁴⁶ Later, “veracity” was added by IBM and formed together with “value” the “5 Vs”. These ”5 Vs“ have been used the most by academics and in business ever since to define BD.⁴⁷ However, in the course of time, the number of “Vs” was increased by different authors up to 24.⁴⁸ Sun (2018) expanded the “5Vs“ by five additional keywords characterizing BD. “Big Data Analytics”, one Sun’s added characteristic, plays a particularly significant role and will be elaborated more extensively in chapter 2.2.3.2.⁴⁹ Please find the complete collection of Sun’s “10 Bigs” under Appendix A.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1 Annual size of the global datasphere⁵⁰

“Big Volume” represents, as already depicted previously, the ever increasing and extensive amount of data (Figure 1). According to the International Data Corporation (IDC), the size of the global datasphere was around 33 ZB in 2018 and will reach 175 ZB in 2025. This analysis, however, was carried out in 2018. With the Coronavirus sweeping the world, nearly every aspect of life has moved online and is presumably driving the annual global data volume to even higher levels than predicted in 2018. An illustration by Domo (2020) reflects this increase impressively (Figure 2).⁵¹

Abbildung in dieser Leseprobe nicht enthalten

Figure 2 Data generated every minute in 2020⁵²

Due to the quickly increasing data volume, there is no absolute limit or number of bytes that would define BD. Such a definition or limit would quickly be outdated. BD is not “big” as soon as it reaches a specific limit but as soon as it leads to results that would have been impossible with less data.⁵³ Great volume creates promising chances, because the larger the amount of data the higher the probability of insights.⁵⁴ On the other hand, companies are overwhelmed by the size of such data volumes and struggle with their current infrastructure and know-how to translate all such information into valuable insights and finally into economic benefits.⁵⁵ The challenge is to identify what is relevant and then quickly extract that data for timely analysis.⁵⁶

“Big Velocity” refers to the increasing speed at which data is produced and needs to be analyzed and processed.⁵⁷ It describes the high rate at which data flows in and out interconnected systems in real-time. In some developing countries, such IT-systems are not advanced enough to harness BD effectively, even though BD would be accessible.⁵⁸

From an analytical perspective, “Big Variety” is probably the biggest obstacle for the effective usage of large data volumes. “Big Variety” describes the highly variable sources, such as mobile devices, e-mails, distributed sensors, web pages, etc. that provide data in different form, e.g., audio, text, images, video, and multimedia. These data sets can be divided into structured, semi-structured and unstructured data. Structured data is characterized by a unified data format and can be analyzed and managed easily with standard database tools. Unstructured and semi-structured data, on the other hand, are more difficult to analyze and understand. Hence, a more advanced infrastructure is indispensable for processing and adequately managing data from various sources. BD platforms therefore have to deal with a high degree of complexity, which represents a major challenge for the value translation.⁵⁹

“Big Veracity” represents the accuracy and quality of the data. It defines whether data can be trusted when significant decisions need to be made based on the collected data. Inconsistency, ambiguity, or incompleteness, for example, are reasons why data are categorized as good, bad, or undefined. With increasing volume and diversity of sources, it is more challenging to quantify the accuracy of incoming data.⁶⁰

“Big Analytics” constitutes the techniques that are used to acquire, analyze and visualize intelligence and knowledge from BD.⁶¹ BD has little value without sophisticated Big Data Analytics (BDA), similar to the availability of oil without having access to the significant advances of the petrochemical industry. Only by deep, smart, and intelligent processing of BD, value translation can take place. And thus, BDA represents the core technology that turns BD into valuable insights. It therefore holds a more important function than BD itself.⁶²

Finally, “Big Value” represents probably the most important “Big” compiled by Sun (2019). It signifies the direct value, measured for example as Return on Investment (ROI) that BD creates for the company.⁶³ “Big Value”, however, can only be generated if BD can be transformed into valuable insights and domain-specific conclusions in the first place.⁶⁴ Therefore, it is crucial to develop a deep understanding for BD and how to harness it since valuable insights do not emerge from data alone.⁶⁵ Hence, the value lies in the meticulous analysis of precise and accurate data.⁶⁶ Figure 3 summarizes these “6 Bigs”

Abbildung in dieser Leseprobe nicht enthalten

Figure 3 The “6 Bigs” of BD⁶⁷

2.2.2 Most significant Big Data sources

As described in the previous subchapters, BD is everywhere these days.⁶⁸ Figure 2 in 2.2.1 impressively shows the many different sources where data is generated and where it can be extracted from. Nonetheless, this subchapter does not serve to give an overview about all available sources but rather the most significant BD sources from a marketer’s perspective.

Data sources can be classified as internal or external sources. External data sources represent all sources outside of the company, publicly available or bought from third parties. Typically, great market research institutes or other external service providers are used as external data sources to analyze, for instance, a company’s performance in relation to the market average, the market leader, direct competitors, etc.⁶⁹

Abbildung in dieser Leseprobe nicht enthalten

Figure 4 BD sources⁷⁰

Figure 4 points out the most significant BD sources. The probably greatest and most important BD source is represented by the Internet of Things (IoT), which refers to a variety of "things" that are connected via the internet so that they can exchange data with each other.⁷¹ It enables the connection of people (e.g., workers, consumers, customers), systems (e.g., business applications, analytical systems, enterprise resource planning systems) and ”things“ (e.g., machines, computers, mobile devices, and sensors) with each other allowing them to exchange information and generate data. These days, billions of connected objects are interacting, partly autonomously, with each other all around the globe. Machine to Machine communication (M2M) via, for instance, automatic sensors generates a huge amount of data and is seen as subset and integral part of the IoT.⁷² Therefore, a variety of data from IoT sources can be used for marketing purposes. This data includes, for instance, control or setting data entered by the product owner, sensor data, automatically arriving at real-time speed from the product itself and informing the receiver, e.g., about speed, temperature, and location of the owner by making use of the Global Positioning System (GPS), as well as others.⁷³ In other words, marketers can understand and analyze how the customer is using or reacting to a product in real-time and respond adequately.⁷⁴ IoT can be an internal BD source when it is used to acquire and transfer data from company-owned products and systems but can as well represent an external BD source when it is used to access products or sources that do not belong to the company. Furthermore, IoT lays the basis for the following external BD sources.

Smartphones are another major BD source. Apple’s first iPhone was introduced only 13 years ago in 2007 but revolutionized the industry and gave smartphones its mainstream popularity. Nowadays, around 1.56 bn smartphones are sold each year to end users.⁷⁵ In total, around 3.2 bn smartphone users exist around the world in 2019. The number is expected to rise up to 3.5 bn in 2020 and up to 3.8 bn in 2021.⁷⁶ Already in 2017, almost every developed country reached a smartphone penetration of over 90% of the population. Furthermore, the gap of smartphone penetration initially observed in developing countries has been closing considerably over the last years.⁷⁷ Modern smartphones are computationally powerful and sensor-rich companions constantly recording information about, e.g., movement patterns, daily activities, and social behaviors of their owners. Smartphone sensing systems run consumer-oriented applications (“apps”), which allow the collection, storage, and streaming of large data volumes from consumers.⁷⁸ Additionally, the mobile device is now a fixture of modern living and used for a vast amount of activities such as surfing the internet, texting, watching a favorite show, but also for paying bills, shopping, and even remotely controlling other devices. Thus, smartphones give marketers deep insights into the lives of consumers and potential customers.⁷⁹ But it is not only the variety of data but also the intensity and trend of smartphone use that make these devices so interesting for marketers. In 2019, the average US adult spent almost 3h per day with his/her smartphone.⁸⁰ And since the consumption of traditional media such as newspapers, magazines, television, radio, and desktop internet is more and more declining, the importance of mobile internet consumption via smartphones is expected to keep on increasing in the future.⁸¹

Social Media also represents a significant external BD source enabling users to generate data much faster than ever before.⁸² Apart from prominent social media companies like Facebook, Instagram, YouTube, Google+ and Twitter, social media in a broader sense also includes blogs, forums, messaging services, wikis, location-based services, crowd-sourced content, as well as product reviews.⁸³ All these platforms reveal consumer interests and preferences including, e.g., favorite artists and brands. However, in-depth analysis of information shared publicly by many users even allows conclusions about sensitive aspects such as the sexual and political orientation, the family status, or the financial situation.⁸⁴ According to a survey by BITKOM from 2018 in Germany, 87% of the population replied to be registered on social media. Among the 14-29 years old people 98% stated to be social media users and this was still the case for 65% of respondents being older than 65.⁸⁵ Social media enables a very direct way of communication. Users can participate in conversations and share their opinions regarding products, services, and companies.⁸⁶ BDA permits to automatically monitor these opinions in large volumes by mining the social media data. The conclusions drawn by marketers from these analyses then represent key factors for strategic marketing decisions.⁸⁷ Hence, it is not surprising that already in 2017 90% of United States (US) companies with more than 100 employees analyzed social media content for marketing purposes, and the proportion is predicted to rise.⁸⁸ Companies have long understood that social media has a huge impact on customers’ decisions.⁸⁹ However, social media’s massive amount of data is widely unstructured and difficult to analyze.⁹⁰ The analysis is a complex procedure due to subjectivity in text review and additional features in raw data like irony and sarcasm.⁹¹ Moreover, it has to be taken into account how representative the extracted data is and which part of the population is the most likely to show up on such sites. Several sociodemographic factors predict who is more likely to adopt certain sites and platforms. Businesses cannot assume having a sample of the whole population, even though an intensive adoption of social media throughout the population is given.⁹²

Search Engines are another important external BD source. According to an international survey from 2019 by MOZ, an international consulting firm, 77% of the respondents stated to use Google three times or more a day.⁹³ The latest exact data provided by Google revealed approximately 3.3 bn Google searches worldwide per day.⁹⁴ However, this number origins from 2012 and, due to the advancing digitalization, it is more than likely that this number has increased substantially up to date. But similarly to the other BD sources, it is not only the data volume that make search engines interesting for economic purposes but also the specific insights they allow. People use search engines primarily for research, shopping and entertainment, and thus, internet searches directly reveal what users are currently interested in.⁹⁵ Highly “intelligent” algorithms analyze how long you stay on a page, whether you directly change pages or whether you ignored an ad in the first place. Consequently, Google and other search engines understand your interests and motivation with every interaction slightly better and reveal insights into their users’ personality that are highly valuable for economic exploitation.⁹⁶

Change of lifestyle and habits, documented digitally as “quantified self”, is another major driver of BD. More and more consumers record and willingly share sensitive personal data by using smart-watches or fitness-trackers. This quantitative data on, e.g., nutritional habits, sports, daily sleeping time, or body temperature allows deep insights in the daily activities, fitness, and health status of the sharing person. Obviously, this information can be exploited economically when being made available for outsiders.⁹⁷

Technological adaption and development will continuously generate and improve new ways to obtain insightful data. Google glasses, e.g., bear the potential to track consumers’ focus and vision in daily life.⁹⁸ Apple applied for a patent that enables the measurement of heart rate, body temperature and oxygen content in the blood using in-ear headphones⁹⁹, potentially enabling the receiver of this data to analyze a consumer’s health and current mood. As a third example, IBM applied for a patent regarding a touch-sensitive flooring that could ideally measure and visualize streams of visitors in supermarkets and their lengths of stay at certain point of sales (POS).¹⁰⁰

Internal data sources, on the other hand, comprise any data sources within a given company such as customer data from the marketing department or financial reports from the controlling department. Internal data can generate a significant competitive advantage, because insights generated from this data are exclusive and not available for competitors.¹⁰¹ Companies these days already generate large amounts of data internally, which originate from systems and software tools such as Enterprise-Resource-Planning (ERP), CRM, or web tracking. The correspondence by e-mail, letter, and fax as well as protocol and measurement data that is generated during the production of goods or during the provision of services to machines and sensors can be internal sources and contain useful information as well.¹⁰²

The ERP-System represents one of the most significant internal BD sources.¹⁰³ It supports all business processes in a company and contains modules for areas such as procurement, production, sales, asset management, human resources, finance and accounting which are connected via a common database. In other words, the company-wide consolidation of data makes it possible to access the data in a centralized system. Thus, ERP supports planning across all levels of the company, from the group level through various plants, divisions and departments to individual storage locations.¹⁰⁴ When successfully implemented, ERP has the potential to drastically reduce inventory efforts and working capital as well as to gather information about customer needs. Furthermore, a functional ERP enables a company to manage and view the extended enterprise of alliances, suppliers, and customers as an integrated whole.¹⁰⁵

Another major internal BD source is CRM. Within the CRM-Systems all individual customer activities, e.g., shopping behavior, and personal data are stored.¹⁰⁶ Electronical bonus card programs like Payback or Happy Digits serve as example here. They record the frequency, volume, and composition of customer purchases. This data then allows to reveal, e.g., which products are bought by whom as well as which products are bought as a bundle or according to seasonality.¹⁰⁷ CRM therefore increases efficiency and effectiveness resulting, e.g., from a simplification of daily administrative work through process optimization, systematic data integration and distribution as well as a fast and targeted analysis of the respective data.¹⁰⁸

Finally, Web Analytics is another BD source that leads to knowledge about customer needs and consumer opinions and thus helps to recognize new business opportunities.¹⁰⁹ It comprises the measurement, collection, analysis, and evaluation of internet-based data to understand and optimize usage behavior on the internet.¹¹⁰ In other words, you can measure page views, ad impressions, clickthrough rates, use time, new visitors, conversion rates, and other key parameters and use this data to evaluate the performance of your website as well as the users’ behavior on your website. Valuable insights like how users found your website, at which stage users canceled the purchasing process or the overall reach of your website can be extracted from those values.¹¹¹ Table 1 in Appendix B presents a sample of the most common ratios.

Coming from channel management, CRM-systems and web analytics are the most significant data sources for expert 1. According to him, these sources reveal extensive, most precise and detailed information regarding, e.g., customer behavior and conversion rates, which enables companies to evaluate the success of online campaigns. If companies do not use their websites as direct distribution channels and only cooperate with retailers such as MediaMarkt or Amazon instead, the precise measurement of online key performance indicators (KPIs) is strongly hampered since retailers greatly restrict access to their onside data. An interesting recent invention by DEXI.IO applies bots that scan other websites and can reveal the performance of a company’s product on the respective websites. By this means, important information becomes available, e.g., how often a given product appears as first result when a consumer searches for “washing machine”. However, proprietors of such websites start to strike back with short questions to check whether the visitor represents a consumer or bot. ¹¹²

2.2.3 Demands on technology

2.2.3.1 Data warehouse and business intelligence

As data becomes more complex, larger, and more incomprehensible, the limited mind and mental capacities of humans hamper the interpretation and analysis of highly complex data. Only information technology helps and allows us to record plentiful and rich data on consumer behavior in real-time.¹¹³ Therefore, this subchapter serves to investigate the technological prerequisites required to integrate BD into an organization.

Business intelligence (BI) has received expanding consideration in management, science, and business since the late 1980s.¹¹⁴ The systematical analysis of data through BI is typically done within a so-called data warehouse (DW) which builds the basis for optimized strategic and operative decisions.¹¹⁵

Abbildung in dieser Leseprobe nicht enthalten

Figure 5 Traditional business intelligence infrastructure¹¹⁶

A DW acquires or extracts its data from external open sources as well as from operational internal databases (Figure 5), e.g., ERP-systems and CRM-systems, providing a comprehensive data pool that includes current and/or historical data.¹¹⁷ This data acquisition is undertaken by an Extract-Transform-Load-Process (ETL-Process), which extracts the data from its original sources, transforms it into the target data format and finally loads transformed data into the target database. Subsequently, pre-defined reports are generated regularly and automatically from this data stock or selected parts of it, the so-called ”data marts”¹¹⁸. The modification or extension of a report can have a significant impact on the entire process and therefore typically takes take a long time.¹¹⁹

However, BI is facing new challenges due to a substantial development of BD and changes in BD technologies.¹²⁰ The data within a classical DW has been characterized by its clear structure which made the acquisition and analysis a relatively simple process. BD of today, in contrast, is characterized by its big volume, big variety, big velocity, and big veracity (see chapter 2.2.1) which make the process much more difficult and costly.¹²¹ Here, BDA, one of the “10 Bigs”, represents a significant factor to enhance the analytical capacities of BI and to respond to the increased technological demands, and thus, will be closely analyzed in the next sub-chapter.¹²²

2.2.3.2 Big Data Analytics

As elaborated in 2.2.1 and the previous chapter, BDA plays a central role when companies want to extract valuable insights from BD, which makes BDA an indispensable factor of the respective process. Iqbal et al. (2016) defined the process of BDA: “[...] refers to the techniques utilized to examine and process BD so that hidden underlying patterns are revealed, relationships are identified, and other insights concerning the application context under investigation are exposed”¹²³.

Abbildung in dieser Leseprobe nicht enthalten

Figure 6 BDA - Process.¹²⁴

BDA refers to the complete process of collecting, managing, processing, analyzing and visualizing continuously evolving data in terms of volume, velocity, value, variety and veracity (Figure 6).¹²⁵ In other words, BDA allows to model, analyze, and interpret BD.¹²⁶

Abbildung in dieser Leseprobe nicht enthalten

Figure 7 BDA – Ontology part 1.¹²⁷

Moreover, Figure 7 depicts the various parts of BDA. It includes, next to the already mentioned DW, data mining (DM), statistical modelling (SM), machine learning (ML), visualization, and optimization.¹²⁸ Ur Rehman et al. (2019) added text analytics (TA) and natural language processing (NLP) as additional components.¹²⁹ DM solutions enable companies to transform raw data into knowledge. Statistical tools help to collect, summarize, analyze, and interpret vast amounts of data, which lead to knowledge discovery. ML techniques enable industrial machines and devices to cross over into a self-learning mode without being explicitly programmed. TA helps to extract high-quality data from texts by searching and discovering patterns and trends using statistical pattern learning. Finally, NLP tools are used to extract and analyze unstructured data.¹³⁰ This also demonstrates that computer science and information technology play a dominant role in the further development of BDA by providing latest techniques and tools.¹³¹

Abbildung in dieser Leseprobe nicht enthalten

Figure 8 BDA – Ontology part 2¹³²

Figure 8 illustrates that BDA combines descriptive analytics, predictive analytics, prescriptive analytics¹³³ as well as real-time analytics.¹³⁴ Descriptive analytics generates insights into historical data, e.g., the number of defective items in the past and the reason for the defects. Predictive analytics helps revealing potential issues that can occur in the future by using SM and ML techniques, e.g., the prediction of equipment failure, expected inventory levels, and anticipated demand levels. In contrast, real-time analytics helps companies to become aware of current situations, e.g., a product’s current location and status or the detection of a faulty machine. Finally, prescriptive analytics advices and suggests the optimal action that an end user should undertake, e.g., by answering the question whether a machine is receiving the right raw materials in the correct amount.¹³⁵ Hence, BDA comprises the whole process of turning BD into valuable insights.¹³⁶ Figure 8 serves to summarize the above-mentioned characteristics of BDA and illustrates a simplified ontology, due to the depth of BDA. Please find a more complete taxonomy under Appendix C. There are various software tools available that support BI systems to cope with the new challenges of BD and the just described process of BDA. Please find a more detailed description of the most popular one, Apache Hadoop, under Appendix D.

3 State of the art regarding implementation of Big Data in marketing

3.1 State of the art marketing research

This chapter serves to analyze in detail how BD is employed to supplement, optimize, and transform traditional systems into strategic and operative state-of-the-art marketing activities.

Especially the subchapters 2.2.2 and 2.2.3 revealed that marketing researchers have access to a rapidly increasing volume of data. It was described that this data can be structured in different ways and originate from numerous different sources which necessitates new analyzing techniques and often results in new and valuable insights. Thus, it seems evident that BD immensely impacts marketing research.

Meffert et al. (2012) define marketing research as the collection, evaluation, and interpretation of information about the company’s current and future marketing situation and decisions. In other words, marketing research is crucial for a company to gain valuable insights into the market, products, and customers¹³⁷, which subsequently serve as basis for strategic decision-making.¹³⁸ Figure 9 illustrates the traditional marketing research process.

Abbildung in dieser Leseprobe nicht enthalten

Figure 9 Traditional marketing research process¹³⁹

Here, a clear formulation of the problem which results in specific research objectives is fundamental in the beginning.¹⁴⁰ With BD, however, the scientific method can be reversed¹⁴¹ from fitting data to preconceived theories of consumers, products, and competition, to using the collected data to frame new theories.¹⁴² BDA, i.e., methodological and technological advances, allow researchers to identify patterns in BD without establishing any hypotheses in advance, and thus, less existing knowledge is required and the focus shifts towards the unknown.¹⁴³ The approach of focusing not only on knowledge but also allowing some ignorance generates a certain degree of freedom and latitude which stimulates creativity within an organization. Founding marketing decision exclusively on perfect knowledge can represent an obstacle for creative activities. Since the source of competitive advantage shifts from knowledge itself to the velocity of generating creative ideas, and ignorance tends to become a significant driver of creativity within an organization, applying ignorance to a certain degree can be seen as crucial when striving for competitive advantage.¹⁴⁴ Walmart, for example, used BDA to reveal a steep incline of 700% in average in Pop-Tarts (pastry of US-American origin) sales shortly before a hurricane. With such scientific inquiry, retailers can increase their stock and place the products strategically in the store to best leverage the anticipated demand.¹⁴⁵ Hence, BD and BDA can allow researchers to skip the first phase of the traditional market research process, if the company decides to search for hidden consumer insights by analyzing patterns and the “unknown” (Figure 10).

Abbildung in dieser Leseprobe nicht enthalten

Figure 10 Search for the unknown and analysis of patterns¹⁴⁶

The elimination of phase one also helps optimizing the marketing research process in terms of speed. No time is needed anymore for the definition of the problem. Additionally, with BDA minimizing response time latency, i.e., the reduction of the intervals between data collection, analysis, interpretation, and action,¹⁴⁷ companies can analyze events, make decisions, and implement these decisions with real-time market responsiveness.¹⁴⁸ For instance, companies selling smart medical devices could employ heart attack incidences and live patient data to predict in real-time, when a patient is about to have a heart attack and interact instantly with market shareholders such as the next emergency service, the patient’s doctor, or the patient him/herself.¹⁴⁹ Pattern analysis can therefore be utilized to reveal not only unknown correlations but also be harnessed for a better prediction of consumer behavior.¹⁵⁰ Please find under Appendix E business prime examples for predictive marketing and what is needed to ideally exploit pattern and correlation analysis.

[...]

---

¹ Cf. Reinsel, D./ Gantz, J./ Rydning, J. (2018), p. 3.

² Cf. World Economic Forum (w/o Y).

³ Cf. Reinsel, D./ Gantz, J./ Rydning, J. (2018), p. 7.

⁴ Cf. Press, G. (2018).

⁵ Cf. Murphy A. et al. (2020).

⁶ Cf. Holland, H. (2020), p. 1.

⁷ Cf. De Luca, L. M. et al. (2020), p. 1.

⁸ Cf. Toedt, M. (2016), p. 9.

⁹ Cf. De Luca, L. M. et al. (2020), p. 1.

¹⁰ Cf. Sorescu, A. (2017), p. 692ff.

¹¹ Cf. Davenport, T, H./ Bean, R. (2019), p. 1ff.

¹² Cf. Lambrecht, A./ Tucker, C, E. (2015), p. 3.

¹³ Cf. Erevelles, S./ Fukawa, N. / Swayne, L. (2016), p. 897.

¹⁴ Cf. Davenport, T, H./ Bean, R. (2019), p. 7.

¹⁵ Cf. De Luca, L. M. et al. (2020), p. 2.

¹⁶ Cf. for the first half-sentence Toedt, M. (2016), p. 9.

¹⁷ Cf. for the last half-sentence Davenport, T, H./ Bean, R. (2019), p. 7.

¹⁸ Cf. MLex Market Insight (2019).

¹⁹ Cf. European Commission (2020), p. 1ff.

²⁰ Cf. European Union (2016), p. 1ff.

²¹ Cf. European Commission (2018).

²² Cf. Toedt, M. (2016), p. 15.

²³ Cf. PwC (2017), p. 2.

²⁴ Cf. Holland, H. (2020), p. 20.

²⁵ Cf. for the last two sentences Schwarz, T. (2016), p. 44.

²⁶ Cf. Fletcher, K./ Wheeler, C./ Wright, J. (1991), p. 121.

²⁷ Cf. Stone, M./ Shaw, R. (1987), p. 12f.

²⁸ Cf. Jackson, R./ Wang, P. (1997), p. 6f and Seller, M./ Gray, P. (1999), p. 10.

²⁹ Cf. Seller, M./ Gray, P. (1999), p. 10.

³⁰ Stone, M./ Shaw, R. (1987), p. 13.

³¹ Cf. Holland, H. (1993), p. 66f.

³² Cf. Schwarz, T. (2016), p. 44.

³³ Cf. Kotler et al. (2016), p. 65.

³⁴ Cf. Li, J. et al. (2018), p. 1ff.

³⁵ Cf. Dufft, N. (2015), p. 82 und Arthur, L. (2013), p. ix.

³⁶ Cf. Nitsche, M./ Gründig, Christian (2015), p. 27.

³⁷ Cf. Appendix J, interview question 10.

³⁸ Cf. Reinsel, D./ Gantz, J./ Rydning, J. (2018), p. 4f.

³⁹ Cf. Kotler et al. (2016), p. 193.

⁴⁰ Cf. for the first half-sentence Vossen, G. (2015), p. 35.

⁴¹ Cf. for the second half-sentence Fan, S./ Lau, R. Y. K./ Zhao, J. L. (2015), p. 1.

⁴² Cf. Vossen, G. (2015), p. 35.

⁴³ BITKOM (2012), p. 7.

⁴⁴ BITKOM (2012), p. 21.

⁴⁵ Gartner (w/o Y).

⁴⁶ Cf. Laney, D. (2001), p. 71f.

⁴⁷ Cf. Ebner, K./ Buhnen, T./ Urbach, N. (2014), p. 3749; Lycett, M. (2013), p. 381, Jain, A. (2016); Oracle (2013), p. 4 and Holland, H. (2020), p. 2.

⁴⁸ Cf. Holland, H. (2020), p. 3.

⁴⁹ Cf. Sun, Z. (2018), p. 3ff.

⁵⁰ Reinsel, D./ Gantz, J./ Rydning, J. (2018), p. 6.

⁵¹ Cf. Domo (2020).

⁵² Domo (2020).

⁵³ Cf. for the last two sentences Holland, H. (2015), p. 17.

⁵⁴ Cf. Jabbar, A./ Akhtar, P./ Dani, S. (2019), p. 3.

⁵⁵ Cf. Bachem, C. (2015), p. 29f.

⁵⁶ Cf. Oracle (2013), p. 4.

⁵⁷ Cf. Spangenberg, N. et al. (2020), p. 262.

⁵⁸ Cf. for the last two sentences Sun, Z. (2018), p. 4.

⁵⁹ Cf. Khan, N. et al. (2018), p. 55.

⁶⁰ Cf. Khan, N. et al. (2018), p. 54.

⁶¹ Cf. Sun, Z./ Sun, L./ Strang, K. (2016), p. 163.

⁶² Cf. Sun, Z. (2019), p. 5.

⁶³ Cf. Sun, Z. (2019), p. 7.

⁶⁴ Cf. De Luca, L. M. et al. (2020), p. 2f.

⁶⁵ Cf. Lycett, M. (2013), p. 384.

⁶⁶ Cf. Khan, N. et al. (2018), p. 53.

⁶⁷ Own illustration based on Sun, Z. (2018), p. 4ff; Ebner, K./ Buhnen, T./ Urbach, N. (2014), p. 3749; Lycett, M. (2013), p. 381; Jain, A. (2016); Oracle (2013), p. 4; Holland, H. (2020), p. 2 and Laney, D. (2001), p. 71f.

⁶⁸ Cf. Amado, A. et al. (2017), p. 2.

⁶⁹ Cf. for the last three sentences Nielsen (2020) and Kuhlmann, C. (2004), p. 289.

⁷⁰ Own illustration based on Holland, H. (2020), p. 8ff and Prasad, R./ Rohokale, V. (2020), p. 125f.

⁷¹ Cf. SAS (w/o Y).

⁷² Cf. Prasad, R./ Rohokale, V. (2020), p. 125f.

⁷³ Cf. Taylor, M./ Reilly, D./ Wren, C. (2018), p. 5.

⁷⁴ Cf. De Luca, L. M. et al. (2020), p. 4f.

⁷⁵ Cf. Statista (2020a).

⁷⁶ Cf. Statista (2020b).

⁷⁷ Cf. Deloitte (2017), p. 5.

⁷⁸ Cf. Purswani, J. M. et al. (2019), p. 338.

⁷⁹ Cf. Deloitte (2017), p. 1ff.

⁸⁰ Cf. Wurmser, Y. (2019).

⁸¹ Cf. Molla, R. (2020).

⁸² Cf. McAfee, A./ Brynjolfsson, E. (2012), p. 5.

⁸³ Cf. Holland, H. (2020), p. 9.

⁸⁴ Cf. Walter, O. (2016), p. 216.

⁸⁵ Cf. Rohleder, B. (2018), p. 2.

⁸⁶ Cf. Bernecker, M. (2016), p. 189f.

⁸⁷ Cf. Tan et al. (2013), p. 66.

⁸⁸ Cf. eMarketer (2017).

⁸⁹ Cf. Moro, S./ Rita, P./ Vala, B. (2016), p. 1.

⁹⁰ Cf. Holland, H. (2020), p. 9.

⁹¹ Cf. Jimenez-Marquez, J. L. et al. (2019), p. 1.

⁹² Cf. Hargittai, E. (2018), p. 1f.

⁹³ Cf. Ray, L. (2019).

⁹⁴ Cf. Google (2012).

⁹⁵ Cf. Dummies (w/o Y).

⁹⁶ Cf. Clasen, N. (2015), p. 252f.

⁹⁷ Cf. Holland, H. (2020), p. 9.

⁹⁸ Cf. Holland, H. (2020), p. 10f.

⁹⁹ Cf. for the first half-sentence Hughes, N. (2009).

¹⁰⁰ Cf. for the last two sentences Holland, H. (2020), p. 11.

¹⁰¹ Cf. Kotler et al. (2016), p. 196.

¹⁰² Cf. Holland, H. (2020), p. 8f.

¹⁰³ Cf. Ur Rehman, M. H. et al. (2019), p. 251f.

¹⁰⁴ Cf. Vahrenkamp, R. (2018).

¹⁰⁵ Cf. Chen, J. (2001), p. 374.

¹⁰⁶ Cf. Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 61.

¹⁰⁷ Cf. Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 865.

¹⁰⁸ Cf. Holland, H. (2018).

¹⁰⁹ Cf. Chen, H./ Chiang, R. H. L./ Storey, V.C. (2012), p. 1185.

¹¹⁰ Cf. Hassler (2010), p. 28.

¹¹¹ Cf. Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 171f.

¹¹² Cf. Appendix J, interview question 1.

¹¹³ Cf. Erevelles, S./ Fukawa, N. / Swayne, L. (2016), p. 897.

¹¹⁴ Cf. Sun, Z./ Sun, L./ Strang, K. (2016), p. 163.

¹¹⁵ Cf. Holland, H. (2020), p. 7.

¹¹⁶ Own illustration based on Müller, S. (2016), p. 141.

¹¹⁷ Cf. Coronel, C./ Morris, S. (2015), p. 590.

¹¹⁸ “A data mart is a subject-oriented data repository, similar in structure to the enterprise data warehouse, but holding the data required for the decision support and BI needs of a specific department or group within the organization.” - p. Loshin, D. (2012), p. 340.

¹¹⁹ Cf. Holland, H. (2020), p. 7.

¹²⁰ Cf. Sun, Z./ Sun, L./ Strang, K. (2016), p. 162.

¹²¹ Cf. Müller, S. (2016), p. 145.

¹²² Cf. Fan, S./ Lau, R. Y. K./ Zhao, J. L. (2015), p. 1ff.

¹²³ Cf. Iqbal, R. et al. (2016), p. 1.

¹²⁴ Own illustration based on Müller, S. (2016), p. 141ff and De Luca, L.M. et al. (2020), p. 2.

¹²⁵ Cf. Ur Rehman, M. H. et al. (2019), p. 248.

¹²⁶ Cf De Luca, L.M. et al. (2020), p. 2.

¹²⁷ Own illustration based on Ur Rehman, M.H. et al. (2019), p. 252 and Sun, Z./ Sun, L./ Strang, K. (2016), p. 163.

¹²⁸ Cf. Sun, Z./ Sun, L./ Strang, K. (2016), p. 163.

¹²⁹ Cf. Ur Rehman, M.H. et al. (2019), p. 252.

¹³⁰ Cf. Ur Rehman, M. H. et al. (2019), p. 252.

¹³¹ Cf. Sun, Z./ Zou, H./ Strang, K. (2015), p. 203.

¹³² Own illustration based on Ur Rehman, M. H. et al. (2019), p. 252 and Sun, Z./ Sun, L./ Strang, K. (2016), p. 163.

¹³³ Cf. for the first half-sentence Sun, Z./ Sun, L./ Strang, K. (2016), p. 163.

¹³⁴ Cf. Ur Rehman, M. H. et al. (2019), p. 252.

¹³⁵ Cf. Ur Rehman, M. H. et al. (2019), p. 253.

¹³⁶ Cf. De Luca, L. M. et al. (2020), p. 2.

¹³⁷ Cf. for the first half-sentence Müller, S. (2015), p. 39.

¹³⁸ Cf. for the second half-sentence Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 95.

¹³⁹ Own illustration based on Cf. Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 100.

¹⁴⁰ Cf. Cf. Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 100.

¹⁴¹ Cf. for the first half sentence Erevelles, S./ Fukawa, N./ Swayne, L. (2016), p. 897.

¹⁴² Cf. Firestein, S. (2012), p. 15ff.

¹⁴³ Cf. Lycett, M. (2013), p. 383.

¹⁴⁴ Cf. for the last two sentences Erevelles, S./ Fukawa, N./ Swayne, L. (2016), p. 899.

¹⁴⁵ Cf. Germann et al. (2014), p. 588.

¹⁴⁶ Own illustration based on Erevelles, S./ Fukawa, N./ Swayne, L. (2016), p. 899 and Meffert, H./ Burmann, C./ Kirchgeorg, M. (2012), p. 100.

¹⁴⁷ Cf. for the first half-sentence Pigni, F./ Piccoli, G./ Watson, R. (2016), p. 7f.

¹⁴⁸ Cf. for the second half-sentence De Luca et al. (2020), p. 4.

¹⁴⁹ Cf. Marinova, D. (2017), p. 31.

¹⁵⁰ Cf. Tong, S./ Luo, X./ Xu, B. (2019), p. 74.