Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Already in 2000, Kenny and Marshall claimed the need for Contextual Marketing. They challenged companies “…to use the power and reach of the internet to deliver tailored messages and information to customers at the point of need.” [1] Compared to then, technology today provides numerous possibilities for measuring context and delivering not only tailored information but also other context specific marketing reactions. Sensors, embedded systems, wearables, visual and audio recording – all enlarge the scope of digitally visible context parameters. Highly important, however, are suitable algorithms to transform data into knowledge allowing real time reaction. To develop such algorithms for real-time contextual marketing, models are required that can capture the interdependencies between three factors: the context situation, the value generated while using a digital service or product and the marketing reactions influencing the situation. The objective of this paper is to design a conceptual framework for context sensitive digital marketing. For that purpose, we will classify relevant context dimensions and describe theoretical models as a basis for the development of an algorithmic processing context sensitive to real-time marketing reactions. We shall group possible marketing reactions and generate some ideas for innovations in context sensitive online services. Finally, we will draft the challenges on digital brand management within a context sensitive real time world. The guiding principle for these concepts is the “Service Dominant Logic” (SDL) approach.

2 Contextual Digital Marketing Based on the Service Dominant Logic Approach

2.1 “Value in Context” as a Paradigm of Digital Marketing

Marketing must offer benefits, that is, promote value-creation not just for the company, but especially for its customers and target groups. Such values may be informational (“information value”), entertaining (“experience value”), or supporting (“service value”) [2, 3]. An information value can arise, for example, from algorithmically generated, individual product recommendations, such as are provided by Amazon Prime or Netflix. An experience value arises for example through entertaining content or usability that is appropriate to the target group. A comfortable search function within Netflix can provide a service value.

Customer values are always context-dependent which fact justifies the service dominant logic approach (SDL approach). It suggests that value for the customer is generated only while using a product or service (“value in context”) [4,5,6]. It is not the provider who creates value, but the customers themselves who are “value producers” by integrating their resources into an interactive usage process characterized by specific context factors. The supplier offers only value potentials. Condensation of this process is a cognitive and emotional customer experience, from which future expectations of benefit arise.

For providers of digital services, the task is to support this “production process” on the customer side as well as possible. It is this value in context perspective that makes the SDL approach ideal as a conceptual framework for digital marketing. It helps the automated identification of both the resources entered by the customer in an interaction situation and the specific context factors that characterize it. The classification of these two aspects and the diagnosis of their influence on the value in context is a prerequisite to be able to react algorithmically and automatically along with the usage situation.

2.2 Definition of “Context” and “Context Sensitive Digital Marketing”

The term context is used in different scientific disciplines and is correspondingly complex. In the linguistic and literary sciences, context denotes analyzing and understanding of verbal expressions or texts that are created under specific circumstances. In psychology and sociology, context is considered to be an influencing factor on human behavior and the shaping of social network relationships [7]. In computer science, on the other hand, context plays an important role in the development of so-called “context-sensitive” systems within the framework of “pervasive computing”, “context aware computing” [8, 9], “ubiquitous” or “embedded computing” concepts.

Dey provides a frequently cited definition from the point of view of “context aware computing”: “Context is any information that can be used to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves” [10]. Therefore, situational aspects are only contextual if they are closely related to the interacting person or the object. Context thus includes a valuation of a situational aspect about its relevance for a concrete action [11]. Out the perspective of pervasive computing, Ferscha concretizes the concept of context sensitivity as a “… system behavior that considers the present and possibly expected future situation of an artefact or the user and acts in a correspondingly plan-based (intelligently) manner.” [12]Footnote 1

In linking these two perspectives, we define context-sensitive digital marketing as plan-based, automated marketing reactions based on the identification and interpretation of the relevant circumstances - relevant to the respective interaction situation and the related purpose. The identification and interpretation of the context as well as the marketing reactions derived from it are usually automated and real-time. One of the challenges of digital marketing is therefore to filter out the most relevant factors for customer-oriented value generation from the multiplicity of situational factors surrounding an interaction situation in real-time [13].

2.3 Classification of “Context”

Context classifications support the identification of value-relevant factors by providing a “search frame” orientation. With marketing aspects, the linking of the sociological-psychological and the information-technological perspective appears promising. However, we see three broad deficits in the approaches discussed so far in the literature. These are usually enumerative, barely link the mentioned perspectives, or they do not adequately consider the new possibilities arising from developments in sensor technology.

Therefore, we have developed a two-dimensional model for context classification. Each context factor can accordingly be described using two dimensions:

  • Dimension 1: Inner and outer context factors [13].

  • Dimension 2: Latent and acute context factors.

Dimension 1 establishes the reference of the context factor to the user. With the “inner context” we refer to context factors of the usage situation, which are directly connected to the person generating the benefit. This includes the psychological and demographic characteristics of the user, his or her personality as well as cultural character and the associated long-term values. Other context factors in this class may be the belonging to a specific customer segment, a social group or a social milieu. We also consider here the utilization history or intensity and the duration of the relationship with the service provider. Particularly relevant are the motivation and the intention, as well as the current emotional mood that determines the interaction situation. Precisely such factors can be quite different depending on the usage situation even regarding an identical object of action.

Outer context factors are the aspects directly surrounding the acting individual, which are usually consciously perceived and which shape the contact situation. These factors interact with the inner context and influence the intentional, motivational and emotional processes in the contact situation. Outer context factors include the technical, temporal, spatial, social and climatic context within which an interaction takes place. To illustrate: there are experimental studies showing that purchasing behavior, perceived value, and price stability are influenced by the device used (technical context) [14, 15]. The spatial environment, e.g. private or public space [16] influences the emotional condition of the user, the usage behavior and thus the perceived value [17]. Similarly, the temporal or climatic context may also act as a value-influencing factor in a usage situation. Other aspects which indirectly affect the use situation and exert an influence on their experience are, for example, the geographic space within which the usage situation is located (for example, during a holiday trip) and the associated cultural environment of use.

Dimension 2 describes the relationship between the depicted outer or inner context factors and the concrete usage situation. We distinguish between “acute” and “latent” context. The latent context includes those influencing factors that have an indirect value-influencing effect in a concrete usage situation. These are rather static compared with the acute use situation. They exert an influence on the specific interaction situation but are not strongly influenced by it (for example, stable psychographic or demographic characteristics, cultural characteristics, climatic conditions, etc.). Even if they are stable in a largely unchanged manner over several use situations, their influence on the respective value generation may differ due to interactions with other factors.

The acute context refers to influencing factors that define the individual moment in a unique way with respect to a specific usage situation. These can, on the one hand, be distinguished by their degree of variability during the interaction. A supplier could record these in a real-time situation and to influence them through interaction-specific value propositions. These factors include, for example, the acute emotional condition of a user that can be weakened or strengthened by specific digital marketing reactions or the concrete motive behind a usage situation, which can be changed by dynamic reactions. Different from these very volatile context factors, on the other hand, are those which are unique in terms of the value-generating interaction situation (for example, persons present, temperature of the room, public or private interaction site) but that cannot be changed from the supplier’s point of view.

In Table 1 the two-dimensional classification of context factors will be summarized by means of examples.

Table 1. Two-dimensional context classification and example assignment of context factors.
Full size table

The acute context factors must be identified, classified and converted into automated marketing reactions based on real-time data. The latent context factors, on the other hand, can be allocated based on historical data and linked to the real-time data. The combination of context factors of different types allows the definition of diverse contextual scenarios on the basis of which automated, digital marketing reactions can be defined in a real-time plan-based manner.

3 Identification of the Context in the Digital Usage Situation as the Base of Context-Sensitive Digital Marketing

3.1 Necessary Algorithms and Models for Contextual Digital Marketing

In the field of computer science, there are different, in some cases overlapping, research areas that offer approaches for the development of context-sensitive systems. Examples include “pervasive computing”, “context aware computing”, “ambient intelligence”, “ubiquitous” or “embedded computing” [18, 19]. What is common to all is ultimately the question of how information technology can ubiquitously support people discretely in reaching their goals and intentions while considering situational aspects.

The technical possibilities to find answers to this question have grown enormously in recent years. The advancement of sensor technology is at the forefront of capturing context-oriented data. Ever smaller, more powerful sensors, embedded in everyday objects (embedded systems) offer theoretically almost unlimited possibilities to understand the contextual situations in which people are acting. These sensors often do not work independently, but are networked, communicate with each other, and transmit their data to central servers or data warehouses in the cloud (“internet of things”). Data generated from different sources can be aggregated into “big data” pools, which, using appropriate analysis processes and instruments (real time processing), allow a realistic and comprehensive picture of the current context and context-appropriate reactions [20].

Smart phones, but also smart TV, wearables, such as smart watches or smart textiles, smart meters, connected, autonomous vehicles – all these systems include sensor technologies and embedded analysis algorithms for detecting the user context. Smart TV, for example, already offers the option of switching regional adapted TV advertising based on geographic data. If the smart TV is within the same Wi-Fi as the smartphone, tablet or notebook of a user, then the usage data collected there can theoretically be combined with those of the Smart TV. This allows a deep understanding of interests, preferences and moods of a user. The smart phones like those developed as part of Google’s “Tango Project” can actively detect their environment and display it in real-time. Outer and inner context factors become comprehensible in the usage situation. Machines which “see” or understand language provide possibilities for automated cognitive performance in the assessment of context situations, also situations involving emotional or social interactions [12, 21].

The data provided by sensor technology and other sources require algorithms – i.e. unique, automated executable action rules – for dealing with the data collected to give it meaning relevant to the solution of a specific problem.

Depending on their contribution to solving the problem, three different types of algorithms can be distinguished:

  • Algorithms that provide action-guiding information; E.g. Information about the shopping history and the product interests of an online buyer.

  • Algorithms that also link information to knowledge and, on this basis, provide automated recommendations for action; E.g. Information about product interests of an online customer linked to the actual buying behavior of similar customers and resulting purchase recommendations.

  • Algorithms that automatically decide and trigger context-dependent actions as well as independent actions for task fulfillment. E.g. An automatic braking operation of a collision protection system in a car or bots which independently pre-configure the goods baskets or order products upon reaching certain price thresholds.

The development of such algorithms should be based on sound theory and empirically verifiable models. From our point of view, three model types are necessary to implement a context-sensitive digital marketing approach to increasing the value in context during digital interaction situations (see Fig. 1):

Fig. 1.
figure 1

Necessary models for developing digital marketing algorithm.

Full size image

  • Models that attribute the context to data: E.g. the emotional dimension “relaxed” can be attributed to usage time (evening), used device (tablet) and tiredness (slow control activity).

  • Models describing the effect of the context: These help to identify the relevant context dimensions for a specific usage situation, as well as to determine the direction and magnitude of its influence on the value in context [22].

  • Marketing response models: These models show what marketing reactions under specific context conditions are suitable to optimize the customer experience during the interaction and thus the value in context. This requires modeling of the relationship between value proposition, contextual situation and value creation (value in context).

For the development of algorithms of the relations formulated in these models, computer science and statistics provide a variety of concepts and tools. Most recently, there have been significant advances in both the so-called symbolic and sub-symbolic procedures of artificial intelligenceFootnote 2. Taxonomies, ontologies, or neural networks help in the definition of context situations and the systematization of possible value propositions. Logic, rule-based closing, or “case based reasoning” considering historical data, enable the implementation of self-optimizing, learning-capable marketing reaction models. This also applies to non-directional or directed correlations between contextual situations and value propositions [22]. Correlation or similarity measures provide the statistical basis for collaborative or content-based filtering. Netflix, for example, uses this to automatically display product recommendations [23]. Simple or Bayesian probability calculations can help determine the relevance of different context dimensions for the value in context. Causal relations between the context situation and value in context can also be recognized by multivariate statistical methods or data mining techniques.

3.2 The Reaction Possibilities of Contextual Digital Marketing

According to Dey & Abowd context sensitive software systems allow basically three possible responses [10, 20]:

  • Presentation of context-oriented information or functions

  • (Automated) Execution of functions or services

  • Tagging: enhancement of data with context information for subsequent marketing activities

The first aspect is already an important part of a professional online marketing. As part of the Real-Time Bidding and Advertising, Programmatic Marketing or Targeting online marketing actions are controlled depending on different contextual factors.

We see potential for innovations in the offer of context-sensitive, value adding online services. Amazon and Netflix for example show that well-crafted, intelligent recommendation systems deliver valuable information on context-appropriate consumption options. This could be further developed in the future into automated “curated shopping” systems that provide situation-specific, useful and individual recommendations in real time. Further potential for increasing the value in context results from context-sensitive assistance and support systems. “real-time pricing” or “dynamic pricing” models may establish individual and context specific prices depending on the customer-specific payment readiness that has been algorithmically determined in the specific context.

The contextual information of a usage situation need not necessarily always act as trigger for real-time based digital marketing responses. It may be also useful to archive them first and use them later. For this purpose, it is usually necessary to link the manifold of incoming context-based information with data from other sources by using methods like record linkage, data matching and/or data fusion [25]. This creates valuable “big data” pools, which can be exploited by methods of data mining.

4 Challenges for the Digital Brand Management

4.1 The Digital Brand Management Frame

It has been shown that new technological possibilities, created by sensor technologies and the connected internet of things, create almost unlimited options for a data-based, real-time-based digital marketing. Previous approaches, e.g. the already mentioned real-time bidding, targeting or programmatic marketing or the use of the currently strongly discussed “bots” in the marketing communication fall short however. They are by no means sufficient to exploit the available technological tools for identifying relevant context factors and for deriving automated marketing reactions. In addition, their field of application is limited to a comparatively small task of brand management. Brand management overall, however, faces the challenge of exploiting the many new possibilities.

To this end, a rethink or paradigm shift of traditional brand management appears necessary. In the literature, the context-oriented design of the brand appearance is often assigned to the “operational perspective” [26]. This point of view ignores the value in context as the driver of the customer-oriented brand value in a digital real-time world. The value in context perceived by the customer in an interaction situation is essentially determined by the value contribution a brand can achieve, considering in particular the acute context factors.

Therefore, real-life situations should not be dismissed as an operational task, but rather planned strategically. This means putting more focus on the contribution a brand can deliver to the value in context, considering the acute context factors in a specific interaction. Bonchek and France formulate aptly: “A brand is not something you manage over time. It’s something you deliver in the moment” [27]. Rigid positioning models need to be made dynamic without diluting the brand identity. Keller claims in this regard “..provide them (consumers, the author) with a highly customized and tailored brand experience..” [28] and he warns against the risk of dilution of the brand [28, 29]. For these reasons, other authors demand that the brand’s “responsiveness” and “interactivity” be improved in the digital environment [30, 31].

In the real-time digital world, we see a central task of context-sensitive brand management in the re-alignment of the tension field between brand continuity and brand adaptation. We refer to this adjustment as the “brand viscosity”, that is, the ability to adapt and integrate the brand in a specific interaction situation. The brand viscosity must be determined in such a way that an optimal balance is found between the respective value in context on the one hand and the customer-oriented brand value on the other. The illustrated models for context attribution, effect and reaction require a distinctive, brand adequate operationalization. Based on this, algorithms can be developed that realize a defined viscosity of the brand in concrete situations. It is necessary to identify the relevant context factors for the brand, to interpret them in their interactions, to anticipate contextual scenarios and to develop strategy-compliant reaction patterns for the digital interaction.

If one follows the idea of the service dominant logic approach, the generation of value in context is dependent not only on the value offered by the brand, but also on the customer’s responses to specific context factors. Merrilees therefore sees a paradigm shift from a “customer-centric” to a “customer-driving” marketing [31]. The value in context is influenced by the context-dependent behavior and feeling of the customer himself. The customer becomes the “value (co-) creator”. His or her ability and willingness to be involved in the generation of values in the acute use situation is also dependent on, or interacts with, personal emotions and experiences. These, in turn, are ultimately value-creating or determine the value in context of the customer [29, 32].

A further central task of context-sensitive brand management is therefore twofold: First, to positively influence the customer’s context-dependent willingness to integrate into the process of value generation; second, to influence the customer’s experiences with the process during the interaction. To this end, based on the models presented, systems are to be developed that can be addressed to customers in real-time in a context-oriented manner that promote customers’ willingness to integrate. The aim is to increase the brand experience, the associated value in context and ultimately the “customer based brand value”.

Figure 2 summarizes these ideas. It illustrates that the value in context is, on the one hand, influenced by the brand viscosity, i.e. the adaptability of the brand to the specific context of the interaction situation. On the other hand, it is influenced by the customer’s context-dependent willingness to integrate. The value in context, determined by brand viscosity and customer integration, then forms the basis for the creation of a customer-oriented brand value as the central goal of the brand management [33].

Fig. 2.
figure 2

Brand viscosity and customer integration as the central challenges of context-oriented brand management.

Full size image

4.2 Brand Viscosity as Challenge for the Digital Brand Management

Figure 2 shows, on the left, the basic design elements for determining the brand viscosity [34]:

  • The brand substance, i.e. the functional, emotional, aesthetic and social brand performance in the sense of the offered value potential.

  • The brand image, i.e. the associations and perceptions associated with the brand, caused by the communication of the value potential by means of brand-specific symbolism, language and stories.

  • The brand relations, i.e. the definition of the roles and responsibilities between supplier and customer in the context of an interaction situation [27].

With respect to the brand substance, this means for example that the functional, emotional, aesthetic and social performance of the brand is carefully enriched with contextual components. A successful example of this is the context-dependent, event-oriented design of the emotional-aesthetic brand performance of Google’s signature in the form of Google doodles.

Further possibilities include the augmentation of the core performance of a brand by context-dependent additional services. For example, the automotive industry already offers its customers, in a variety of ways, context-sensitive, digital “smart services”. They help to increase the value contribution of the brand to the customer in a specific usage situation. The potential, derived from consideration of, in particular, acute context factors, seems far from being exhausted.

The brand substance forms the basis for the creation of a credible, authentic brand image. This can also be outlined in a context-dependent manner and in real-time. Thus, it is conceivable and technically possible, depending on the context situation, to select from a set of brand attributes predefined within the framework of the positioning strategy those which are specifically relevant to the context, and to emphasize them in communications with the customer. From a predefined set of brand-specific symbols, language styles and stories, it should be possible to select, combine, and digitally display those that promise the highest value in context. To that end, the mentioned elements that enrich the brand with meaning should be modularized, then a suitable combination of those modules should be assigned to context scenarios within the context effect model and integrated into the marketing reaction model.

Finally, it is also necessary to shape brand relationships with customers in a context-specific manner. Depending on the context, the roles between the provider and the customer can be defined differently within the framework of the interaction, and role-specific information or services can be offered in real-time. For example, depending on the proficiency of a viewer, Netflix could assign different roles (for example expert or layman) and use it for corresponding recommendations or information.

4.3 Customer Integration as Challenge for the Digital Brand Management

Within the SDL approach, the customer’s engagement during the interaction situation, his or her willingness to provide resources in the moment, and the resulting experiences become central for the value creation. The principles developed there can be guidelines for brand management in the digital real-world. Especially in the world of social networks, customers are active co-creators of the brand, of its identity and of the value-generating experiences that relate to it [35]. The importance of the customers in shaping the brand has grown so far that the question is increasingly raised as to whether the brand itself still belongs to the company [36].

While suppliers can decide relatively autonomously about the design of brand viscosity, a challenge is to consider the customer’s role in value generation in context-sensitive, algorithm-based marketing concepts. To meet this challenge, we should answer three central, interrelated questions:

  1. 1.

    How do acute context factors (for example, the mood of a user or the location) influence the customer experiences and feelings at the moment of the interaction and how can algorithms respond to different context constellations (interaction-related customer experiences) [31]?

  2. 2.

    How strong is the involvement and engagement of the customer in the specific interaction situation with respect to the brand (interaction-related customer engagement) [28, 31] and what possibilities exist from the supplier’s perspective to influence this with the objective of maximizing the value in context in real time?

  3. 3.

    In the acute use situation, considering the specific context factors, how can the customers’ willingness and ability be positively influenced to integrate themselves and their resources into the value-added process? (value co-creation)

In the comprehensive literature, the “customer experience” and “customer engagement” concepts are usually considered in the role they play in the long-term customer relationship over several interaction phases [37, 38]. From our point of view, this focus must be supplemented in a real-time world by the insights gained from the acute, interaction-related perspective. At the core of context-sensitive brand management is, above all, the dynamics that arise between the supplier and the user at the moment of use of the brand, considering the specific contextual factors in the immediate situation. The context-dependent possibilities of influence in the moment of value creation are to be considered more intensively in the light of the now diverse technological possibilities, while considering the strategic objectives of the brand management.

In particular, Merrilees models the relationship between interactive “brand experiences”, customer engagement and value co-creation [31]. He clarifies that the brand-specific sensations and experiences generated in the interaction situation substantially influence the customer’s commitment and willingness to participate in the use situation. He also assumes that the willingness to value co-creation differs significantly because of different interactive experiences with “hedonic brands” and “functional brands”. While Merrilees assumes a moderate willingness to participate in functional brands due to the dominant cognitive processes, the emotional experiences of hedonic brands lead to a strong willingness to participate in the generation of value [28, 31]. Therefore, comparable context constellations for the same customer are to be interpreted differently from the point of view of brand management, depending on the type of brand, meaning they are to be addressed with different marketing reactions.

But Merrilees hardly considers the significance of acute context factors for the value in context. The concept of brand management needs to formulate “customer experience” as more than experience and sensation related to the moment of the interaction. Experiences are more than “long-term experiences” with the brand. Hence, the need arises to provide value-generating experiences in their dependence on acute contextual factors, e.g. the acute emotional mood of the user, the existing social environment or the location of the use.

For example, Netflix could design its product recommendations very differently depending on whether a user is sad or euphoric, lonely or together with a partner or in a group of friends seeking relaxation. In the same way, the cognitive, emotional or behavior-oriented commitment of the customer should also be considered as context-dependent and possibly open to influence from the marketing strategy. After a long working day, a tired, overworked user will have a different “psychological or physical investment… in that brand” [39] than in a relaxed holiday situation. Finally, the willingness and ability of a customer to bring resources (for example, knowledge, personal data, time, etc. [40]) into the process of value creation must also be viewed and controlled in a context-dependent manner. Digital brand management should better understand customers as a resource of value creation in the usage situation and develop algorithms to influence customer integration depending on the contextual factors characterizing the situation.

5 Summary

In this article, a concept is developed and presented based on the Service-Dominant Logic Approach. The concept can serve as an orientation in the design of context-sensitive digital marketing approaches. It was shown that in a real-world environment, the “utilization moment” and the factors influencing it are central in the creation of a value in context. This is at the same time the driver of the customer-oriented brand value. Brand management in the digital real-time world becomes more complex as brand perception and brand use are context-dependent, changeable and more difficult to influence at any time. To overcome this complexity, it is necessary to formulate brand-specific models for context allocation, context effect and marketing reaction. A central task of brand management in the digital real-time world is to formulate context scenarios and to define possible marketing reactions. For this purpose, it is important to understand which context dimensions are relevant for the brand and how they function in the value creation process. At the same time, to ensure the integrity of brand management, it is imperative to prevent unintended contextual scenarios in digital usage situations. Even in a real-world environment, the importance of the brand is retained: strong brands shape the context. Weak brands are formed by the context!