When Personalization Becomes Intrusion: Explaining the Personalization

DOI:https://doi-001.org/1025/17866321794169

Dr.Bessas Hocine ^a, Dr.Bouchenaf Safia ^a, Dr.Mebarkia Razika

hocinebess@hotmail.com University of sétif-ALGERIA.

Safia88setif@hotmail.fr University of sétif-ALGERIA.

Mebarkiarazika2212@gmail.com University of sétif-ALGERIA.

Received: 30/08/2025 Accepted: 20/03/2026 Published: 20/08/2026

Abstract

Artificial intelligence (AI) is transforming marketing personalization by enabling firms to process increasingly complex customer information and adapt marketing interactions to individual preferences, behaviors, and contexts. These capabilities can improve relevance, usefulness, convenience, and decision efficiency, but they can simultaneously intensify consumers’ perceptions of privacy intrusion. This study examines the personalization–privacy tension through a conceptual integrative literature review. Rather than treating personalization and privacy as mutually exclusive outcomes, the study synthesizes theoretical and empirical literature on AI-based personalization, consumer value, privacy, trust, perceived control, transparency, psychological reactance, and consumer resistance. Drawing on Privacy Calculus Theory, Psychological Reactance Theory, and the Stimulus–Organism–Response perspective, the study develops an integrative theoretical framework in which AI-based personalization generates a dual consumer evaluation involving perceived personalization value and perceived privacy intrusion. The framework further proposes that this evaluation is context-dependent and influenced by data type and sensitivity, personalization intensity, transparency, perceived control, and consumer expectations. Depending on the relative salience of perceived personalization value and perceived privacy intrusion, consumers may respond through acceptance or resistance, with potential consequences for trust, engagement, and relationship continuation. The principal contribution of the study is not to rediscover the personalization–privacy paradox, which is increasingly established in the literature, but to provide a theoretical explanation of how and under what contextual conditions this tension may be resolved toward perceived consumer value or perceived privacy intrusion. The study contributes to the emerging literature on AI-driven marketing by integrating fragmented theoretical perspectives and providing a theoretically grounded basis for future empirical investigation and consumer-centered AI personalization strategies.

Keywords

Artificial intelligence; AI-driven marketing; AI personalization; personalization–privacy tension; personalization–privacy paradox; perceived personalization value; perceived privacy intrusion; privacy calculus; psychological reactance; consumer behavior

Introduction

Artificial intelligence (AI) is fundamentally reshaping contemporary marketing by enabling firms to collect, process, and interpret increasingly large volumes of customer data and translate these insights into highly individualized interactions. Among the most consequential applications of AI in marketing is algorithmic personalization, through which recommendation systems, targeted advertising, predictive models, and other AI-enabled applications adapt content, offers, and interactions to individual consumers. The strategic appeal of such personalization lies in its capacity to increase relevance, convenience, and decision efficiency. At the same time, increasingly individualized marketing depends on the collection and processing of personal and behavioral information, creating a fundamental tension between the benefits of personalization and consumers’ concerns about how their data are obtained, inferred, and used (Aguirre et al., 2015; Awad & Krishnan, 2006). Recent research has further established the personalization–privacy paradox as an important issue in contemporary digital and AI-enabled marketing environments (Ameen et al., 2022; Hsu et al., 2026).

The tension is particularly important because AI changes not only the scale of personalization but also its depth and sophistication. AI systems can combine heterogeneous behavioral signals, infer preferences, predict future actions, and dynamically adapt interactions. Such capabilities can make marketing encounters appear highly relevant and useful, but they can also create perceptions that firms know more about consumers than consumers expect or intend to disclose. Recent empirical research demonstrates that consumer responses to AI personalization cannot be reduced to a uniformly positive or negative effect. For example, experimental evidence indicates that the informational environment surrounding potential privacy risks can influence consumers’ responses to AI-personalized options (Erlei et al., 2026).

The broader personalization literature provides further evidence that these effects are not uniform. Recent meta-analytic evidence indicates that the effectiveness of personalization varies according to characteristics such as the type of consumer data involved and the level of personalization provided (Eisend et al., 2026). This finding is particularly important because it suggests that the personalization–privacy relationship cannot be adequately represented as a simple linear process in which increasing personalization necessarily produces either greater value or greater privacy concern. Instead, the consequences of personalization depend on how consumers evaluate the information practices underlying the personalized interaction and on the context in which personalization is experienced.

Recent scholarship has made substantial progress in explaining this phenomenon. Hsu et al. (2026), for example, conceptualize the personalization–privacy paradox as a persistent tension between personalization and privacy and highlight the need to understand how this tension can be managed rather than simply treated as a dilemma to be eliminated. Other recent work has explicitly examined AI-based personalization and consumer trust, developing conceptual explanations of how personalization benefits may coexist with concerns about privacy, data use, and algorithmic control (Şen, 2026). Similarly, Vu et al. (2026) propose a Utility–Control–Communication–Governance framework for understanding how consumers navigate trust and the personalization–privacy paradox in AI-driven e-commerce, emphasizing the roles of utility, consumer control, communication, and governance.

These developments demonstrate that the field is no longer characterized by a simple absence of research. Rather, research has expanded across several partially overlapping streams, including personalization effectiveness, privacy calculus, consumer trust, perceived control, transparency, algorithmic decision-making, and resistance to AI-enabled marketing. The resulting literature provides important explanations for particular components of the phenomenon, but its theoretical foundations remain distributed across different perspectives and levels of analysis. Consequently, an important question remains: how can these complementary perspectives be integrated to explain how consumers resolve the tension between the value generated by AI-based personalization and the privacy intrusion that may accompany it?

This question is important because the same personalization practice may be interpreted differently by different consumers or even by the same consumer in different contexts. An individualized recommendation may be interpreted as evidence that a firm understands the consumer’s needs and therefore provides additional value. Yet the same recommendation may alternatively signal extensive monitoring or the use of personal information in ways that exceed the consumer’s expectations. The critical issue is therefore not simply whether AI personalization generates benefits or privacy concerns, but under what conditions consumers may interpret the same personalization practice primarily as valuable or as intrusive. Recent empirical research supports the relevance of this contextual perspective by showing that privacy-related responses to AI personalization can depend on the nature and communication of data-related risks (Erlei et al., 2026).

From a theoretical perspective, this unresolved issue calls for an integration of complementary explanations. Privacy Calculus Theory provides a useful foundation for understanding how consumers may weigh anticipated benefits against perceived privacy costs when deciding whether to disclose information or accept personalized services (Beke et al., 2022). Psychological Reactance Theory complements this perspective by explaining how perceptions of excessive monitoring, intrusion, or reduced autonomy may generate resistance (Chen et al., 2019; Chen et al., 2022). The Stimulus–Organism–Response (S-O-R) perspective, in turn, provides a broader structure for connecting characteristics of AI-enabled marketing stimuli with consumers’ internal evaluations and subsequent behavioral responses (Mehrabian & Russell, 1974). The value of combining these perspectives is therefore not to claim a novel theoretical combination in itself, but to use their complementary explanatory mechanisms to organize fragmented findings around the central personalization–privacy tension.

Accordingly, the present study does not seek to establish that the personalization–privacy paradox exists, nor does it propose another direct personalization–privacy–trust model. Instead, it develops an integrative theoretical explanation of the conditions and mechanisms through which AI-based personalization may be interpreted as perceived consumer value or as perceived privacy intrusion. The central argument is that AI personalization can generate a dual evaluative possibility: it can increase perceived relevance, usefulness, and convenience while simultaneously increasing perceptions of surveillance, intrusion, or loss of control. The resolution of this tension is theoretically expected to depend on contextual characteristics, including the type and sensitivity of data used, the level of personalization, transparency regarding data practices, perceived consumer control, and consumers’ expectations concerning the personalization process.

On this basis, the study proposes a conceptual framework organized around the following theoretical logic:

AI-based personalization → dual consumer evaluation (perceived personalization value ↔ perceived privacy intrusion) → contextual interpretation → acceptance/resistance → relational outcomes.

The framework conceptualizes the personalization–privacy relationship not as a binary choice between personalization and privacy, but as a context-dependent evaluative tension whose resolution may influence consumer acceptance, resistance, trust, engagement, and continuation of the relationship. This positioning is particularly relevant in light of recent research showing that consumer trust in AI-enabled commerce is shaped by factors including utility, control, communication, and governance (Vu et al., 2026).

The study therefore pursues three related objectives. First, it synthesizes fragmented literature on AI-based personalization, consumer privacy, perceived value, trust, and resistance. Second, it identifies the theoretical mechanisms through which personalization may generate simultaneously beneficial and intrusive perceptions. Third, it develops theoretical propositions specifying the contextual conditions under which the evaluation of AI personalization may shift toward perceived value or perceived privacy intrusion. The intended contribution is consequently one of theoretical reconciliation and integration, rather than the introduction of a new construct or the claim of a previously unidentified paradox.

Methodologically, the study adopts a conceptual integrative literature review. Rather than collecting primary survey or experimental data, it critically synthesizes theoretical, conceptual, and empirical research to construct an integrated explanation of the phenomenon. This approach is particularly appropriate for a research domain characterized by rapidly evolving AI personalization practices, privacy concerns, and consumer responses (Torraco, 2005; Whittemore & Knafl, 2005). The resulting framework is intended to provide a theoretically grounded basis for future empirical investigation while also offering marketers a more nuanced understanding of when personalization may enhance consumer value and when it may be perceived as intrusive.

The remainder of the article is organized as follows. The next section presents the methodology of the integrative literature review. Section 3 establishes the theoretical foundations through Privacy Calculus Theory, Psychological Reactance Theory, and the S-O-R perspective. Section 4 critically synthesizes the literature on AI-based personalization and the personalization–privacy tension. Section 5 develops the proposed integrative conceptual framework and theoretical propositions. Section 6 discusses the theoretical and managerial contributions, limitations, and avenues for future research. The final section concludes the study.

2. Research Methodology

2.1. Research Design and Review Approach

This study adopts a conceptual integrative literature review approach. This design is appropriate for a research problem that draws on multiple theoretical and empirical streams, including artificial intelligence, marketing personalization, consumer privacy, perceived value, trust, transparency, perceived control, psychological reactance, and consumer resistance. An integrative review enables the critical synthesis of evidence from diverse research traditions and supports the development of an integrated theoretical perspective (Torraco, 2005; Whittemore & Knafl, 2005).

The objective of the review is not to estimate an aggregate statistical effect or to claim an exhaustive systematic review of the entire literature. Rather, the study seeks to identify, compare, and integrate relevant theoretical and empirical contributions in order to explain how consumers evaluate the tension between the benefits of AI-based personalization and the potential privacy intrusion associated with data-driven personalization.

Accordingly, the review follows a theory-building orientation consisting of four interconnected stages: literature identification, evidence organization, critical synthesis, and theoretical integration.

2.2. Literature Search Strategy

The literature search focused on academic publications addressing AI-based personalization, consumer privacy, personalization value, consumer responses, and the theoretical mechanisms underlying the personalization–privacy tension. Searches were conducted using combinations of keywords and Boolean operators such as “AND” and “OR”.

The principal search concepts included combinations of the following terms:

“artificial intelligence” OR “AI”;

“AI marketing” OR “AI-driven marketing”;

“AI personalization” OR “AI-based personalization” OR “algorithmic personalization”;

“personalization” OR “personalisation”;

“personalization–privacy paradox” OR “personalisation-privacy paradox”;

“privacy” OR “privacy concern” OR “privacy intrusion”;

“perceived value” OR “personalization value”;

“privacy calculus” OR “privacy calculus theory”;

“psychological reactance” OR “reactance”;

“consumer trust” OR “trust in AI”;

“transparency” OR “algorithmic transparency”;

“perceived control” OR “consumer control”;

“consumer resistance” OR “acceptance”;

“consumer engagement” OR “relationship outcomes”.

The search strategy primarily targeted peer-reviewed academic literature available through established scholarly databases and academic publishing platforms, including Scopus, Web of Science, and ScienceDirect. Relevant publications were also identified through backward and forward citation searching of highly relevant studies, particularly foundational theoretical contributions, recent reviews, meta-analyses, and studies directly addressing AI-based personalization and privacy.

Particular attention was given to recent research because of the rapid development of AI-enabled personalization. However, foundational theoretical studies were retained when they provided concepts or mechanisms directly relevant to the proposed framework.

2.3. Eligibility and Selection Criteria

The literature was selected according to its relevance to the research question and its contribution to the theoretical development of the study.

Studies were considered relevant when they met one or more of the following criteria:

1. They examined artificial intelligence, personalization, privacy, consumer value, trust, transparency, perceived control, resistance, or closely related concepts.

2. They provided a theoretical, conceptual, empirical, review-based, or meta-analytic contribution relevant to the personalization–privacy relationship.

3. They examined consumer responses to personalized or algorithmically mediated marketing interactions.

4. They contributed evidence concerning the benefits, costs, mechanisms, boundary conditions, or consequences of personalization.

5. They provided theoretical foundations relevant to Privacy Calculus Theory, Psychological Reactance Theory, or the Stimulus–Organism–Response perspective.

Studies were excluded when they were primarily commercial or promotional, lacked sufficient academic relevance, duplicated another identified contribution, or addressed artificial intelligence, personalization, or privacy in contexts that were insufficiently connected to the central research question.

The selection process was therefore based primarily on conceptual relevance and evidential contribution rather than on publication year alone.

2.4. Evidence Extraction and Mapping

For each relevant publication, information was organized according to a common analytical structure. The extraction process focused on the following dimensions:

Author and publication year;

Research context;

Principal concepts;

Theoretical perspective;

Methodological approach, where applicable;

Principal findings;

Identified limitations;

Relevance to the personalization–privacy tension;

Relevance to the proposed conceptual framework.

The resulting evidence was organized into thematic categories corresponding to the main mechanisms examined in the study:

AI-based personalization and personalization effectiveness;
Perceived personalization value;
Privacy concerns and perceived privacy intrusion;
Privacy calculus and benefit–cost evaluation;
Psychological reactance and threats to autonomy;
Consumer trust;
Transparency and perceived control;
Consumer acceptance and resistance;
Contextual characteristics of personalization and data practices.

This evidence-mapping procedure was used to establish traceability between the reviewed literature and the theoretical arguments developed in the present study. It also enabled the identification of areas of convergence, divergence, and theoretical complementarity across the reviewed literature.

2.5. Critical Synthesis Procedure

The analysis did not consist of a descriptive summary of individual studies. Instead, the literature was critically compared to identify areas of convergence, divergence, and theoretical complementarity.

The synthesis followed a structured analytical sequence: what is established in the literature; where the literature converges; where findings diverge; what mechanisms may explain these differences; what contextual conditions may influence these mechanisms; and what form of theoretical integration may account for the observed patterns.

Particular attention was given to apparently divergent findings concerning the consequences of personalization. Rather than treating differences across studies as simple inconsistencies, the analysis examined whether they could be theoretically associated with contextual characteristics such as data type and sensitivity, personalization intensity, transparency, perceived control, and consumer expectations.

This approach allows the review to move beyond a fragmented collection of findings toward an integrated explanation of why the same AI-based personalization practice may generate different consumer evaluations.

2.6. Theoretical Integration

The final stage of the review consisted of integrating the identified evidence through three complementary theoretical perspectives.

Privacy Calculus Theory provides the basis for understanding how consumers may evaluate the anticipated benefits of personalization against perceived privacy costs.

Psychological Reactance Theory provides an explanation of how perceived intrusion may become resistance when consumers experience personalization as a threat to autonomy, control, or freedom of choice.

The Stimulus–Organism–Response (S-O-R) perspective provides an overarching structure for connecting characteristics of AI-based personalization with consumers’ internal evaluations and subsequent behavioral or relational responses.

These perspectives are not presented as a novel theoretical combination in themselves. Instead, they are employed as complementary explanatory lenses through which fragmented findings can be organized around the central personalization–privacy tension.

The resulting theoretical synthesis conceptualizes AI-based personalization as capable of generating two simultaneous consumer evaluations:

Perceived Personalization Value ↔ Perceived Privacy Intrusion

These evaluations are considered within a contextual environment characterized by factors such as data type and sensitivity, personalization intensity, transparency, perceived control, and consumer expectations.

2.7. Development of Theoretical Propositions

The theoretical propositions were developed through the synthesis of multiple streams of literature rather than from primary empirical data.

Each proposition represents a theoretically reasoned relationship derived from the converging and complementary evidence identified in the reviewed literature. The propositions address five principal questions:

Why can AI-based personalization generate perceived consumer value?
Why can the same personalization practice generate perceived privacy intrusion?
Why may consumers interpret the same personalization practice differently across contexts?
Which contextual conditions may influence this interpretation?
How may the resulting evaluation contribute to consumer acceptance or resistance?

The propositions therefore represent theoretically derived extensions of the reviewed literature. They are intended to specify relationships that can subsequently be examined and refined through empirical research rather than to establish empirically validated causal effects.

2.8. Rigor, Transparency, and Traceability

To strengthen methodological rigor, the review distinguishes between evidence directly supported by previous research and theoretical relationships proposed through the synthesis conducted in the present study.

The study does not claim to be a systematic review or meta-analysis, and it does not report numerical screening outcomes or statistical estimates that were not directly generated through a documented systematic procedure. This distinction is important because the purpose of the study is conceptual integration and theory development rather than statistical aggregation.

The analytical process nevertheless follows explicit principles of literature identification, relevance assessment, evidence extraction, critical comparison, thematic organization, and theoretical synthesis. These procedures are consistent with established methodological guidance for integrative literature reviews (Torraco, 2005; Whittemore & Knafl, 2005).

The resulting framework is therefore grounded in an identifiable body of theoretical and empirical evidence while clearly distinguishing established findings from the theoretical propositions advanced by the present study.

3. Theoretical Foundations

3.1. Privacy Calculus Theory: Understanding the Value–Privacy Trade-off

Privacy Calculus Theory provides the first theoretical foundation for explaining why consumers may accept certain forms of personalized marketing despite concerns regarding the collection and use of their personal information. The central premise of the privacy calculus perspective is that consumers evaluate the anticipated benefits and costs associated with information disclosure and data use. Privacy-related decisions are therefore not necessarily determined by a desire to protect personal information under all circumstances; rather, consumers may accept information practices when the perceived benefits are considered sufficiently valuable relative to the perceived risks or costs (Beke et al., 2022).

Applied to AI-driven personalization, this perspective is particularly relevant because personalization can generate tangible consumer benefits. AI systems may use customer information to improve recommendation relevance, reduce search effort, anticipate preferences, and adapt marketing communication to individual needs. These benefits can increase the perceived value of allowing firms to use personal or behavioral information. At the same time, however, the same information practices may generate perceived privacy costs when consumers believe that the amount, sensitivity, or use of their data exceeds what they consider appropriate.

The privacy calculus perspective therefore helps explain why personalization and privacy should not necessarily be treated as mutually exclusive outcomes. A consumer may simultaneously recognize the usefulness of a personalized recommendation and experience concern regarding the information required to produce it. The resulting evaluation depends on the perceived balance between anticipated benefits and perceived privacy costs.

This logic is particularly important for the present study because it provides the theoretical basis for conceptualizing the central dual evaluation:

Perceived Personalization Value ↔ Perceived Privacy Intrusion

However, Privacy Calculus Theory alone does not fully explain what happens when consumers interpret personalization as an intrusion or threat to their autonomy. For this reason, it is complemented by a perspective capable of explaining resistance and negative psychological responses.

3.2. Psychological Reactance Theory: From Perceived Intrusion to Resistance

Psychological Reactance Theory provides the second theoretical foundation of the proposed framework. The theory explains how individuals may react when they perceive that their freedom of choice or behavioral autonomy is being threatened. In marketing environments, highly personalized communication may become problematic when consumers perceive that firms are monitoring them excessively, inferring information they did not intentionally provide, or attempting to influence their decisions in an overly intrusive manner.

This mechanism has been documented in research on personalized advertising. Chen et al. (2019), for example, identify consumer reactance as an important inhibiting mechanism in the personalization paradox and show that privacy concerns can contribute to negative reactions to online personalized advertising.

More recent research further demonstrates the relevance of reactance to personalization–privacy relationships. Chen, Sun, and Liu (2022) found that privacy concerns can alter the mechanism through which web personalization affects consumer loyalty: when privacy concerns are relatively low, personalization can strengthen trust and loyalty, whereas under higher privacy concerns, psychological reactance becomes an important mechanism linking personalization to relational outcomes.

The relevance of this perspective has also been demonstrated in recent AI-related research. Recent research explicitly combines the S-O-R framework and Psychological Reactance Theory to explain how AI-driven personalization can generate perceived threats and subsequent consumer responses. Similarly, research applying S-O-R and reactance theory to personalized advertising indicates that perceived privacy intrusiveness can trigger psychological reactance, irritation, and skepticism, with transparency influencing these relationships.

For the present framework, Psychological Reactance Theory therefore explains the transition from perceived privacy intrusion to consumer resistance. It does not imply that personalization automatically creates reactance. Rather, reactance becomes theoretically relevant when personalization is interpreted as threatening autonomy, privacy, or freedom of choice.

The theoretical sequence can consequently be represented as:

Perceived Privacy Intrusion → Perceived Threat to Autonomy → Psychological Reactance → Resistance

This mechanism complements Privacy Calculus Theory by explaining the negative-response pathway that may emerge when perceived privacy costs or intrusion become sufficiently salient to threaten the consumer’s sense of autonomy or control.

3.3. Stimulus–Organism–Response: Structuring the Consumer-Side Mechanism

The third theoretical perspective is the Stimulus–Organism–Response (S-O-R) framework. Originally developed by Mehrabian and Russell (1974), the S-O-R perspective conceptualizes behavioral responses as emerging from the interaction between environmental stimuli and individuals’ internal psychological states.

Applied to AI-driven marketing, characteristics of AI personalization can be conceptualized as stimuli, including the degree of personalization, recommendation relevance, data-driven targeting, interactional characteristics, transparency, and perceived intrusiveness.

The organism corresponds to the consumer’s internal evaluation of these stimuli. In the present study, this level primarily includes perceived personalization value and perceived privacy intrusion, while trust, perceived control, satisfaction, and related psychological states may represent additional internal evaluative or relational mechanisms depending on the specific personalization context.

The response represents the resulting behavioral or relational reaction, which may include acceptance, continued use, engagement, resistance, avoidance, or relationship continuation.

Recent empirical research supports the applicability of the S-O-R perspective to AI personalization. For example, research in AI-driven tourism has examined personalization as a stimulus and perceived value, trust, and privacy concern as organismic states that subsequently influence behavioral intentions.

The S-O-R perspective therefore provides an organizing structure for the present framework:

AI Personalization Stimuli → Consumer Evaluation → Behavioral/Relational Response

Importantly, S-O-R is not treated as the source of the study’s novelty. Its role is structural: it allows the different theoretical mechanisms to be located within a coherent consumer-response sequence.

3.4. Complementarity of the Three Theoretical Perspectives

The three theoretical perspectives employed in this study are complementary rather than competing explanations.

Privacy Calculus Theory addresses the question:

How does the consumer evaluate the benefits and privacy costs associated with personalization?

It therefore provides the foundation for understanding the value–privacy evaluation.

Psychological Reactance Theory addresses the question:

What happens when personalization is interpreted as excessive intrusion or a threat to autonomy?

It explains the psychological mechanism through which perceived intrusion may contribute to resistance.

S-O-R addresses the question:

How can characteristics of AI personalization be connected to internal consumer evaluations and subsequent responses?

It provides the broader structure within which the two psychological mechanisms can be organized.

The resulting theoretical articulation can therefore be represented as:

AI-Based Personalization

↓

Dual Consumer Evaluation

↙　　　　　　　　　↘

Perceived Personalization Value　Perceived Privacy Intrusion

↓　　　　　　　　　↓

Favorable Evaluation　Perceived Threat to Autonomy

↓　　　　　　　　　↓

Acceptance　　　　　 Psychological Reactance

↓

Resistance

↘　　　　　　　　　↙

Consumer and Relational Outcomes

This articulation should not be interpreted as a claim that the combination of these theories is unprecedented. Previous research has already applied combinations of reactance, S-O-R, privacy concerns, and personalization to related contexts.

The contribution of the present study instead lies in using these perspectives as complementary lenses to address a broader theoretical question: under what conditions does AI-based personalization become interpreted primarily as value, and under what conditions does it become interpreted as intrusion?

3.5. Theoretical Positioning of the Proposed Framework

Based on the preceding discussion, the present study conceptualizes AI-based personalization as a potentially ambivalent marketing stimulus. Its effects cannot be adequately understood through a simple assumption that personalization either benefits or harms consumers.

Instead, personalization may simultaneously activate two evaluative pathways.

The first is a value pathway:

AI Personalization → Perceived Relevance/Usefulness → Perceived Personalization Value → Acceptance

The second is a privacy-intrusion pathway:

AI Personalization → Perceived Privacy Intrusion → Perceived Threat to Autonomy/Privacy → Psychological Reactance → Resistance

Privacy Calculus Theory explains the evaluative relationship between anticipated benefits and perceived privacy costs, while Psychological Reactance Theory explains the negative psychological mechanism that may emerge when privacy-related concerns become interpreted as threats to autonomy. S-O-R provides the overarching structure connecting personalization characteristics, consumer evaluations, and behavioral responses.

The framework therefore moves beyond a simplified representation such as:

AI Personalization → Consumer Response

And instead proposes:

AI Personalization → Dual Consumer Evaluation → Contextual Resolution → Acceptance/Resistance → Relational Outcomes

The central theoretical issue is consequently the resolution of the personalization–privacy tension. Consumers do not necessarily choose between personalization and privacy in an absolute sense. Rather, they evaluate personalized interactions in light of their perceived benefits, the sensitivity and appropriateness of the underlying data practices, the level of personalization, the degree of transparency and control, and their existing expectations regarding the relationship with the firm.

This theoretical positioning provides the foundation for the critical synthesis and theoretical propositions developed in the following sections.

4. Critical Synthesis of the Literature

4.1. From Personalization Benefits to the Personalization–Privacy Tension

The literature on personalization has traditionallyemphasizeditsability to improve the relevance and effectiveness of marketing communication. By adapting messages, recommendations, offers, and content to individualcharacteristics or behavioral patterns, personalizationcanreducesearchcosts, improvedecisionefficiency, and increase the perceivedusefulness of marketing interactions (Aguirre et al., 2015; Ameen et al., 2022; Cloarec, 2020). Recentevidenceconfirmsthatpersonalizationcanenhance marketing effectivenesswhilealsodemonstratingthatitseffects are not uniform. A large meta-analysispublished in the Journal of Marketing in 2026 synthesized 1,536 effect sizes from 290 studies and foundthatpersonalizationeffectsvaryaccording to characteristicssuch as the type of consumer data used and the level of personalization (Eisend et al., 2026).

The emergence of AI intensifies this duality. AI-based systems can process heterogeneous data, infer consumer preferences, identify behavioral patterns, and dynamically adapt interactions. Consequently, personalization may become more accurate and context-sensitive, while the same capabilities may increase consumers’ awareness of the amount of information that firms possess or infer about them (Şen, 2026).

This creates what the literature commonly describes as the personalization–privacy paradox: consumers may desire relevant and individualized experiences while simultaneously expressing concerns about the collection and use of the data required to generate those experiences (Awad & Krishnan, 2006; Ameen et al., 2022; Hsu et al., 2026). Earlier research established this tension in digital personalization, while recent studies have increasingly examined its implications in AI-enabled environments.

The central implication is that personalization should not be conceptualized as an inherently beneficial technological capability. Its consequences depend on how consumers interpret the relationship between the benefits generated by personalization and the privacy implications of the underlying data practices.

4.2. AI-Based Personalization as a Source of Consumer Value

The value pathway represents the positive side of the personalization tension. AI can transform customer data into recommendations, offers, content, and interactions that are more closely aligned with individual needs. From the consumer’s perspective, such personalization may generate relevance, convenience, reduced effort, and improved decision efficiency (Aguirre et al., 2015; Şen, 2026).

The 2026 meta-analysis by Eisend, Niewiadomska, and van Noort provides particularly important evidence in this regard. Rather than assuming a uniform effect of personalization, the study demonstrates systematic variation in personalization effects according to personalization characteristics and data-related conditions (Eisend et al., 2026).

This finding is theoretically important because it shifts attention from the question of whether personalization is beneficial to the question of when personalization is perceived as beneficial. A recommendation may be valuable when it appears relevant, timely, and appropriate to the consumer’s needs. The same recommendation may generate a different reaction when the consumer cannot understand why a particular inference was made or perceives the underlying data use as excessive (Awad & Krishnan, 2006; Ham & Lee, 2026).

Thus, perceived personalization value should be understood as an evaluative outcome rather than as an automatic consequence of increasing personalization intensity.

4.3. AI-Based Personalization as a Source of Privacy Intrusion

The second side of the tension concerns the potential costs associated with personalization. AI systems can infer preferences and characteristics that consumers may not have explicitly disclosed. As personalization becomes more sophisticated, consumers may therefore perceive that firms possess an unusually detailed understanding of their behavior (Şen, 2026).

Privacyconcernsmay arise fromseveral sources, including the sensitivity of the data involved, the unexpected use of information, uncertaintyregarding how information isprocessed, and perceivedlack of control over data practices (Awad&Krishnan, 2006; Al Helaly et al., 2025; Beke et al., 2022). Importantly, privacyconcerns do not necessarilyrequire an actual violation of privacy. The perception thatpersonalizationinvolves excessive or unexpected data use mayitself influence consumer responses.

Recent empirical research reinforces this contextual interpretation. The 2026 CHI study by Erlei et al. Demonstrates that the informational environment surrounding potential data-leak risks can affect responses to AI personalization, indicating that consumer reactions depend not only on the existence of a privacy risk but also on how that risk is communicated and interpreted (Erlei et al., 2026).

Accordingly, the privacy side of the paradox should not be reduced to objective data protection alone. It also involves perceived privacy intrusion, which represents the consumer’s subjective interpretation of the appropriateness and extent of data use.

4.4. Why the Literature Produces Divergent Outcomes

A central finding emerging from the literature is that personalization does not generate a single predictable consumer response. The same technological capability can produce positive outcomes in one context and less favorable outcomes in another.

The recent meta-analysis by Eisend et al. Is particularly relevant because it demonstrates systematic variation in personalization effects rather than treating differences across studies as simple random inconsistency (Eisend et al., 2026). This suggests that apparently divergent findings may reflect differences in the characteristics of personalization and the information environment surrounding it.

Several factors appear particularly important.

First, data type and sensitivity may influence consumer reactions. Personalization based on relatively ordinary behavioral information may be perceived differently from personalization involving highly sensitive or unexpected information (Eisend et al., 2026; Beke et al., 2022).

Second, personalization intensity may affect perceptions of relevance and intrusion simultaneously. Increasing personalization can improve the usefulness of an interaction while also making the extent of consumer monitoring more salient (Eisend et al., 2026; Ham & Lee, 2026).

Third, transparency can influence how consumers interpret AI personalization. Recent research specifically examines algorithmic transparency as a potential mechanism for addressing the personalization–privacy paradox (Ham & Lee, 2026).

Fourth, perceived control may determine whether consumers regard personalization as an acceptable exchange or as an imposed practice. Recent work on AI-driven e-commerce identifies utility, control, communication, and governance as important components shaping how consumers navigate trust and the personalization–privacy paradox (Vu et al., 2026).

Finally, consumer expectations matter. A personalized recommendation that corresponds to what a consumer reasonably expects from a platform may be interpreted as useful, whereas an unexpectedly precise recommendation may signal extensive monitoring or data use.

These factors suggest that the personalization–privacy tension is inherently contextual rather than purely technological.

4.5. From Privacy Calculus to Dual Consumer Evaluation

The preceding literature can be synthesized through the logic of Privacy Calculus Theory. Consumers may evaluate personalization according to the benefits they expect to receive and the costs they associate with providing or allowing access to personal information (Beke et al., 2022).

However, the AI context requires a broader interpretation of this calculus. The consumer is not simply deciding whether to disclose information. The consumer may also evaluate the output of the personalization process itself. A recommendation can reveal to the consumer that the system has inferred information about them, thereby making the underlying data practices more salient.

Consequently, personalization can generate two simultaneous evaluations:

Perceived Personalization Value

And

Perceived Privacy Intrusion

These evaluations should not be assumed to be perfectly opposite or mutually exclusive. A consumer may perceive a recommendation as highly useful while also feeling uncomfortable about how accurately the system inferred the underlying preference (Aguirre et al., 2015; Beke et al., 2022).

The theoretical implication is that the personalization–privacy paradox is better understood as a dual evaluation problem than as a simple choice between accepting personalization and protecting privacy.

4.6. From Perceived Intrusion to Psychological Reactance

The literature on psychological reactance provides an additional explanation for the negative side of this dual evaluation. When consumers perceive personalization as excessive, intrusive, or threatening to their autonomy, privacy concern may develop into a stronger psychological response.

Previous research has demonstrated that personalization can generate reactance when consumers perceive a loss of control or excessive intrusion (Chen et al., 2019; Chen et al., 2022). These findings reinforce the relevance of autonomy, perceived threat, and control in explaining negative consumer responses to personalized marketing.

This distinction is theoretically important. Privacy concern does not necessarily imply rejection. Consumers may remain concerned about data use while continuing to use a personalized service when the perceived benefits remain sufficiently high. Resistance becomes more likely when privacy-related concerns are interpreted as a meaningful threat to autonomy, control, or freedom of choice.

The resulting mechanism can therefore be expressed as:

Perceived Privacy Intrusion → Perceived Threat → Psychological Reactance → Resistance

This mechanism complements the benefit–cost evaluation described by Privacy Calculus Theory by explaining how privacy-related concerns may develop into a more active form of consumer resistance.

4.7. The Role of Transparency, Control, and Communication

Recent research increasingly emphasizes that the personalization–privacy tension is not determined exclusively by the technology or the quantity of data collected. The manner in which data practices are communicated and governed also matters.

Ham and Lee (2026), for example, examine algorithmic transparency and algorithm literacy as potential mechanisms for addressing the personalization–privacy paradox. Their findings indicate that algorithmic transparency can play an important role in reducing privacy-related concerns associated with highly personalized advertising. Similarly, Vu et al. (2026) identify utility, control, communication, and governance as important components of a consumer pathway framework for AI-driven e-commerce.

These findings suggest that transparency and control should not be conceptualized simply as independent predictors of trust. Rather, they may influence how consumers interpret the personalization process itself.

When consumers understand why personalization occurs and retain meaningful control over relevant data practices, the same level of personalization may be interpreted as more legitimate and valuable. Conversely, limited transparency or perceived lack of control may increase the likelihood that personalization is interpreted as intrusive (Ham & Lee, 2026; Vu et al., 2026).

Thus, transparency and control are better positioned in the present framework as contextual conditions that may influence the resolution of the value–intrusion tension.

4.8. Theoretical Fragmentation in the Existing Literature

The literature reviewed reveals substantial theoretical progress but also considerable fragmentation.

One stream emphasizes personalization effectiveness, focusing on relevance, persuasion, engagement, and consumer value. Another emphasizes privacy, examining privacy concerns, information disclosure, and perceived risk. A third focuses on trust, while another examines reactance and resistance. More recent research has begun integrating some of these perspectives into AI-specific frameworks.

For example, Şen (2026) develops a conceptual framework connecting AI personalization and consumer trust within the personalization–privacy paradox. Vu et al. (2026) develop a consumer pathway framework centered on utility, control, communication, and governance. Ham and Lee (2026) examine algorithmic transparency and algorithm literacy as potential mechanisms for addressing the paradox.

These contributions significantly advance the field. They also demonstrate that the research gap can no longer be defined as an absence of conceptual explanations.

The remaining issue is instead one of theoretical reconciliation. Different studies illuminate different portions of the phenomenon, but an integrated explanation is still needed to clarify how these mechanisms may interact in determining whether consumers interpret AI personalization primarily as a source of value or as a source of intrusion.

4.9. Toward a Resolution-Based Interpretation of the Personalization–Privacy Tension

The synthesis therefore leads to a resolution-based interpretation of the personalization–privacy paradox. Rather than conceptualizing it as a fixed contradiction between two incompatible consumer preferences, the paradox can be understood as a context-dependent evaluative tension.

AI-based personalization generates potential benefits while simultaneously making data practices more salient. Consumers evaluate these two dimensions through their perceptions of relevance, usefulness, privacy, control, transparency, and appropriateness. The resulting interpretation may favor perceived value, perceived intrusion, or a coexistence of both.

The proposed theoretical logic can therefore be summarized as:

AI-Based Personalization

↓

Dual Consumer Evaluation

Perceived Personalization Value ↔ Perceived Privacy Intrusion

↓

Contextual Resolution

Data Type – Personalization Level – Transparency – Perceived Control – Consumer Expectations

↓

Acceptance / Resistance

↓

Relational Outcomes

This formulation does not claim that the underlying paradox is newly discovered. Instead, it provides a synthesis of the literature around the unresolved question of how and under what conditions the tension is resolved in consumers’ evaluations.

4.10. Theoretical Gap and Positioning of the Present Study

The critical synthesis indicates that the literature has reached a stage in which the existence of the personalization–privacy tension is well established. Recent research has provided systematic reviews, conceptual frameworks, empirical studies, and meta-analytic evidence addressing different aspects of the phenomenon (Hsu et al., 2026; Eisend et al., 2026).

The remaining theoretical opportunity lies in connecting these findings into a coherent explanatory architecture. Specifically, there is a need to distinguish between:

The value generated by personalization;
The privacy intrusion perceived by consumers;
The contextual factors that shape their relative evaluation;
The psychological mechanism through which perceived intrusion may contribute to resistance; and
The relational consequences associated with acceptance or resistance.

The present study therefore positions itself as an integrative theoretical synthesis rather than as another direct model of personalization, privacy, and trust.

Its central theoretical proposition is that the consequences of AI personalization depend on the contextual resolution of a dual consumer evaluation in which perceived personalization value and perceived privacy intrusion may coexist and are interpreted within a specific data and interaction context.

This positioning provides the conceptual foundation for the integrated framework and theoretical propositions developed in the next section.

5. Proposed Conceptual Framework and Theoretical Propositions

5.1. General Logic of the Proposed Framework

The preceding literature synthesis indicates that the consequences of AI-based personalization cannot be adequately explained through a direct relationship between personalization and consumer outcomes. Personalization may simultaneously enhance the relevance and usefulness of marketing interactions while increasing consumers’ perceptions of privacy intrusion. Recent research confirms that the personalization–privacy relationship is context-dependent and that factors such as utility, control, communication, governance, and characteristics of personalization influence consumer responses (Eisend et al., 2026; Vu et al., 2026).

The present study therefore conceptualizes AI-based personalization as an ambivalent marketing stimulus capable of generating two simultaneous consumer evaluations: perceived personalization value and perceived privacy intrusion.

The proposed theoretical logic is:

AI-Based Personalization → Dual Consumer Evaluation → Contextual Resolution → Consumer Response → Relational Outcomes

The central theoretical argument is that consumers do not necessarily evaluate personalization as either beneficial or harmful from the outset. Rather, they interpret the personalized interaction according to the benefits it provides and the privacy implications they perceive. The relative salience of these evaluations may subsequently influence whether personalization is accepted or resisted.

This approach differs from simply proposing another personalization–privacy–trust relationship. Recent research has already developed conceptual frameworks directly addressing AI personalization, privacy, and trust, including the Utility–Control–Communication–Governance framework emphasizing utility, control, communication, and governance (Vu et al., 2026). The contribution of the present framework instead lies in integrating these fragmented insights around the question of how the personalization–privacy tension is resolved in consumer evaluation.

5.2. AI-Based Personalization as the Initial Stimulus

AI-based personalization refers to the use of artificial intelligence to adapt marketing content, recommendations, offers, communication, or interactions according to information concerning individual consumers.

Compared with conventional personalization, AI enables more extensive processing and inference of behavioral and contextual information. This can increase the relevance and usefulness of marketing interactions, but it can simultaneously make consumers more aware of the extent to which firms collect, infer, and utilize information about them (Aguirre et al., 2015; Şen, 2026).

Accordingly, AI-based personalization constitutes the initial stimulus in the proposed framework. Its consequences are not assumed to be inherently positive or negative. Instead, the same personalization practice may activate both a value pathway and a privacy-intrusion pathway.

5.3. Perceived Personalization Value

The first component of the dual evaluation is perceived personalization value.

Consumers may perceive AI-based personalization as valuable when it improves relevance, convenience, usefulness, decision efficiency, or the fit between a marketing interaction and their individual needs (Aguirre et al., 2015; Eisend et al., 2026).

The literature therefore supports the proposition that personalization can create consumer value, but the existence and magnitude of this value depend on how personalization is experienced rather than simply on its technological sophistication. Recent meta-analytic evidence demonstrates that personalization effects vary according to characteristics such as the type of consumer data and the level of personalization (Eisend et al., 2026).

Accordingly:

Proposition 1 (P1).

AI-based personalization can generate perceived personalization value when consumers perceive personalized interactions as relevant, useful, and consistent with their needs and expectations.

5.4. Perceived Privacy Intrusion

The second component is perceived privacy intrusion.

AI personalization may require firms to collect, combine, infer, or process information that consumers consider personal or sensitive. When such practices are perceived as excessive, unexpected, opaque, or insufficiently controlled, personalization may generate a perception of intrusion (Awad & Krishnan, 2006; Beke et al., 2022).

Importantly, perceived privacy intrusion concerns the consumer’s interpretation of data practices rather than necessarily an objectively established privacy violation. A consumer may perceive an interaction as intrusive because the degree of personalization reveals an unexpectedly detailed understanding of their behavior.

Recent experimental evidence shows that privacy-related information environments can influence consumers’ responses to AI personalization, reinforcing the importance of how data practices and privacy risks are perceived and communicated (Erlei et al., 2026).

Therefore:

Proposition 2 (P2).

AI-based personalization can generate perceived privacy intrusion when consumers perceive the collection, inference, or use of personal information as excessive, unexpected, opaque, or insufficiently controlled.

5.5. The Dual Consumer Evaluation

The central element of the framework is the coexistence of perceived personalization value and perceived privacy intrusion.

These two evaluations should not be conceptualized as perfectly opposite or mutually exclusive states. A consumer may simultaneously consider a recommendation useful and feel uncomfortable about how the system obtained or inferred the information necessary to generate it.

This distinction is important because it helps explain why the same personalization practice can produce apparently contradictory responses. The consumer may appreciate the outcome while questioning the underlying data practice.

The personalization–privacy paradox is therefore conceptualized here as a dual evaluation process rather than as a simple binary choice between personalization and privacy. This interpretation is consistent with recent research characterizing the personalization–privacy relationship as a persistent tension rather than a uniformly resolved trade-off (Hsu et al., 2026).

5.6. Contextual Resolution of the Personalization–Privacy Tension

The relative salience of perceived personalization value and perceived privacy intrusion is unlikely to be determined by personalization intensity alone. The literature identifies several contextual characteristics that may influence consumer interpretation.

First, data type and sensitivity may affect perceptions of appropriateness. Personalization based on ordinary behavioral information may be interpreted differently from personalization involving highly sensitive or unexpected information (Eisend et al., 2026; Beke et al., 2022).

Second, personalization intensity may simultaneously increase relevance and make data use more salient. Higher levels of personalization may therefore strengthen perceived value in some contexts while increasing perceptions of intrusion in others (Eisend et al., 2026).

Third, transparency may influence consumers’ understanding of why and how personalization occurs. Recent research specifically investigates algorithmic transparency as a potential mechanism for addressing the personalization–privacy paradox (Ham & Lee, 2026).

Fourth, perceived control may influence whether consumers experience personalization as an acceptable exchange or as an imposed practice. Recent research identifies control as one of the central mechanisms shaping trust and consumer responses in AI-driven e-commerce (Vu et al., 2026).

Finally, consumer expectations may determine whether personalization appears appropriate or surprising. Personalization that corresponds to consumers’ expectations may be perceived as helpful, whereas unexpectedly precise personalization may increase perceptions of surveillance or intrusion.

These considerations lead to the central contextual proposition:

Proposition 3 (P3).

The consumer evaluation of AI-based personalization is context-dependent, such that the relative salience of perceived personalization value and perceived privacy intrusion may vary across consumer, data, and interaction contexts.

5.7. Boundary Conditions of the Dual Evaluation

The preceding discussion suggests that the personalization–privacy tension is not determined solely by the presence of AI. Rather, its resolution depends on characteristics of the technological, informational, and interactional environment.

The present framework therefore identifies five principal contextual conditions:

Data type and sensitivity

→ What information is being used?

Personalization intensity

→ How precisely is the interaction tailored?

Transparency

→ Does the consumer understand how and why personalization occurs?

Perceived control

→ Does the consumer feel able to influence relevant data practices?

Consumer expectations

→ Does the personalization correspond to what the consumer reasonably expects?

These conditions are not treated as additional independent constructs merely added to increase model complexity. Instead, they are conceptualized as boundary conditions that may shape how consumers interpret the dual value–intrusion evaluation.

This positioning is consistent with recent research emphasizing transparency, control, communication, and governance in the management of personalization–privacy tensions (Ham & Lee, 2026; Vu et al., 2026).

Proposition 4 (P4).

The relative salience of perceived personalization value and perceived privacy intrusion is conditioned by contextual factors, particularly data type and sensitivity, personalization intensity, transparency, perceived control, and consumer expectations.

5.8. From Evaluation to Consumer Acceptance or Resistance

The resolution of the value–intrusion tension may influence the consumer’s subsequent response.

When personalization is perceived as relevant and beneficial while its privacy implications remain acceptable, consumers may be more willing to accept personalized interactions. Conversely, when personalization is interpreted as excessive or intrusive, consumers may experience concerns regarding autonomy and control and become more likely to resist.

Psychological Reactance Theory provides a complementary explanation for this negative pathway. Perceived intrusion can become particularly consequential when consumers interpret personalization as a threat to their autonomy or freedom of choice (Chen et al., 2019; Chen et al., 2022).

Recent research also indicates that personalization–privacy outcomes may involve resistance rather than simply trust or acceptance, reinforcing the importance of distinguishing positive and negative consumer-response pathways.

Accordingly:

Proposition 5 (P5).

When perceived personalization value is more salient than perceived privacy intrusion, consumers are more likely to exhibit acceptance of personalized interactions; when perceived privacy intrusion becomes more salient than perceived personalization value, consumers are more likely to exhibit resistance.

5.9. Relational Consequences

Consumer acceptance and resistance may subsequently influence the quality and continuity of the relationship between consumers and firms.

Positive responses may contribute to trust, engagement, continued use, and relationship continuation. Conversely, persistent perceptions of intrusion may contribute to avoidance, resistance, reduced engagement, and deterioration of the consumer–firm relationship (Chen et al., 2022).

The framework does not claim that these relational consequences have been empirically established as direct consequences of the proposed model. Rather, they represent theoretically plausible downstream outcomes that future empirical research can examine.

This distinction is essential because the present study is conceptual and does not attempt to establish causal relationships statistically.

5.10. Integrated Conceptual Framework

The complete framework can therefore be summarized as follows:

AI-Based Personalization

↓

Dual Consumer Evaluation

↙　　　　　　　　　　　↘

Perceived Personalization Value ↔ Perceived Privacy Intrusion

↓

Contextual Resolution

Data Type and Sensitivity

Personalization Intensity

Transparency

Perceived Control

Consumer Expectations

↓

Consumer Response

↙　　　　　　　　　　　↘

Acceptance　　　　　 Resistance

↓

Relational Outcomes

Trust

Engagement

Relationship Continuation

The conceptual logic can consequently be expressed as:

AI-Based Personalization → Dual Consumer Evaluation → Contextual Resolution → Acceptance/Resistance → Relational Outcomes

The principal theoretical contribution of this framework is not the identification of a new construct or the claim that the personalization–privacy paradox itself is unexplored. Recent research has already established the importance of this paradox and has proposed alternative conceptualizations of it (Hsu et al., 2026; Vu et al., 2026).

Instead, the framework seeks to provide an integrative explanation of the resolution of the tension by connecting personalization value, privacy intrusion, contextual conditions, and consumer responses within a single theoretical architecture.

5.11. Theoretical Positioning and Contribution

The proposed framework contributes to the literature in three principal ways.

First, it conceptualizes the personalization–privacy relationship as a dual consumer evaluation rather than as a simple trade-off in which consumers must choose between personalization and privacy.

Second, it identifies contextual resolution as the central explanatory mechanism through which the same personalization practice may generate different consumer interpretations.

Third, it connects previously fragmented theoretical mechanisms. Privacy Calculus Theory explains the evaluation of benefits and privacy costs; Psychological Reactance Theory explains the emergence of resistance when perceived intrusion threatens autonomy; and the S-O-R perspective provides the broader structure linking AI personalization stimuli to consumer evaluations and responses.

The framework therefore offers a theoretical synthesis that complements recent AI-personalization research rather than attempting to replace existing models. In particular, it differs from recent UCCG-based work by focusing specifically on the theoretical resolution of the value–intrusion evaluation, while recognizing utility, control, communication, and governance as relevant contextual considerations (Vu et al., 2026).

6. Discussion and Theoretical Contributions
- 6.1. General Discussion

The present study addresses the personalization–privacy tension from an integrative theoretical perspective. Rather than treating AI-based personalization as inherently beneficial or inherently intrusive, the proposed framework conceptualizes it as an ambivalent marketing stimulus capable of generating two simultaneous consumer evaluations: perceived personalization value and perceived privacy intrusion.

This perspective is particularly relevant given the rapid development of research on AI-enabled personalization. Existing studies have examined the personalization–privacy paradox from different perspectives, including consumer trust, utility, control, communication, governance, psychological reactance, privacy concerns, and personalization effectiveness (Hsu et al., 2026; Vu et al., 2026; Şen, 2026). The recent meta-analysis of personalization in marketing communication further demonstrates that personalization effects vary according to characteristics such as the type of data used and the level of personalization, indicating that its consequences cannot be reduced to a uniform positive or negative effect (Eisend et al., 2026).

The central contribution of the present framework is therefore to interpret these apparently divergent findings as manifestations of a broader context-dependent evaluation process. The same personalization practice may create value in one context and intrusion in another, or may generate both perceptions simultaneously. The theoretical question is consequently not whether personalization is beneficial or harmful, but how consumers interpret and resolve the tension between these two possibilities.

6.2. First Theoretical Contribution: FromParadox Identification to Tension Resolution

The first contribution concerns the conceptualization of the personalization–privacy paradox.

Recent research has already established the existence of the paradox and developed frameworks for understanding it. For example, recent AI-personalization research explicitly connects personalization with privacy and trust, while the UCCG framework identifies utility, control, communication, and governance as mechanisms shaping consumer trust and the navigation of the personalization–privacy paradox (Şen, 2026; Vu et al., 2026).

The present study therefore does not claim to discover the paradox. Instead, it shifts the analytical question from paradox identification to tension resolution.

The proposed logic is:

Personalization → Value and Intrusion → Contextual Evaluation → Acceptance or Resistance

This reframing provides a way to interpret apparently contradictory findings without treating them as theoretical failures. Divergent outcomes may instead reflect differences in how consumers evaluate the benefits and privacy implications of the same personalization practice.

6.3. Second Theoretical Contribution: Conceptualizing a Dual Consumer Evaluation

The second contribution is the explicit conceptualization of personalization as generating a dual consumer evaluation.

Previous research has separately examined personalization value, privacy concerns, trust, and resistance. The proposed framework brings these strands together by distinguishing between:

Perceived Personalization Value

And

Perceived Privacy Intrusion

The two evaluations are not assumed to be mutually exclusive. A consumer can perceive a recommendation as highly relevant while simultaneously questioning how the firm acquired or inferred the information necessary to generate it.

This distinction helps explain why the personalization–privacy relationship cannot be represented simply as:

More personalization → More value

Or:

More personalization → More privacy concern

Instead, the consequences depend on how the consumer interprets the interaction and its underlying information practices.

6.4. ThirdTheoretical Contribution: ContextualResolution

The third and most important contribution concerns the role of contextual conditions.

The framework proposes that the resolution of the value–intrusion tension may depend particularly on data type and sensitivity, personalization intensity, transparency, perceived control, and consumer expectations.

This contribution is consistent with recent evidence showing that personalization effects vary according to personalization and data characteristics (Eisend et al., 2026). It also complements recent AI research emphasizing control, communication, and governance in shaping consumer trust and responses to AI-driven personalization (Vu et al., 2026).

The theoretical implication is that these conditions should not simply be treated as independent predictors of consumer outcomes. Their more important role in the present framework is to influence how consumers interpret personalization in the first place.

Thus:

Context → Interpretation of Personalization → Resolution of Value/Intrusion Tension → Consumer Response

This provides a more precise explanation of why similar technological capabilities may produce different consumer reactions across contexts.

6.5. FourthTheoretical Contribution: ConnectingPrivacyCalculus and PsychologicalReactance

The fourth contribution lies in connecting two complementary psychological mechanisms.

Privacy Calculus Theory explains how consumers may evaluate the benefits and costs associated with personalization and data use (Beke et al., 2022). Psychological Reactance Theory explains how perceived intrusion may contribute to resistance when consumers experience a threat to autonomy or control (Chen et al., 2019; Chen et al., 2022).

Recent research has already applied S-O-R and Psychological Reactance Theory to personalization-related contexts, meaning that the theoretical combination itself is not claimed as novel.

The contribution instead lies in positioning these mechanisms within the broader resolution process:

Value–Privacy Evaluation → Perceived Intrusion → Threat to Autonomy → Reactance → Resistance

This allows the framework to distinguish between privacy concern and active resistance, which should not be treated as identical consumer responses.

6.6. Managerial Implications

The framework also provides several implications for marketing managers.

First, firms should not seek to maximize personalization intensity without considering how consumers interpret the underlying data practices. Greater personalization may increase relevance, but it may also increase perceptions of intrusion.

Second, firms should consider the type and sensitivity of data used for personalization. Personalization based on relatively ordinary behavioral information may be interpreted differently from personalization based on sensitive or unexpected information.

Third, transparency and perceived control should be treated as components of personalization design rather than as purely regulatory requirements. Recent AI-commerce research specifically identifies control, communication, and governance as important mechanisms shaping consumer trust and responses to AI-driven personalization (Vu et al., 2026).

Fourth, firms should align personalization practices with consumer expectations. Unexpectedly precise personalization may draw attention to the underlying data practices and increase perceptions of surveillance or intrusion.

Finally, managers should evaluate AI personalization using both positive and negative indicators. Measures such as relevance, usefulness, engagement, and satisfaction should be considered alongside privacy concern, perceived intrusion, resistance, and avoidance.

The managerial objective should therefore not be maximum personalization, but appropriate personalization.

6.7. Implications for Future Research

The framework generates several opportunities for empirical research.

First, future studies could empirically test whether perceived personalization value and perceived privacy intrusion constitute distinct but simultaneous consumer evaluations.

Second, researchers could examine whether data sensitivity, personalization intensity, transparency, perceived control, and consumer expectations function as moderators or contextual conditions.

Third, experimental research could examine whether identical AI-personalization stimuli produce different consumer responses when the data source, transparency, or control mechanism is manipulated.

Fourth, longitudinal research could investigate whether consumers’ evaluations change as they become more familiar with AI personalization.

Finally, future studies could examine whether the proposed resolution mechanism differs across sectors such as retail, banking, tourism, healthcare, and other high-sensitivity environments.

6.8. Limitations

The principal limitation of this study is its conceptual nature. The proposed relationships and propositions have not been statistically tested and should therefore be understood as theoretically derived rather than empirically established.

A second limitation concerns the rapid evolution of AI technologies. Recent publications demonstrate how quickly new forms of AI personalization, e-commerce systems, and consumer interaction models are emerging. Consequently, the theoretical framework may require refinement as AI practices and consumer expectations evolve.

Third, the framework integrates several theoretical perspectives and therefore remains relatively broad. Future empirical research will be necessary to determine which mechanisms have the strongest explanatory power in particular contexts.

Finally, the framework should not be interpreted as claiming that all consumers resolve personalization–privacy tensions in the same way. Individual differences, cultural context, sector characteristics, and regulatory environments may further influence the evaluation process.

6.9. Final TheoreticalPositioning

The central theoretical position of the study can be summarized as follows:

AI personalization does not have a predetermined consumer meaning. Its value or intrusiveness is contextually interpreted.

The proposed framework therefore moves beyond the question:

“Does AI personalization create value or privacy concerns?”

Toward the more theoretically informative question:

“Under what conditions does AI personalization become interpreted as value rather than intrusion, and how does this interpretation shape consumer responses?”

This represents the principal contribution of the study. It does not replace existing personalization–privacy frameworks, but provides an integrative theoretical architecture for interpreting their complementary and sometimes divergent findings.

7. Conclusion

Artificial intelligence is fundamentally transforming marketing personalization by enabling firms to process increasingly complex customer information and adapt marketing interactions to individual preferences, behaviors, and contexts. These capabilities can improve relevance, usefulness, convenience, and decision efficiency, but they can simultaneously intensify consumers’ perceptions of privacy intrusion. The resulting personalization–privacy tension represents a central challenge associated with increasingly intelligent and data-intensive marketing practices.

This study addressed this tension through a conceptual integrative literature review. Rather than treating personalization and privacy as mutually exclusive outcomes, the study synthesized theoretical and empirical literature to develop an integrated explanation of how consumers may interpret AI-based personalization as a source of value, a source of intrusion, or both simultaneously. The review brought together literature on AI personalization, consumer value, privacy, trust, perceived control, transparency, psychological reactance, and consumer resistance.

Drawing on Privacy Calculus Theory, Psychological Reactance Theory, and the Stimulus–Organism–Response perspective, the proposed framework conceptualizes AI-based personalization as an ambivalent marketing stimulus capable of generating a dual consumer evaluation. On one side, personalization may create perceived personalization value when consumers experience it as relevant, useful, and consistent with their needs and expectations. On the other side, the same personalization may generate perceived privacy intrusion when consumers perceive the collection, inference, or use of personal information as excessive, unexpected, opaque, or insufficiently controlled.

The central theoretical argument is therefore that the personalization–privacy paradox should not be understood simply as a contradiction between consumers’ desire for personalization and their desire for privacy. Instead, it can be conceptualized as a context-dependent evaluative tension whose resolution depends on how consumers interpret the benefits and privacy implications of personalized interactions.

The framework further identifies several contextual conditions that may influence this resolution, particularly data type and sensitivity, personalization intensity, transparency, perceived control, and consumer expectations. These conditions may shape whether a particular personalization practice is interpreted primarily as valuable and beneficial or as intrusive and threatening. The resulting evaluation may subsequently influence consumer acceptance or resistance and, potentially, relational outcomes such as trust, engagement, and relationship continuation.

The main theoretical contribution of the study therefore does not lie in claiming to have discovered the personalization–privacy paradox. Recent research has already established the importance of this phenomenon and proposed several frameworks for understanding it. Rather, the contribution lies in reconciling fragmented explanations of the phenomenon around the question of how and under what conditions consumers resolve the tension between perceived personalization value and perceived privacy intrusion.

This positioning also distinguishes the proposed framework from existing approaches. Recent research has emphasized trust, utility, control, communication, governance, transparency, and other mechanisms. The present framework does not attempt to replace these perspectives. Instead, it integrates their insights into a broader explanation in which contextual conditions influence the interpretation of personalization and, consequently, the direction of consumer response.

The study also contributes conceptually by distinguishing privacy concern from privacy-driven resistance. Privacy concerns do not necessarily lead consumers to reject personalized services. Consumers may tolerate or accept perceived privacy costs when personalization generates sufficient perceived value. Resistance becomes theoretically more likely when perceived intrusion is interpreted as a threat to autonomy, control, or acceptable data use. This distinction provides a conceptual bridge between Privacy Calculus Theory and Psychological Reactance Theory.

From a managerial perspective, the framework suggests that firms should not pursue maximum personalization as an unconditional objective. Instead, they should seek an appropriate level of personalization that enhances consumer relevance while minimizing perceptions of unnecessary or unexpected intrusion. Particular attention should be given to data sensitivity, transparency regarding data practices, meaningful consumer control, and alignment between personalization practices and consumer expectations.

Finally, the conceptual nature of the study represents an important limitation. The proposed framework and propositions have not been empirically tested and should therefore be understood as theoretically derived rather than empirically established. Future research should test the proposed dual-evaluation mechanism and examine whether data sensitivity, personalization intensity, transparency, perceived control, and consumer expectations moderate or condition the relationship between AI personalization and consumer responses. Experimental and longitudinal research could be particularly valuable in determining whether and how consumers’ evaluations change as AI personalization becomes increasingly sophisticated.

Ultimately, the study argues that the strategic challenge of AI-driven personalization is not simply to determine whether personalization or privacy is more important. The more fundamental question is how consumers interpret the relationship between personalization benefits and privacy implications within a particular context. Understanding this contextual resolution may provide a more comprehensive theoretical basis for explaining why the same AI personalization capability can generate perceived value for some consumers while producing perceived intrusion, resistance, or distrust for others.

References

Aguirre, E., Mahr, D., Grewal, D., de Ruyter, K., & Wetzels, M. (2015). Unraveling the personalization paradox: The effect of information collection and trust-building strategies on online advertisement effectiveness. Journal of Retailing, 91(1), 34–49. https://doi.org/10.1016/j.jretai.2014.09.005

Al Helaly, Y., Dhillon, G., & Oliveira, T. (2025). The impact of privacy intrusiveness on individuals’ responses and engagement toward personalization in online interactive advertising. Journal of Interactive Advertising, 25(1), 38–60. https://doi.org/10.1080/15252019.2024.2440318

Ameen, N., Hosany, S., & Paul, J. (2022). The personalisation-privacy paradox: Consumer interaction with smart technologies and shopping mall loyalty. Computers in Human Behavior, 126, 106976. https://doi.org/10.1016/j.chb.2021.106976

Awad, N. F., & Krishnan, M. S. (2006). The personalization privacy paradox: An empirical evaluation of information transparency and the willingness to be profiled online for personalization. MIS Quarterly, 30(1), 13–28. https://doi.org/10.2307/25148715

Beke, F. T., Eggers, F., Verhoef, P. C., & Wieringa, J. E. (2022). Consumers’ privacy calculus: The PRICAL index development and validation. International Journal of Research in Marketing, 39(1), 20–41. https://doi.org/10.1016/j.ijresmar.2021.05.005

Chen, Q., Feng, Y., Liu, L., & Tian, X. (2019). Understanding consumers’ reactance of online personalized advertising: A new scheme of rational choice from a perspective of negative effects. International Journal of Information Management, 44, 53–64. https://doi.org/10.1016/j.ijinfomgt.2018.09.001

Chen, X., Sun, J., & Liu, H. (2022). Balancing web personalization and consumer privacy concerns: Mechanisms of consumer trust and reactance. Journal of Consumer Behaviour, 21(3), 572–582. https://doi.org/10.1002/cb.1947

Cloarec, J. (2020). The personalization–privacy paradox in the attention economy. Technological Forecasting and Social Change, 161, 120299. https://doi.org/10.1016/j.techfore.2020.120299

Eisend, M., Niewiadomska, D., & van Noort, G. (2026). EXPRESS: Personalization in marketing communication: A meta-analysis. Journal of Marketing. Advance online publication. https://doi.org/10.1177/00222429261460888

Erlei, A., Abbas, T., Bizer, K., & Gadiraju, U. (2026). The Data-Dollars Tradeoff: Privacy Harms vs. Economic Risk in Personalized AI Adoption. In N. Oliver, D. A. Shamma, H. Candello, P. Cesar, P. Lopes, A. Bozzon, T. Kosch, V. Liao, X. Ma, V. Artizzu, F. Draxler, G. Lopez, A. V. Reinschluessel, X. Tong, & P. O. Toups Dugas (Eds.), CHI 2026 – Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems, Article 1261. Association for Computing Machinery (ACM). https://doi.org/10.1145/3772318.3791427

Ham, M., & Lee, S. W. (2026). Antidote for the personalization-privacy paradox: Does algorithm transparency trigger higher ad click-through intention than algorithm literacy? Internet Research. https://doi.org/10.1108/INTR-08-2023-0672

Hsu, M.-W., Parry, G. C., & Ng, I. (2026). Personalisation-privacy paradox: Systematic review and survey evidence on personal data stores. Expert Systems, 43(7), e70314. https://doi.org/10.1111/exsy.70314

Mehrabian, A., & Russell, J. A. (1974). An approach to environmental psychology. MIT Press.

Şen, B. (2026). AI-Based Personalization and Consumer Trust: A Conceptual Framework for Understanding the Personalization–Privacy Paradox. Journal of International Management Educational and Economics Perspectives, 14(1), 26-37.

Torraco, R. J. (2005). Writing integrative literature reviews: Guidelines and examples. Human Resource Development Review, 4(3), 356–367. https://doi.org/10.1177/1534484305278283

Vu, C. B., Ozuem, W., Willis, M., & Katz, J. (2026). Navigating trust and the personalization-privacy paradox in AI-driven e-commerce: A consumer pathway framework. Journal of Retailing and Consumer Services, 93, 104938. https://doi.org/10.1016/j.jretconser.2026.104938

Whittemore, R., & Knafl, K. (2005). The integrative review: Updated methodology. Journal of Advanced Nursing, 52(5), 546–553. https://doi.org/10.1111/j.1365-2648.2005.03621.x

Original Version Master file–Mapan Journal-Dr.Bessas Hocine Download

When Personalization Becomes Intrusion: Explaining the Personalization–Privacy Tension in AI-Driven Marketing

Leave a Reply Cancel reply