Skip to main content
main-content
Top

Hint

Swipe to navigate through the chapters of this book

2022 | OriginalPaper | Chapter

2. Algorithmic Marketing

share
SHARE

Abstract

This chapter provides an overview of the “algorithmic business” concept. The latter is a socio-technical notion describing the use of intelligent computational processes by business entities which allow making assessments, predictions and decisions about consumers. The phenomenon represents the last stage of the evolution of marketing technologies caused mainly by the increase in consumer data and powerful new applications based on machine learning and cognitive computing to analyse such data and make automated decisions in real-time. The chapter includes an analysis of the historical development, leading technologies, current applications, and new organisational models.
Footnotes
1
For example, Privacy Choice (2010) explain that practitioners call AI simple learning algorithms trivializing the original project of building “artificial intelligence.” On the problem of using “AI” as a label, see also McCorduck (2004).
 
2
Kurzweil et al. (1990).
 
3
Turing (1950).
 
4
An overview on how AI technologies are applied in marketing processes can be found in Gentsch (2018).
 
5
I follow the outlook of Alter’s (2019) research, explaining current business practices by applying socio-technical analysis.
 
6
For the history of AI technologies, two of the leading manuals have been used: Russell and Norvig (2010) and Nilsson (2009).
 
7
This idea was expressed by Cristianini (2021), drawing on Thomas Kuhn’s notion of a paradigm shift to interpret current developments in intelligent algorithmic systems and their impact on social organisations as the product of significant changes in the theoretical underpinnings of AI research.
 
8
Cristianini (2021).
 
9
Schutzer (1990).
 
10
Wierenga (1990).
 
11
Burke et al. (1990).
 
12
McCann et al. (1990).
 
13
For a detailed overview of expert systems in marketing, see, e.g., Wierenga et al. (2000).
 
14
As noted by Wierenga, this is mainly due to an underdeveloped discussion on behavioural approaches to marketers’ decision-making and metrics to evaluate the system’s performance (Wierenga et al. 2000).
 
15
Stevenson et al. (1990) observe that the main drawback of developing a marketing expert system is money and time, with an estimated cost of $15 million, and a period of three to five years to produce results.
 
16
Wagner (2017).
 
17
As noted by Paul Smolensky, the approach that neural networks take in working toward computer intelligence is completely different from the approach taken in symbolic AI. He used the term “sub-symbolic AI” to express the idea of an intellectual activity performing at an intermediate level between the human brain’s symbolic level and the pure neural level. In sub-symbolic AI, knowledge (concepts and rules) is not explicitly represented by symbols but implicitly learned from data through operations expressed in mathematical operations. Unlike in symbolic AI, the sub-symbolic hypothesis holds that it is impossible to give a complete representation of mental processes at the level of symbols. Instead, it assumes that such processes should be represented beneath the conceptual level, that of neuron-like nodes and synapses-like edges. The intelligent behaviour of sub-symbolic computer systems cannot be broken down into single logical operations but is the inseparable product of the operations performed by the neuronal units within the network. Smolensky (1988), p. 12.
 
18
On the structure of neural networks, see Sect. 2.3.3.
 
19
Russell and Norvig (2010), p. 19. The main problem was linked to the fact that back in the 1950s, computers did not have enough computing power to allow neural networks to compute different input examples and learn effectively.
 
20
Plasek (2016).
 
21
Russell and Norvig (2010), p. 18.
 
22
Petrison et al. (1997).
 
23
Venugopal and Baets (1994).
 
24
Curry and Moutinho (1993).
 
25
Tkác and Verner (2016).
 
26
Russell and Norvig (2010), p. 36.
 
27
Berners-Lee and Cailliau (1990).
 
28
Guttman et al. (1998).
 
29
Jennings and Wooldridge (1996).
 
30
Alonso (2002).
 
31
Russell and Norvig (2010), pp. 46–58.
 
32
Nilsson (2014).
 
33
For a thorough analysis of software agents in e-commerce, see Maes (1999).
 
34
Russell and Norvig (2010), p. 27.
 
35
Yarowsky (1992).
 
36
Evidence of this shift lies in one of the many patents that Google registered at the time, described as “a computer-implemented method for determining user profile information for a user, the computer-implemented method comprising: determining, by a computer system including at least one computer on a network, initial user profile information for the user; obtaining, by the computer system, inferred user profile information for the user; determining, by the computer system, user profile information for the user using both the initial user profile information and the inferred user profile information; serving, by the computer system, an advertisement to the user using the user profile information.” See Bharat et al. (2005).
 
37
Cristianini (2016).
 
38
Markoff (2011).
 
39
Clark (2012).
 
40
TechCrunch (2014).
 
41
Smith and Linden (2017).
 
42
Team (2014).
 
43
O’Leary (2013) observes that “AI has been used in several different ways to facilitate capturing and structuring Big data, and it has been used to analyse Big data for key insights.” The author holds that the scale of today’s big data is likely to be little or small data in 5 to 10 years, and most probably, the terms will splinter, just like what happened with the concept of artificial intelligence. Different approaches and new subdomains will emerge.
 
44
For a review of different big data definitions, see De Mauro et al. (2015), p. 103. The authors provide the following definition of big data: “Big data represents the Information assets characterised by such a High Volume, Velocity and Variety to require specific Technology and Analytical Methods for its transformation into Value”.
 
45
IDC (2020).
 
46
Stackscale (2020).
 
47
CIO reports that 80–90% of the data that are generated today are unstructured and include photos, text, and videos. See (CIO 2019).
 
48
Vahn (2014).
 
49
For a general introduction to machine learning, see Mitchell (1997, p. 1–19). More recently, Domingos (2015) and Alpaydin (2020).
 
50
The problem of learning was indeed anticipated by the father of AI, Alan Turing. In his 1950 seminal paper, he proposed a thought experiment where human-like intelligence could be derived not from predefined knowledge installed in computers, but from the activity of learning from a teacher. One analogy he used is that of a child: “Instead of trying to produce a programme to simulate the adult mind, why not rather try to produce one which simulates the child’s? If this were then subjected to an appropriate course of education one would obtain the adult brain. Presumably the child brain is something like a notebook as one buys it from the stationer’s. Rather little mechanism, and lots of blank sheets.” Another analogy is that of learning from punishments and rewards: “We normally associate punishments and rewards with the teaching process. Some simple child machines can be constructed or programmed on this sort of principle. The machine has to be so constructed that events which shortly preceded the occurrence of a punishment signal are unlikely to be repeated, whereas a reward signal increased the probability of repetition of the events which led up to it.” See Turing (1950).
 
51
In some cases, labelled data can be found ready-made on the Web. Often, however, labelling requires human work, which can be low-skill (such as training the system to recognise images of cats and dogs) or high-skill (as in the case of training the system to recognise specialised texts). Sometimes the acquisition of labelled data can be facilitated with semi-supervised learning techniques. Semi-supervised learning falls halfway between supervised and unsupervised learning, where a small amount of labelled data is combined with a large amount of unlabelled data.
 
52
Alpaydin (2020), p. 143.
 
53
An introduction specific on reinforcement learning can be found in Sutton and Barto (1998).
 
54
Goodfellow et al. (2016).
 
55
Mitchell (1997), pp. 81–197.
 
56
In unsupervised learning, input data of similar types are combined in order to compose different clusters. When a new input pattern is applied, the neural network gives an output response indicating the class to which the input pattern belongs. There would be no feedback from the environment as to the desired output and whether it is it correct or incorrect. So here the network must discover the patterns, features from the input data, and the relation for the input data over the output. On a reinforcement learning approach, the network will adopt the behaviour that maximises its score (e.g., the reward points linked to gains in investments or victories in games).
 
57
Zhang et al. (2019).
 
58
Jurafsky (2000).
 
59
Halevy et al. (2009).
 
60
For example, through the technique called “tokenisation,” NLP based on machine learning based enables a computer to distinguish the different words in a sentence. The data scientist creates a training dataset with lots of textual data in which the different words are labelled to form “tokens.” With enough examples, the system can learn the rules of syntax by looking directly into the data and applying them to process new textual sources. Another example, which often follows tokenisation, is part-of-speech tagging (PoS), in which the teacher also determines the token’s part of speech (for example, whether a specific word is a noun, adverb, or adjective).
 
61
Hirschberg and Manning (2015).
 
62
Martinez and Walton (2014).
 
63
Gatt and Krahmer (2018).
 
64
Adamopoulou and Moussiades (2020).
 
65
Yu and Deng (2016).
 
66
For an introduction to computer vision, see Szeliski (2010).
 
67
Szeliski (2010), pp. 575–637. For example, supervised learning can be employed to classify different images according to different classes (image classification), while unsupervised learning can be used to classify images according to similarities (image clustering). Image-recognition techniques can be applied to recognise different entities: objects (object recognition), handwritten characters (optical character recognition), codes (reading 2D codes), and human faces (facial recognition).
 
68
Guo and Zhang (2019).
 
69
Brynjolfsson and Mcafee (2017).
 
70
Metelskaia et al. (2018).
 
71
A W3Techs survey shows that, to December 2020, Google Analytics is used by 54.6% of all world websites, that is a market share of 83.9% compared to its competitors (W3Techs 2020).
 
72
Mordor Intelligence (2020).
 
73
MarTech Series (2019).
 
Literature
go back to reference Adamopoulou E, Moussiades L (2020) An overview of chatbot technology. In: Maglogiannis I, Iliadis L, Pimenidis E (eds) Artificial intelligence applications and innovations. Springer International Publishing, pp 373–383 CrossRef Adamopoulou E, Moussiades L (2020) An overview of chatbot technology. In: Maglogiannis I, Iliadis L, Pimenidis E (eds) Artificial intelligence applications and innovations. Springer International Publishing, pp 373–383 CrossRef
go back to reference Alonso E (2002) AI and agents: state of the art. AI Magazine 23(3):25–25 Alonso E (2002) AI and agents: state of the art. AI Magazine 23(3):25–25
go back to reference Alpaydin E (2020) Introduction to machine learning. MIT Press Alpaydin E (2020) Introduction to machine learning. MIT Press
go back to reference Alter S (2019) Applying socio-technical thinking in the competitive, Agile, lean, data-driven world of knowledge work and smart, service-oriented, customer-centric value creation ecosystems. Complex Syst Inf Model Q 18:1–22 Alter S (2019) Applying socio-technical thinking in the competitive, Agile, lean, data-driven world of knowledge work and smart, service-oriented, customer-centric value creation ecosystems. Complex Syst Inf Model Q 18:1–22
go back to reference Bharat K, Lawrence S, Sahami M (2005) Generating user information for use in targeted advertising, issued June 2005 Bharat K, Lawrence S, Sahami M (2005) Generating user information for use in targeted advertising, issued June 2005
go back to reference Burke RR, Rangaswamy A, Wind J, Eliashberg J (1990) A knowledge-based system for advertising design. Mark Sci 9(3):212–229 CrossRef Burke RR, Rangaswamy A, Wind J, Eliashberg J (1990) A knowledge-based system for advertising design. Mark Sci 9(3):212–229 CrossRef
go back to reference Cristianini N (2021) Shortcuts to artificial intelligence. In: Pelillo M, Scantamburlo T (eds) Machines we trust machines we trust – perspectives on dependable AI. MIT Press Cristianini N (2021) Shortcuts to artificial intelligence. In: Pelillo M, Scantamburlo T (eds) Machines we trust machines we trust – perspectives on dependable AI. MIT Press
go back to reference Curry B, Moutinho L (1993) Neural networks in marketing: modelling consumer responses to advertising Stimuli. Eur J Mark 27(7):5–20 CrossRef Curry B, Moutinho L (1993) Neural networks in marketing: modelling consumer responses to advertising Stimuli. Eur J Mark 27(7):5–20 CrossRef
go back to reference De Mauro A, Greco M, Grimaldi M (2015) What is Big Data? A consensual definition and a review of key research topics. In: AIP Conference Proceedings, 1644. American Institute of Physics, pp 97–104 De Mauro A, Greco M, Grimaldi M (2015) What is Big Data? A consensual definition and a review of key research topics. In: AIP Conference Proceedings, 1644. American Institute of Physics, pp 97–104
go back to reference Domingos P (2015) The master algorithm: how the quest for the ultimate learning machine will remake our world. Basic Books Domingos P (2015) The master algorithm: how the quest for the ultimate learning machine will remake our world. Basic Books
go back to reference Gatt A, Krahmer E (2018) Survey of the state of the art in natural language generation: core tasks, applications and evaluation. J Artif Intell Res 61:65–170 CrossRef Gatt A, Krahmer E (2018) Survey of the state of the art in natural language generation: core tasks, applications and evaluation. J Artif Intell Res 61:65–170 CrossRef
go back to reference Gentsch P (2018) AI in marketing, sales and service: how marketers without a data science degree can use AI, big data and bots. Springer Gentsch P (2018) AI in marketing, sales and service: how marketers without a data science degree can use AI, big data and bots. Springer
go back to reference Goodfellow I, Bengio Y, Courville A, Bengio Y (2016) Deep learning. MIT Press, Cambridge Goodfellow I, Bengio Y, Courville A, Bengio Y (2016) Deep learning. MIT Press, Cambridge
go back to reference Guo G, Zhang N (2019) A survey on deep learning based face recognition. Comput Vision Image Understand 189:102805 CrossRef Guo G, Zhang N (2019) A survey on deep learning based face recognition. Comput Vision Image Understand 189:102805 CrossRef
go back to reference Guttman RH, Moukas AG, Maes P (1998) Agent-mediated electronic commerce: a survey. Knowl Eng Rev 13(2):147–159 CrossRef Guttman RH, Moukas AG, Maes P (1998) Agent-mediated electronic commerce: a survey. Knowl Eng Rev 13(2):147–159 CrossRef
go back to reference Halevy A, Norvig P, Pereira F (2009) The unreasonable effectiveness of data. IEEE Intell Syst 24(2):8–12 CrossRef Halevy A, Norvig P, Pereira F (2009) The unreasonable effectiveness of data. IEEE Intell Syst 24(2):8–12 CrossRef
go back to reference Hirschberg J, Manning CD (2015) Advances in natural language processing. Science 349(6245):261–266 CrossRef Hirschberg J, Manning CD (2015) Advances in natural language processing. Science 349(6245):261–266 CrossRef
go back to reference Jurafsky D (2000) Speech & language processing. Pearson Education Jurafsky D (2000) Speech & language processing. Pearson Education
go back to reference Kurzweil R, Richter R, Kurzweil R, Schneider ML (1990) The age of intelligent machines, vol 579. MIT Press, Cambridge Kurzweil R, Richter R, Kurzweil R, Schneider ML (1990) The age of intelligent machines, vol 579. MIT Press, Cambridge
go back to reference Maes P (1999) Smart commerce: the future of intelligent agents in cyberspace. J Interact Mark 13(3):66 CrossRef Maes P (1999) Smart commerce: the future of intelligent agents in cyberspace. J Interact Mark 13(3):66 CrossRef
go back to reference Martinez MG, Walton B (2014) The wisdom of crowds: the potential of online communities as a tool for data analysis. Technovation 34(4):203–214 CrossRef Martinez MG, Walton B (2014) The wisdom of crowds: the potential of online communities as a tool for data analysis. Technovation 34(4):203–214 CrossRef
go back to reference McCann J, Tadlaoui A, Gallagher J (1990) Knowledge systems in merchandising: advertising design. J Retail 66(3):257 McCann J, Tadlaoui A, Gallagher J (1990) Knowledge systems in merchandising: advertising design. J Retail 66(3):257
go back to reference McCorduck P (2004) Machines who think: a personal inquiry into the history and prospects of artificial intelligence. CRC Press CrossRef McCorduck P (2004) Machines who think: a personal inquiry into the history and prospects of artificial intelligence. CRC Press CrossRef
go back to reference Metelskaia I, Ignatyeva O, Denef S, Samsonowa T (2018) A business model template for AI solutions. In: Proceedings of the International Conference on Intelligent Science and Technology, pp 35–41 CrossRef Metelskaia I, Ignatyeva O, Denef S, Samsonowa T (2018) A business model template for AI solutions. In: Proceedings of the International Conference on Intelligent Science and Technology, pp 35–41 CrossRef
go back to reference Mitchell TM (1997) Machine learning. McGraw-Hill Science Mitchell TM (1997) Machine learning. McGraw-Hill Science
go back to reference Nilsson NJ (2009) The quest for artificial intelligence. Cambridge University Press CrossRef Nilsson NJ (2009) The quest for artificial intelligence. Cambridge University Press CrossRef
go back to reference Nilsson NJ (2014) Principles of artificial intelligence. Morgan Kaufmann Nilsson NJ (2014) Principles of artificial intelligence. Morgan Kaufmann
go back to reference O’Leary DE (2013) Artificial intelligence and big data. IEEE Intell Syst 28(2):96–99 CrossRef O’Leary DE (2013) Artificial intelligence and big data. IEEE Intell Syst 28(2):96–99 CrossRef
go back to reference Petrison LA, Blattberg RC, Wang P (1997) Database marketing: past, present, and future. J Direct Mark 11(4):109–125 CrossRef Petrison LA, Blattberg RC, Wang P (1997) Database marketing: past, present, and future. J Direct Mark 11(4):109–125 CrossRef
go back to reference Plasek A (2016) On the cruelty of really writing a history of machine learning. IEEE Ann Hist Comput 38(4):6–8 Plasek A (2016) On the cruelty of really writing a history of machine learning. IEEE Ann Hist Comput 38(4):6–8
go back to reference Russell SJ, Norvig P (2010) Artificial intelligence: a modern approach, 3rd edn. Pearson Russell SJ, Norvig P (2010) Artificial intelligence: a modern approach, 3rd edn. Pearson
go back to reference Schutzer D (1990) Business expert systems: the competitive edge. Exp Syst Appl 1(1):17–21 CrossRef Schutzer D (1990) Business expert systems: the competitive edge. Exp Syst Appl 1(1):17–21 CrossRef
go back to reference Smith B, Linden G (2017) Two decades of recommender systems at Amazon.com. IEEE Inter Comput 21(3):12–18 CrossRef Smith B, Linden G (2017) Two decades of recommender systems at Amazon.com. IEEE Inter Comput 21(3):12–18 CrossRef
go back to reference Smolensky P (1988) On the proper treatment of connectionism. Behav Brain Sci 11(1):1–23 CrossRef Smolensky P (1988) On the proper treatment of connectionism. Behav Brain Sci 11(1):1–23 CrossRef
go back to reference Stevenson TH, Anthony Plath D, Bush CM (1990) Using expert systems in industrial marketing. Ind Mark Manage 19(3):243–249 CrossRef Stevenson TH, Anthony Plath D, Bush CM (1990) Using expert systems in industrial marketing. Ind Mark Manage 19(3):243–249 CrossRef
go back to reference Sutton RS, Barto AG (1998) Reinforcement learning: an introduction. The MIT Press, Cambridge Sutton RS, Barto AG (1998) Reinforcement learning: an introduction. The MIT Press, Cambridge
go back to reference Szeliski R (2010) Computer vision: algorithms and applications. Springer Science & Business Media Szeliski R (2010) Computer vision: algorithms and applications. Springer Science & Business Media
go back to reference Tkác M, Verner R (2016) Artificial neural networks in business: two decades of research. Appl Soft Comput 38:788–804 CrossRef Tkác M, Verner R (2016) Artificial neural networks in business: two decades of research. Appl Soft Comput 38:788–804 CrossRef
go back to reference Vahn G-Y (2014) Business analytics in the age of big data. Bus Strategy Rev 25(3):8–9 CrossRef Vahn G-Y (2014) Business analytics in the age of big data. Bus Strategy Rev 25(3):8–9 CrossRef
go back to reference Venugopal V, Baets W (1994) Neural networks and statistical techniques in marketing research. Mark Intell Plan 12(7):30–38 CrossRef Venugopal V, Baets W (1994) Neural networks and statistical techniques in marketing research. Mark Intell Plan 12(7):30–38 CrossRef
go back to reference Wagner WP (2017) Trends in expert system development: a longitudinal content analysis of over thirty years of expert system case studies. Exp Syst Appl 76:85–96 CrossRef Wagner WP (2017) Trends in expert system development: a longitudinal content analysis of over thirty years of expert system case studies. Exp Syst Appl 76:85–96 CrossRef
go back to reference Wierenga B (1990) The First Generation of Marketing Expert Systems Wierenga B (1990) The First Generation of Marketing Expert Systems
go back to reference Wierenga B, Van Bruggen G, Harm G, van Bruggen. (2000) Marketing management support systems: principles, tools, and implementation. Springer CrossRef Wierenga B, Van Bruggen G, Harm G, van Bruggen. (2000) Marketing management support systems: principles, tools, and implementation. Springer CrossRef
go back to reference Yarowsky D (1992) Word-sense disambiguation using statistical models of Roget’s categories trained on large corpora. In: COLING 1992 Volume 2: The 15th International Conference on Computational Linguistics Yarowsky D (1992) Word-sense disambiguation using statistical models of Roget’s categories trained on large corpora. In: COLING 1992 Volume 2: The 15th International Conference on Computational Linguistics
go back to reference Yu D, Deng L (2016) Automatic speech recognition - a deep learning approach. Springer Yu D, Deng L (2016) Automatic speech recognition - a deep learning approach. Springer
go back to reference Zhang S, Yao L, Sun A, Tay Y (2019) Deep learning based recommender system: a survey and new perspectives. ACM Comput Surv (CSUR) 52(1):1–38 CrossRef Zhang S, Yao L, Sun A, Tay Y (2019) Deep learning based recommender system: a survey and new perspectives. ACM Comput Surv (CSUR) 52(1):1–38 CrossRef
Metadata
Title
Algorithmic Marketing
Author
Federico Galli
Copyright Year
2022
DOI
https://doi.org/10.1007/978-3-031-13603-0_2