1 Introduction

Although advertising expenses have increased, advertising effectiveness has suffered dramatically in recent years. Consumers are overwhelmed by the quantity of advertising messages they receive, and it is getting more difficult to attract consumers’ attention (Ha and McCann 2008). One of the best ways to break through the information clutter in digital media and to increase advertising effectiveness is to target advertisements dynamically—to “provide the right content in the right format to the right person at the right time” (Tam and Ho 2006).

In marketing, scholars use the term personalization to refer to the tailoring of products, services, or content to consumer needs, goals, knowledge, interests, or other characteristics (Zimmermann et al. 2005). Various terms (e.g., customization, profiling, segmentation) are used to refer to personalization, which itself is associated with many different strategies of adaptation (e.g., adapting the layout of a message to a display, tailoring content to consumer characteristics) (cf. Vesanen 2007).

Personalization has proven successful at increasing the effectiveness of advertisements [e.g., users show increased attention to an advertisement (Malheiros et al. 2012), perceive its usefulness (Tam and Ho 2006), or better remember an advertisement (Adolphs and Winkelmann 2010)]. However, recent research has revealed some of personalization’s drawbacks. For instance, consumers can react negatively towards advertisements once they become aware of the personalization (Malheiros et al. 2012). Furthermore, the availability of person-related data (required for personalization) is often limited (Tucker 2012) or restricted by privacy regulations (Goldfarb and Tucker 2011). Hence, personalization is not always possible, nor is it always the strategy of choice.

Another challenge relates to the design and implementation of adaptive advertising systems that rely on sophisticated information systems. These systems need to unite two very different domains (marketing and IT). Thus, it is important to create both a common base of understanding (to enable effective cross-domain communication) and well-defined business and functional requirements (to align business requirements with technical designs and implementations).

Therefore, advertisers (business) and system designers (IT) need guidance to communicate effectively and choose the right advertising adaptation strategy. From a strategic point of view, different forms of advertising adaptation lead to different advertising effects. From a technical point of view, different modalities of adaptation require adaptive advertising systems to perform distinct functions, meaning they will need specific technical infrastructures. It is thus necessary to differentiate between the adaptation strategies that are currently associated with the concept of personalization (Lasinger and Bauer 2013).

This paper proposes a novel concept, the PERSIT matrix, which provides a systematic overview of strategic options for advertising adaptation. The paper distinguishes between different strategies of advertising adaptation, relates them to their respective advertising effects, and outlines their distinct functional requirements. In line with Lewin (1953) and communication theory (Schulz von Thun et al. 2003), the PERSIT matrix differentiates between person-centric and situation-centric adaptation strategies and relates them to each other. Person-centric adaptation strategies use a consumer’s person-related data, whereas situation-centric adaptation strategies leverage data about the situation where an advertisement is presented.

The proposed PERSIT matrix enables marketing experts to choose the adaptation strategy that best fits the desired advertising goals. Because the matrix also reflects a (technical) design viewpoint on adaptation, it facilitates communication between marketing experts and designers of adaptive advertising systems. In fact, the PERSIT matrix’s utility was confirmed in expert discussions with marketing specialists.

The remainder of the article proceeds as follows: First, we briefly discuss the conceptual foundations of personalization. Next, we detail the PERSIT matrix, a classification scheme for studies in the field of adaptive advertising and a central contribution of this paper. Then we outline our methodology for performing a thorough literature review and evaluating the matrix’s utility. After that, we discuss the benefits and challenges of the adaptation strategies that result from the PERSIT matrix, as well as the functional requirements and potential advertising effects of each strategy. Then, based on our literature review, we pinpoint research gaps that should be addressed with future research. After discussing the utility of the PERSIT matrix as considered by marketing experts, we conclude with a summary of the paper’s contributions.

2 Conceptual foundations

While personalization in advertising has a long history (Vesanen 2007), the dynamic adaptation of advertisements in real time, which is enabled by digital media, is rather new (Adams 2004).

2.1 Personalization of advertisements

The concept of personalization arose as early as the 1870s, and by the 1970s, it was discussed mainly in connection with segmentation and targeting. Nowadays, several terms (e.g., profiling, segmentation, targeting, filtering, tailoring) are used to describe the concept of personalization; it is a broad concept that includes topics such as diagnosing location, fitting the visual layout of the message to data terminal equipment, and tailoring message content (Vesanen 2007).

A basic form of personalization is grounded in the idea of ‘market segmentation’. A target population (i.e., the anonymous mass market) is divided into segments that are internally as homogenous as possible and externally as heterogeneous as possible; each segment is then provided with a specific tailored advertisement to increase advertising effectiveness for that segment. In the specific case where the size of a segment equals one, we speak of ‘individualization’ or ‘one-to-one marketing’: Consumers are addressed individually with customized messages or services (‘mass customization’) (Reutterer 2003). The basis for dividing a broad target market into subsets of consumers who have common needs is flexible; the market can, for instance, be segmented based on culture, lifestyle, or purchasing behaviour.

Besides consumer characteristics, the current situation influences thinking and behavior (Lewin 1953); this fact has long been acknowledged by research in psychology and marketing: Studies of television advertising (De Pelsmacker et al. 2002) and print advertisements (Dahlén 2005) have shown that ‘editorial context’ (the content surrounding an advertisement) highly impacts how advertising affects its audience.

This fundamental idea of accounting for situational context has been successfully applied to advertising in classical media: in all the approaches, advertisers anticipate a specific constellation of situational context and select an advertisement based on their assessment of that situation.

The situation-specific adaptation of advertisements has various effects. For example, a thematic fit between an advertisement and the respective magazine was also reported to increase positive feelings towards the advertised brand (Jun et al. 2003).

However, the marketing discipline frequently reduces situational context to ‘editorial context’—the content that surrounds the advertisement in a magazine, on the radio, or on television. Although researchers acknowledge that this narrow view of situational context is limiting, to date little work has offered a broader perspective of situational context in advertising: “...virtually all studies have focused on the immediate editorial context, such as articles or television shows or segments, rather than on the total media context” (Dahlén 2005). The rare exceptions to this trend are studies on the impact of ‘atmospherics’ on purchase behavior at the point of sale (Newman et al. 2007).

2.2 Real-time personalization of advertisements

Digital media enable the adaptation of advertisements in real-time, supported or automated by information systems. Real-time personalization of advertisements can, hence, be described as the dynamic changing of advertising messages on digital media in real time based on information that is sourced from the targeted consumer or the situation shared by the advertising system and the consumer.

More than a decade ago, researchers described information systems that were aware of their context and could adapt to it. On the Web, personalization methods adapting in real time are already very advanced, making use of a wide range of techniques such as content-based filtering, collaborative filtering, and hybrid filtering, which combines the first two techniques (e.g., Adomavicius and Tuzhilin 2005). These filtering techniques are based on modern statistical machine-learning methods. Real-time personalization has been proven to increase advertising effectiveness. For instance, it increases the user’s level of attention (Malheiros et al. 2012) and perception of the usefulness of the information displayed (Tam and Ho 2006). Thus, personalized advertising seems to be particularly effective when addressing consumers whose preferences are (already) narrowly construed, since the advertisements focus on specific product details.

Typically, personalization is based on user information such as identity, demographics, lifestyle, specified preferences, past purchases, and historical visit patterns. Nowadays, situational aspects are rarely considered in real-time personalization settings. Situational adaptation strategies are merely understood as an addition to existing personalization approaches (e.g., Zimmermann et al. 2005).

3 The PERSIT matrix and its constructs

While existing literature typically considers personalization as one holistic construct, we differentiate between person-centric and situation-centric adaptation in respect to the design of an adaptive advertising system. As we will show, it is beneficial to explicitly distinguish between these two adaptation concepts, since, besides different functional requirements for designing such systems, these two concepts have different implications for advertising effectiveness.

To address these differences, this paper suggests the PERSIT matrix, which differentiates between PERson-centric and SITuation-centric adaptation strategies. In this section, we define the constructs that provide the foundation for this categorization matrix as represented in Table 1. In doing so, we draw on theory from the fields of advertising, communication psychology, and information systems.

Table 1 The PERSIT matrix with adaptation strategies
Full size table

As outlined in the introduction, decades of advertising research has shown that personalization improves advertising effectiveness (Vesanen 2007). While advertisements are traditionally tailored to an individual consumer or consumer segment, other factors that impact advertising effectiveness include the editorial context (Dahlén 2005) and the broader atmosphere (Newman et al. 2007). We used this (implicit) differentiation between person- and situation-related factors to build the PERSIT matrix.

Likewise, psychological research theorizes that consumer behavior is derived from personal and situational factors (Lewin 1953). Specifically, communication theory states that communication has two important anchors of reference, the person and the situation (Schulz von Thun et al. 2003). Effective communication addresses both. We transferred this concept from the field of interpersonal communication to the field of advertising adaptation.

Research on information systems has provided insight into the technical implementation and implications of personalization. On the one hand, some information systems use information about the user (e.g., personal characteristics, past transactions, the user’s task) to adapt their services (Tam and Ho 2006); other information systems use awareness of their situational context (e.g., weather, location, time) to adapt their services. Both system types have specific functional requirements and face distinct challenges. The PERSIT matrix links these two system types to adaptation strategies, which emerge from advertising and psychological research.

Building on these theories, the PERSIT matrix presents person-centric adaptation on the vertical axis and situation-centric adaptation on the horizontal axis. As we have outlined in Sect. 2.1, there are basically three archetypes for person-centric adaptation (Reutterer 2003): no person-centric adaptation (addressing the anonymous mass market), adaptation to a group of individuals (segmentation, one-to-many person-centric adaptation), and addressing a particular individual (individualization, one-to-one person-centric adaptation). Considering this differentiation, the PERSIT matrix includes six possible adaptation strategies. All of these strategies inherit the advantages and challenges associated with person-centric adaptation or situation-centric adaptation, or both. Because every strategy implies the use of different information sources, different hardware and software will be required; in turn, these requirements lead to adaptive advertising systems of different technical complexity and expense.

3.1 Person-centric adaptation

In the style of Vesanen (2007), person-centric adaptation can be defined as the use of personal information about a consumer or group of consumers to tailor information communication to an individual or group of consumers. Using information about the consumer—either previously obtained or provided in real time—the provision of information is altered to fit that consumer’s needs, whether those needs have been stated directly by the consumer or perceived by an information provider (e.g., an advertiser).

The strategy of not performing any person-centric adaptation (i.e., addressing the mass market) can be thought of as a ‘one-size-fits-all’ strategy. Narrowing down the mass market to relatively homogenous subsegments and addressing those subsegments separately is a one-to-many person-centric adaptation; this personalization concept is also frequently referred to as ‘segmentation’ (e.g., Vesanen 2007). Further narrowing down the market to a single individual (i.e., concentrating on a specific person) is a one-to-one person-centric adaptation, a strategy that may also be referred to as ‘individualization’ (e.g., Reutterer 2003).

Particularly for the design of adaptive advertising systems, it is crucial to distinguish between advertisements that are tailored to an individual and those that address several people at once. When a system adapts for an individual, it must consider only the characteristics of that individual. When a system adapts for a group, it must categorize individuals according to characteristics that the individuals share. Each option requires different kinds of identification and uses different approaches to analyze consumer characteristics.

Advertising systems for devices that are typically used by one person at a time (private screens such as desktop computers, smartphones, tablets, etc.) may only utilize the strategies of ‘no person-centric adaptation’ and ‘one-to-one person-centric adaptation’ since only in very rare cases will a second or third person join the main user. Public screens (e.g., digital-out-of-home advertising), in contrast, are placed in areas where several people can view the screen’s contents and, accordingly, all three person-centric adaptation strategies are feasible: no adaptation (one-size-fits-all), adaptation to a group of individuals viewing the screen at the same time (segmentation), and one-to-one person-centric adaptation (individualization) for instances where only one person is within the service area of a public screen.

3.2 Situation-centric adaptation

Instead of basing adaptation efforts on personal data, situation-centric adaptation draws on environmental information that is unrelated to an individual. We define situation-centric adaptation as delivering information that is relevant (to an individual or a group of individuals) in the present situation based on information about that situation. Information about the current situation is retrieved, transformed, or deduced from information sources.

An example application of situation-centric adaptation is advertising ear protection in a loud environment. The relevance of the advertisement is enhanced by the situation, regardless of the characteristics (e.g., age, gender, preferences) of the consumers. According to communication psychology, this situation-centric adapted advertisement fits the definition of a communication that is adequate in a particular situation (Schulz von Thun et al. 2003).

3.3 Distinguishing person-centric from situation-centric adaptation

As outlined above, the constructs of person-centric and situation-centric adaptation are orthogonal. According to communication psychology, effective communication fits a particular person and situation (Schulz von Thun et al. 2003). In advertising, person-centric adaptation aims to fit an advertisement to a person, while situation-centric adaptation aims to fit an advertisement to a situation. Ideally, both dimensions are taken into account; however, depending on the purpose, or in the case that one option is unavailable, the other option can still be used to optimize communication efforts (see Sect. 5.1).

There are two major differences between person-centric and situation-centric adaptation: the targeted entity and the data used for the adaptation. Person-centric adaptation targets a defined person, group, or segment. In contrast, situation-centric adaptation does not target an individual, group, or segment, but instead tailors a message to everybody in a given situation. Person-centric adaptation relies on person-related data, data that has meaning only in relation to a particular person or group of people; situation-centric adaptation, however, relies on situation-related data, data that has meaning in relation to a situation but not to a person.

This also means that it is possible to optimize advertisements for a defined person as well as the general population, and to use data sources that relate to a situation and an individual. For instance, imagine advertisements on an online search engine. Traditionally, advertisements are selected based on the keywords entered. This adaptation is person-centric because the advertisements are only relevant to the person who is searching, since the search terms are meaningful only to this person. If these advertisements were shown to any other person, they would very likely not be relevant. However, suppose that a search interface presents searchers with an advertisement based on the time of the search and the location of the searcher by using the most likely language and form of greeting that one would expect for these parameters. This advertisement would likely be meaningful to any other person in the same place and time. In this case, we speak of situation-centric adaptation. Person-centric and situation-centric adaptation can be combined. For example, the search term ‘rock’ could refer to a naturally occurring solid aggregate of minerals or a form of popular music. Advertisements could be selected based on personal information (e.g., previous searches, the user’s job, user preferences), situational information (e.g., a person is more likely to look for rock music when in a bar or at home, while for an aggregate of minerals when taking an expedition), or both.

4 Methodology

In this section, we describe the methodologies employed for performing the literature review and evaluating the matrix’s utility among marketing experts.

4.1 Literature review for digital-out-of-home advertising (DOOHA)

In its essence, the PERSIT matrix was designed to categorize any kind of adaptive advertising. However, reviewing studies on every type of adaptive advertising would be tremendously difficult and is beyond the scope of this paper. Instead, we decided to examine studies on only one notable field of application: digital-out-of-home advertising (DOOHA).

A myriad of descriptors have been used for networks of displays in public spaces. When such displays are primarily used for advertising, we call them digital out-of-home advertising (DOOHA) (Telschow and Loose 2008). We define DOOHA as digital screens in public spaces that can be controlled independently via a centralized network and that are mainly used for advertising purposes.

DOOHA is a rapidly emerging marketing channel that promises to reach out to consumers at any time and any place in physical space. It can be used to display advertisements to consumers who are in transit, waiting, or at commercial locations such as a retail venue. In contrast to revenues from other forms of advertising, DOOHA revenues are growing at an accelerating pace. From 2007 to 2012, DOOHA revenues grew at a compound annual rate of 12.1 %, and they contributed 7.88 billion USD to the global economy in 2012 (PQ Media 2013). DOOHA is said to be particularly influential in the realm of adaptive advertising because, equipped with various sensors, DOOHA allows adaptation of advertising messages in real time and for various purposes (Telschow and Loose 2008). As DOOHA is a public screen solution, all six adaptation strategies of the PERSIT matrix can potentially be leveraged with it. Person-centric information may, for instance, be gathered through Bluetooth connections to individuals’ smartphones where profiles may be stored; cameras may be installed to detect a passerby’s gender; and microphones may allow for emotion recognition in speech. Situation-centric information may, for instance, be sourced from installed sensors such as thermometers (e.g., weather), cameras (e.g., light conditions), or microphones (e.g., noise), or may be sourced via information providers (e.g., web services for weather forecasts) (Bauer et al. 2012).

Besides DOOHA, mobile advertising revenue is growing rapidly, having contributed 9.6 billion USD to the global economy in 2012 (Gartner 2012). Mobile advertising, however, uses individuals’ private screens and is therefore restricted to four adaptation strategies only, because one-to-many person-centric adaptation cannot be realized.

As a result, we decided to focus on DOOHA in our analysis. The procedure for analyzing and categorizing published studies in the field of DOOHA is outlined below.

4.1.1 Sample

We sampled articles that we found in the most common scientific online databases in the field: ACM, EBSCO, IEEE, SpringerLink, and Wiley Online Library. First, we searched these online databases in full text for articles containing the term ‘digital out of home’ or any of the different (and often synonymously used) definitions for DOOHA that were reported by Telschow and Loose (2008) and Kaupp (2010) (Table 2).

Table 2 Retrieved articles per source and search term
Full size table

Using these search terms, 838 articles published before 30 June 2011 were retrieved from the five online databases. The number of articles retrieved from each online database is as follows: ACM (220), EBSCO (117), IEEE (194), SpringerLink (209), and Wiley Online Library (98). A detailed breakdown of the 838 articles regarding search terms and sources is depicted in Table 2. After reviewing the titles and abstracts of these articles, 742 were excluded because they were either retrieved from multiple online databases or unrelated to the topic at hand. The remaining 96 articles made up our final sample.

4.1.2 Coding

The 96 articles were independently reviewed and coded by two reviewers. In the few instances (nine articles) where disagreement emerged, the reviewers discussed the issue until consensus was established. We included the following coding dimensions: (1) type of adaptation (individual, group, situation, none); (2) variables that were used for the adaptation, such as identity or weather; (3) purpose of the application, such as advertising, information presentation, or entertainment; and (4) meta-information (citation information).

Based on coding dimension (3) ‘purpose of the application’, we split the sample into 16 advertising-related articles and 80 articles falling into the broader category of ‘digital signage’ (excluding advertising-related articles). For both sample parts, each of the articles was assigned to one of the six fields of the PERSIT matrix based on the following coding dimensions: (1) ‘type of adaptation’ and (2)‘variables used for adaptation’. The results of this review and analysis are discussed in Sect. 5.

4.2 Expert discussions

We employed expert discussions to ensure the applicability and utility of the PERSIT matrix. The matrix was discussed with five experts (three marketing managers and two operating advertisement design officers) from a global advertising agency. Each of the meetings started with a presentation teaser about the matrix and outcomes concerning advertising effects; a semi-structured discussion followed. After each meeting, insights from the expert discussions were discussed in analytical group reflection sessions by the research team.

5 Results and discussion

5.1 The benefits and challenges of the advertising adaptation strategies

The main advantages of person-centric adaptation are as follows: increased advertising effectiveness (e.g., improved memory or improved attitude towards the brand) (Adolphs and Winkelmann 2010), increased perceived usefulness of information (Tam and Ho 2006), and increased attention towards the advertisement (Malheiros et al. 2012). Person-centric adaptation is also well suited for any situation where just one consumer views an advertisement (one-to-one). Furthermore, person-centric adaptation techniques are based on established technologies and have been studied and optimized for years (Adams 2004).

However, recent research pinpoints the limitations of person-centric adaptation: first, person-centric adaptation is not possible in all situations, and second, its benefits are limited by a phenomenon referred to as ‘personalization reactance’. The first implication is based on the concept of a trade-off between personalization and privacy (Lee et al. 2011). An increase in required privacy or decrease in available personal data reduces the options for person-centric adaptation. The second implication—personalization reactance—describes a phenomenon where consumers feel that advertising using person-centric adaptation intrudes on their privacy; as a result, consumers develop negative emotions towards the advertised product or company, or both (Malheiros et al. 2012; Tucker 2012). We expect that personalization reactance will be particularly high when person-centric adapted advertisements are shown on public displays, since sensitive information is potentially disclosed to nearby people.

The most significant benefit of situation-centric adaptation is that it resolves the main challenges of personalization. Because the adaptation process is based on situational information (shared and available to the general public) and is not based on person-related data, no person-related data can be revealed. Consumers will not feel observed and personalization reactance is avoided. Also, no personal data is required, avoiding a privacy trade-off (Lee et al. 2011). In addition to avoiding the challenges of person-centric adaptation, situation-centric adaptation has other benefits. As stated previously, research has revealed that perceived thematic closeness between television advertisements and the programs during which they air enhances recall (De Pelsmacker et al. 2002). Furthermore, consumers have more positive attitudes when they perceive that an advertisement is relevant and related to its medium (Dahlén 2005).

However, conceptual and technical challenges arise when situation-centric concepts are applied to advertising. For instance, it is not yet clear how situations (context) can best be conceptualized and which parameters should be chosen for adapting advertisements. Existing research mainly focuses on rather ‘simplistic’ variables such as time, place, or weather. Consequently, the empirical implications of situation-centric adaptive advertising are not clear. Technical challenges include the deployment of required sensor infrastructures and the lack of algorithms to match advertisement content to situations.

5.2 Research concerning adaptive advertising in the DOOHA domain

As outlined in the methods section, we reviewed published studies in the field of DOOHA. All 16 articles from the sample that were attributed to the DOOHA domain could be directly assigned to the matrix. As can be seen from Table 3, while eight articles deal with person-centric adaptation, only three articles consider situation-centric adaptation. Yu et al. (2010) and Stalder (2011) describe situation-centric adaptation without person-centric adaptation, basing the adaptation on Global Positioning System (GPS) localization and atmospherics (e.g., weather, light), respectively. In contrast, Eriksson and Åkesson (2008) describe a combination of situation-centric and person-centric adaptation for an individual (based on consumer schedule, time of day, and location).

Table 3 Populated PERSIT matrix with DOOHA-related articles
Full size table

As can be seen from Table 3, more than one third of the articles in the sample (6) follow no adaptation strategy at all; neither person-centric nor situation-centric adaptation options are considered. Most of this research provides general recommendations for DOOHA (Dräger 2011; Huang et al. 2008) or analyzes its effects (Silberer 2010; Taylor and Franke 2011; Burke 2010) without addressing adaptation options.

Another six articles describe one-to-one person-centric adaptation strategies (individualization). The articles all note the need to identify the consumer (e.g., via RFID as described in Kalyanam et al. (2006)) and address privacy considerations (Geiger 2011). Two of the articles refer to using personal information on a user device (a shopping list on a mobile device) to adapt the screen (Krüger et al. 2011; Strohbach 2011).

Surprisingly, only one DOOHA article describes a strategy of person-centric adaptation for a group (one-to-many; segmentation): Harrison and Andrusiewicz (2003) deployed video cameras to capture demographic data that was used to adapt the screen.

No article addresses situation-centric adaptation and person-centric adaptation for a group. Also, none of the retrieved research explicitly considers situation-centric adaptation as an alternative to person-centric adaptation. Instead, deciding how to perform an adequate adaptation is solely based on functional requirements and technical possibilities.

Interestingly, little research about DOOHA describes information systems that dynamically adapt advertisements to the situation (e.g., Eriksson and Åkesson 2008; Stalder 2011; Yu et al. 2010). This finding is surprising since several researchers emphasize the importance of situational context in this setting. For instance, Silberer (2010) indicates that the effects of advertising largely depend on the display’s environment. Also, Telschow and Loose (2008) state that advertisements that are directly related to the current purchasing situation are remembered better. Where researchers do consider adaptation to the situation, they generally perceive it not as an independent form of adaptation, but rather as a complement of person-centric adaptation (Zimmermann et al. 2005).

5.3 Research concerning adaptive advertising in digital signage

We also reviewed the remaining 80 articles of our sample, expanding the scope to the entirety of digital signage (Table 4). Again, all 80 articles could be matched to the PERSIT matrix. 26 articles described person-centric adaptation for an individual, and only five considered person-centric adaptation for a group. In total, 16 papers dealt with situation-centric adaptation, with 11 papers focusing on situation-centric adaptation only (without personalization), and five papers describing situation-centric adaptation in combination with person-centric adaptation for the individual.

Table 4 Populated PERSIT matrix with articles not related to DOOHA
Full size table

5.4 Functional requirements of each adaptation strategy

When analyzing the DOOHA literature, we identified technical challenges associated with the different adaptation strategies of the PERSIT matrix. These challenges lead to specific functional requirements for each adaptation strategy. The results are presented in Table 5.

Table 5 Functional requirements associated with the advertising adaptation strategies of the PERSIT matrix
Full size table

When reading Table 5, note that the advertising situation for DOOHA differs from the one on the Web. On the Web, a single user typically browses a website on his or her computer or mobile device. In contrast, DOOHA is encountered in a public space, where several people can see the advertisement concurrently. Therefore, person-centric adaptation must consider several people at once, for example by adapting for the closest consumer (individualization) or by using characteristics that are shared between people (segmentation).

5.5 The PERSIT matrix’s utility as considered by marketing experts

The PERSIT matrix is more than a classification of literature. In this section, we outline the utility of the matrix for marketing practitioners.

As outlined in the methods section, we presented the PERSIT matrix to experts in the field so that they could evaluate the utility of the matrix. They were strongly convinced of the benefits of person-centric adaptation but were surprised to hear about possible drawbacks, such as personalization reactance. Interestingly, they did not worry about the privacy–personalization trade-off; in their opinion, lobbying will ensure continuous and sufficient access to personal data. The interviewees valued the structuring of the different adaptation options in the PERSIT matrix. In fact, they had already experimented with situation-centric adaptation without being aware of it. For instance, they had advertised painkillers after each foul in a soccer match on television. However, the marketing experts had not been aware of the actual benefits of this adaptation strategy. They stated they would consider the matrix in their future client sessions to discuss different adaptation options, particularly with respect to automating advertising adaptations in digital media. The benefits, disadvantages, and open questions for each adaptation strategy from a marketing perspective are outlined in Table 6.

Table 6 Marketing perspective on the advertising strategies of the PERSIT matrix
Full size table

From the experts’ point of view, the PERSIT matrix supports them in (1) considering all possible adaptation strategies, (2) generating new ideas to increase advertisement relevance, and (3) structuring the decision process by evaluating the feasibility and balancing the benefits and challenges of the considered adaptation options (for instance, as part of technical, legal, and business due diligence).

6 Conclusions

The personalization of advertising messages promises to break through the information clutter that confronts consumers. While personalization is considered a rather broad concept, from a strategic (advertising effect) and technical (system design) point of view, it is necessary to break it down to its basic adaptation strategies. For this purpose, this paper proposes the PERSIT matrix, a systematic overview of available adaptation strategies for personalized advertising organized according to the dimensions of person-centric and situation-centric adaptation. This overview allows advertisers to consider various adaptation opportunities systematically, and to select a feasible adaptation strategy that accounts for regulatory, time, and cost constraints. In other words, it supports systematic, strategic decision making for advertising adaptation that reflects the business requirements of marketing experts as well as the functional requirements of system designers.

This work contributes to the interactive marketing research field in three ways. First, the proposed PERSIT matrix enables marketing experts to choose the best adaptation strategy from among ones with different advertising effects, because the matrix (1) reflects the benefits and drawbacks (regarding business requirements) of the respective adaptation strategies and (2) outlines the different strategic options for adapting advertisements. Second, as the PERSIT matrix provides a level of abstraction that both business people and system designers can understand, it facilitates communication between these groups, since the matrix reflects (1) the different functional requirements for different advertising adaptation strategies and (2) the strategies’ advertising effects. Advertisers can use the strategic options that the matrix offers to create coherent advertising messages. Its utility for marketing experts was evaluated and the experts perceived the matrix to be useful.

Third, the work provides a literature review of adaptive advertising in the field of DOOHA and structures it along the six identified adaptation strategies. A main finding is that research that combines person-centric and situation-centric adaptation is still underrepresented, particularly with respect to one-to-many person-centric adaptation. Further research in this field is encouraged.

A limitation of our work is that the advertising effects of different adaptation strategies have only been analyzed in terms of the DOOHA field. Accordingly, future research should evaluate the PERSIT matrix in a broader advertising context. Furthermore, as the PERSIT matrix deals with adaptation strategies generally, its findings could be transferred to contexts other than advertising. For instance, applying the matrix to product recommendation systems is also possible. A respective evaluation is missing, and research in this direction is strongly encouraged.