Practical application of artificial intelligence in TV market and TV audience research. IT and digital marketing

Essai , 2020 , 32 Pages , Note: 60

Gestion d'entreprise - Étude de marché

This paper is about the practical application of artificial intelligence in TV market and TV audience research.

Digital transformation has a colossal impact on the media industry, the World Wide Web is rapidly changing the media landscape and introducing new game rules. Traditional media are losing their audience, their weight in the eyes of customers and consequently, the advertising revenues. Media companies other than from the internet sector need to counteract and adapt, especially, as far as the audience measurement is concerned.

This project reviews the necessary changes in audience management that are supported by Artificial Intelligence technology. Specific solutions offered in the Russian media market have served as an example for possible development directions. It was concluded that new products such as multi-source evaluation, the transition from a limited sample to the whole universe measurement are required to be put into effect in order to deliver holistic and consumer-focused data by research companies.

This tremendous change in the mindset of the researchers will enable usage of a single metric like the Reach across all media by advertising managers. Moreover, the dynamic pricing at TV advertising will become possible too. Given the magnitude of the data to be processed only AI is capable of offering economically reasonable industrial solutions.

Abstract

Digital transformation has a colossal impact on the media industry, the World Wide Web is rapidly changing the media landscape and introducing new game rules. Traditional media are losing their audience, their weight in the eyes of customers and consequently, the advertising revenues. Media companies other than from the internet sector need to counteract and adapt, especially, as far as the audience measurement is concerned. This project reviews the necessary changes in audience management that are supported by Artificial Intelligence technology. Specific solutions offered in the Russian media market have served as an example for possible development directions. It was concluded that new products such as multi-source evaluation, the transition from a limited sample to the whole universe measurement are required to be put into effect in order to deliver holistic and consumer-focused data by research companies. This tremendous change in the mindset of the researchers will enable usage of a single metric like the Reach across all media by advertising managers. Moreover, the dynamic pricing at TV advertising will become possible too. Given the magnitude of the data to be processed only AI is capable of offering economically reasonable industrial solutions.

Abbildung in dieser Leseprobe nicht enthalten

The pictures have been removed for copyright reasons

1.Introduction

Informational revolution, the emergence of new data processing and communicational technologies pose crucial challenges for mass media, including traditional television. TV that has been the primary advertising vehicle for decades started speedily losing the audience, and consequently, attractiveness for advertisers, revenues, and market share in favour of digital services.

TV content production and distribution have fallen victim of the internet. TV value creation chains are exposed to threats from different sides. The high-speed internet penetrated almost every household and mobile devices which performance can exceed that of standard PC’s enables the delivery of HD video content for individual consumption practically in every corner of the world. Concurrently, smartphones cameras are capable of shooting HD videos, and users can upload their self-generated content in clouds accessible to everyone that might sometimes compete in terms of quality with the content created by professionals. In contrast to addressing TV audiences nationwide or regionally in mass, web communication allows to convey personalized content and engage users in information exchange, i.e. retrieve immediate feedback. New interactive tools enable a supply of online video inventory with a guaranteed contact frequency and dynamic pricing and poach customary TV advertisers. Hence, when run as commercial enterprises TV stations and networks need to keep pace with innovations and respond to commercial pressure from the internet sector in a competition for the audience. They need to refurbish themselves, and several visionaries dream even of convergence with internet platforms.

Contemporary media economics teaches that media produces a twofold commodity: content and audience (Andrew, 2019). The content, the first commodity, is sold or delivered “for free” to users-subscribers, and used to attract audiences that can be marketed to those who are interested in reaching out to the audience with different types of messages either commercial or political. Advertisers, political associations and entities eventually pay mass media suppliers for their audience, the second or seemingly “auxiliary” commodity, when placing advertising orders with them. A commodity can have a value if it has a measure for exchange, and the classic media industry has developed longitudinal sample surveys to conclude on media consumption of mass just based on registering and observing viewing, listening or reading of a few. TV ratings measured by research companies have become a single “currency” of advertising trade for decades.

All setbacks, experienced by the TV industry now, are spinning around audience research and TV viewing measurement. While TV keeps up supplying contacts with mass audiences, internet inventory suppliers are capable of identifying users and tracing their preferences by deciphering audience footprints (cookies, tracking pixels and URLs). They installed counters to verify their contacts with individuals making possible sales and delivery of customized marketing messages. They enable advertisers to decide whom, when, where and with which marketing message to approach using bidding services and machine learning in real-time. Hence, TV measurement occupies a pivotal point in the traditional value chain.

The first question we are going to answer in this inquiry can be formulated as follows - how audience measurement is going to change in the nearest future and how the value chain is transforming.

TV industry decision-makers have realized the urge to change belatedly as the TV share in global advertising spendings has declined by a quarter from 42% to 32% just during 5 last years, and in Russia - from 49% to 33% respectively (Guttmann, 2019).

Distribution of Global Ad spendings by medium 2015 - 2020

Abbildung in dieser Leseprobe nicht enthalten

Figure 2

Source: https://www.statista.com/statistics/245440/distributuion-of-global-advertising-expenditure-by-media/statista.com

Initially, confronted with new distribution options, TV channels and producers took advantage of OTT (Over-The-Top), YouTube making possible time-shift viewings of their linear programs etc. They attempted to find more reliable and cost-effective ways to account for incremental audiences emerging in a new digital environment using their standard measurement toolkit. But, it turned out to be inappropriate and outdated (Fulgoni,2017). Now, the continued migration from traditional TV consumption to digital platforms put the squeeze on the industry to rethink measurement approaches. As a number of users’ devices inclusive TV sets, set-top-boxes etc. connected to the web are rapidly growing, plenty of viewing data can be collected from multi-system operators (MSO) distributing linear and non-linear video content of TV channels. A mass of unstructured raw data is collected, stored and handled by points of sales, multiple system telecommunication operators, hospitals, banks and pension funds, credit card institutes, tax offices, travel and forwarding agents, utilities, social networks etc. The internet platforms are combining their tracking data with metadata, other structured and unstructured data for creating individual user profiles. Sampling is no longer required to represent the whole audience as they can reach out to the focused audience. These datasets can be combined with those of TV viewing to benefit the TV industry.

Thus, Big Data if correctly processed and evaluated, gives a chance to transfer the practice of automated and programmatic buying and selling applied by internet platforms on television-advertising inventory. Oppose to the previous research standard using a prolonged panel measurement, i.e. a single-source observation, researches will have to deal with a wide variety of sources that they need to refine samples from audience duplications and to find computational and analytical tools to pool the data from different sources.

The second issue that we are dealing with in this project is what part Artificial Intelligence can play to support the audience measurement enabling programmatic sales of the TV audience. Audience researchers will take on the issue properly if leaning on the experience of internet giants, and they avail themselves of the advantages that Artificial Intelligence offers for accomplishing analytical tasks. This paper attempts to give an account on AI adoption and impact on the TV advertising market and audience research.

To start with, it is worth making a relevant reservation as to possible general conclusions on the inquiry subject. Both the TV market and audience research market carry sharply delineated, national and local features that make any generalization contestable. In contrast to the global nature of the internet, the television industry has evolved originally as a mass media product made exclusively for home consumption. Both bargaining power of stakeholders (content producers, TV stations, broadcast platforms, research companies, advertising agencies etc.) and historical peculiarities of market genesis and social importance of news coverage and narration for political ends determine the structure of national markets.

Given this circumstance, we will focus on reviewing the implementation of AI techniques in the Russian market using a specific business case of the media research company Mediahills Ltd. We have chosen the Russian market for review as Russian media industry was commercialized just early in 90ie, even though Russians had been those who invented TV in 30ies. Due to a late commercial start they could jump over several development stages and gain the benefits of modern technologies sparing plenty of efforts.

Despite the mentioned reservation, we will attempt to draw several general conclusions on the TV and audience research business models and their value chain.

2.Methodology applied

While working on this project we used the research approach of systematic qualitative analysis, including observation, textual or visual analysis (for example, from books or videos). The main tool was a qualitative in-depth analysis of sources indicated in the references list. We attempted to get insights from the collected data, decipher market reality, stakeholders' behaviour and their representations in the academic and professional literature. The qualitative analysis allows for the discovery of broad patterns and themes from complex business activities, interpretations of various aspects of the research topics, and enables granular research analysis subsequently (Boyatzis, 1998).

Research in AI’ application in the media industry, and, especially in media research is a relatively new area. Unsurprisingly, there are few empirical investigations and theoretical analytic works. We conducted a database search for literature using keywords such AI, digital marketing, programmatic trading and audience research during the data collection stage. About 100 handbooks, reports, essays, news releases, articles, industry whitepapers were reviewed to identify applications of AI techniques in the media sector and, specifically, in the audience measurement. In the collected data we searched for the manifestations of the three basic rules of dialectical cognition when we dealt with thematic analysis. An iterative, inductive process of interpretive analysis within the context of practical IT applications was adopted to conclude from the specific on the universal that reflect common, recurring patterns. To complement the mentioned, induction and deduction allowed us to make hypotheses about answers to research questions.

On top of that, we took advantage of a case study used for explaining how an organization or business process works. Fortunately, I had a chance to conduct a couple of discussions with executives of Russian media companies while being an intern at Mediahills that was the subject of the case study. The executives were engaged in the digital business of TV broadcasters, multiplatform operators, and advertising agencies but they asked not disclose their personalities. The insights gained during the internship and knowledge shared by skilled media personnel were incredibly helpful and meaningful to elaborate on the grounded theory.

To substantiate my conclusions, I used the results of an empirical research. I conducted a short qualitative poll. A questionnaire for the quantitative survey contained specific items relating to AI applications in advertising and media research. Indeed, we were not able to achieve the scientifically required magnitudes to make the findings fall within the fidelity interval. Although the results are not representative, and thus, universally valid, they can serve as a basis for our conclusion within reasonable limits.

The findings below are structured to reflect the logic of my thoughts on how the research questions can be answered.

3.Background and Literature review

The subject-matter of the present research is articulated as analysis of current and anticipated revolutionary changes in the TV audience measurement, their impact on the television value chain in the digital age and the prospects for supporting these changes by artificial intelligence. The area of research lies at the intersection of various disciplines: marketing, media analytics, sociology, micro- and macroeconomics, business and organisation studies, strategic management and information technology. However, since the research questions are narrowly delineated in terms of business strategy options for media companies and their practical implementation, it is not surprising that there is pretty little literature on this particular topic. Nevertheless, relevant publications can be found in the literature in different research domains like digital marketing, programmatic trading, media research, adoption of artificial intelligence in specific business cases. Although academics offer several comprehensive handbooks and specialised textbooks (Russel, 2018, Russel, 2019, Shapiro 1987, Warwick, 2012, Winston, 1995, Kotler 2019 etc.), most of the publications take the form of research papers accessible in recurrent releases of scientific or trade news journals, where scholars and businessmen share results of their research or observations. These texts served as a valuable secondary source of data and media research. Wilkinson and Merle (2013) ascertain that ”the accelerated pace of changes in twenty-first-century media necessitates academic researchers’ ready access to up-to-date information about industry, content and technology”. The secondary sources used for drafting this paper includes global and local websites of the media and advertising industry, press releases, corporate handbooks and manuals (see the bibliography). Useful pieces of knowledge are scattered across many unrelated sources where it can’t be mined without significant effort and synthetic analysis.

Consequently, at first we need to take a short look at AI, digital marketing and TV media research as separate subjects before we can draw synthetic conclusions on the topic of our interest.

3.1. Relevance of Concept of Artificial Intelligence for Media Industry

The term ‘Artificial Intelligence” was coined at Dartmouth college conference held in 1956 and sponsored by DARPA (Defense Advanced Research Projects Agency) (Winston, 1995). Since the Dartmouth conference AI, proponents have not reached a consensus yet about what AI is and how it is to be defined. A more comprehensive picture of understanding AI is given by Russel and Norvig (2018) when they tend to describe AI in terms of the pursued goal. "AI is aimed at…. ( see the table that can be depicted) ( Shapiro, 1987).

Abbildung in dieser Leseprobe nicht enthalten

Figure 3. Source: Shapiro, S. C. (1987). Encyclopedia of artificial intelligence. Wiley

AI concept unfolds as the most efficient form of the search for problem solutions from the space of all possible solutions contrasting to the classical approach when a detailed algorithm is prescribed, and all inputs are established. According to Russel (2019), AI makes itself out when programs that are designed calculatingly rational start constructing themselves if executed infinitely fast, and this would result in perfectly rational behaviour. Russel (2019) describes AI as the field devoted to creating optimal programs for intelligent agents under time and space constraints on the machines implementing these programs.

Sridhar (2018) denotes the following differences between traditional and AI programming.

Abbildung in dieser Leseprobe nicht enthalten

Figure 4 Source: Sridhar, S., (2018). ARTIFICIAL INTELLIGENCE (WAY TO FUTURE), London

Hence, there is no unifying definition of AI, but "the main unifying theme is the idea of an intelligent agent. We define AI as the study of agents that receive precepts from the environment and perform actions'' (Russel, 2018). In his lectures at MIT, Winston limits the definition just with “ algorithms enabled by constraints imposed by representations that support thinking, reception and actions'' (YouTube). Indeed, Searle's Chinese Chamber experiment proves that running the right program does not necessarily generate understanding (Preston, 2007). Thus, at the time being we can conclude that AI is not about machines or high-end computers capable of perceiving, thinking and acting but just all about programming with self-learning algorithms.

Despite these limitations, 92% of surveyed business leaders said AI is vital to their business processes, and 76% are sure that cognitive technologies will transform their businesses (Deloitte, 2020). 80% of the media practitioners agreed that AI would have a significant impact on the value chain of their industry (Shields, 2018). Still, the benefits of AI applications might be controversial and overestimated due to a specific "hype noise." Almost half (47 per cent) of marketers consider Al to be overhyped, far more than other industry buzzwords, while 43 per cent of marketers believe vendors overpromise and underdeliver when it comes to AI" ( Nicalls, 2018).

3.2.General vision of Digital marketing for TV Industry

When assessing the effects of AI and digitalisation on marketing, we refer to Phillip Kotler who elaborated Marketing 4.0 approach. This concept takes into account the convergence of the offline and online worlds of businesses and customers (Kotler, 2018). The latter is characterised by a fusion of technologies that is blurring the lines between the physical, digital, and biological spheres (Schwab, 2016). As connected devices become more commonplace Kotler concludes on transformation of 4 P’s into 4C’s based on the adoption of the Big Data analysis and AI techniques for creating unique selling propositions, specifically, customised products (Co-creation), dynamic pricing (Currency), peer-to-peer distribution (Communal activation), customer-rating systems (Conversation) (Kotler, 2018). Applying Kotler’s Marketing 4.0 C’s principles Gentsch (2019), Sterne (2017), Mather (2018) determined the following major domains for AI adoption that intersect with scope of our interest in the present essay

a) Extended market and customized audience research
b) Personalised targeted advertising
c) Programmatic sales and Real-Time-Bidding advertising.

They exemplified their research with practical cases of Amazon, Bosch/Siemens, UPS, Otto Group, Netflix, Creditreform etc.. In this paper we will analyze how this model works in modern Russia.

[...]

• Cours: Business management
• Note: 60
• Auteur: Igor Kolesnichenko (Auteur)
• Année de publication: 2020
• Pages: 32
• N° de catalogue: V1147042
• ISBN (PDF): 9783346552310
• Langue: anglais
• mots-clé: #IT #Advertising #TV #Audience# mesurement