Genomarketing: the potential for hyper-personalization through consumer DNA

Hyper-personalization represents the next step in the nexus of consumer-company interaction, encompassing both customization and personalization options (Rosenbaum et al., 2021). In an era where “Your DNA is one click away” (Buiten, 2020), hyper-personalization goes beyond conventional marketing metrics (Daviet and Nave, 2024), capitalizing on a consumer’s unique genetic makeup. That is, it uses an individual’s set of genes, derived from DNA samples typically obtained from saliva, to tailor highly personalized products and services (Rosenbaum et al., 2017, 2021). This approach lies at the intersection of genetic science and marketing, transforming the Marketing 5.0 toolkit (Ivanova-Kadiri, 2022). Kotler et al. (2021) characterize the “Technology 5.0” era as one focused on designing technology that serves humanity. This shift reflects a growing trend among consumers towards preferring increasingly personalized experiences. The consumption of new technology has not only moved the needle but also expanded the possible scope of personal data to include biological and neurological markers. These markers, which reflect an individual’s unique biological composition, raise questions about privacy. Driven by genetic data and technological advancements (Fumagalli, 2019), (hyper-) personalization is now redefining marketing strategies through enhanced segmentation, targeting and positioning (Daviet et al., 2022; Daviet and Nave, 2024). Genetic data, once confined to medical breakthroughs, has emerged as a valuable commercial asset. This technological leap sets the stage for a neologism introduced by this study, “Genomarketing,” which leverages genetic data to refine marketing strategies and enhance personalized consumer interactions.

A key enabler of this shift is direct-to-consumer genetic testing (DTC-GT), which allows consumers to access their DNA data without medical intermediaries. In 2023, the global DTC-GT market was valued at approximately $1.98bn and is projected to grow at a compound annual growth rate (CAGR) of 15.80% from 2024 to 2033, reaching over $8.57bn by the end of this period (Precedence Research, 2024). As of 2022, more than 22 million people have taken personalized DNA tests (Brady, 2022). Although most companies (e.g. 23andMe, AncestryDNA) reside in the USA, consumers worldwide can obtain genetic tests via the Internet. Yet, the exponential expansion of DTC-GT and its potential for hyper-personalization also raise significant ethical and regulatory challenges. This tension is evident in global regulatory approaches.

Despite its promise, this move towards hyper-personalization presents a dilemma, raising critical questions about the future course of human civilization:

Europe’s GDPR sets a high bar for consumer protection, treating genetic data as sensitive and requiring explicit consent for its use (Shabani and Borry, 2018). By contrast, the US leans towards a more aggressive commercialization model, where privacy protections are less stringent. In markets like China, regulatory oversight is notably weaker; over half of the DTC-GT providers lack privacy policies, and informed consent forms are rarely provided (94%) (Du and Wang, 2020).

At its inception, the DTC-GT market generated revenue through the sale of testing kits. Over time, however, many have signaled intentions to monetize the genetic data collected, often under ambiguous consent terms. This commodification of genetic data often leaves consumers unaware of how their data is monetized. Privacy is increasingly treated as a commodity whose value can be quantified (Hann et al., 2007) with genetic data ownership being as much an economic issue as it is an ethical one.

To date, marketing scholars have largely neglected the applications of genetics, with only a few exceptions (Bowen et al., 2005; Daviet et al., 2022; Daviet and Nave, 2024; Fumagalli, 2019; Gabel, 2011; Gil and Guerreiro, 2025; Ivanova-Kadiri, 2022; McGuire, 2023; Moorman et al., 2024; Nill and Laczniak, 2022; Patsiaouras, 2017; Pearson and Liu-Thompkins, 2012; Raeiro et al., 2025; Zheng and Alba, 2021). To the best of our knowledge, this research represents one of the first empirical attempts to investigate hyper-personalization via DNA sequencing. Against this background, this study advances consumer research by seeking to explore the drivers behind consumer willingness-to-share their DNA data with third parties for hyper-personalized products and services. In doing so, this study contributes, in an exploratory manner, to the marketing literature by uncovering the role of DNA data in hyper-personalization.

All too often, we hear about DNA but not what it can and cannot tell us. DNA is often regarded as merely a biological blueprint, but its implications are far more than this simplistic view. It contains unique and immutable identifiers that render it exceptionally sensitive and deserving of special consideration, a view known as “genetic exceptionalism” (Sulmasy, 2015). Unlike other types of sensitive data, DNA holds the keys to deeply personal insights into who we are, influencing not only our biological traits but also our understanding of heritage and health outcomes (Steverson et al., 2024). Given DNA’s complex and sensitive nature, consumer privacy must be scrutinized, particularly as biological mechanisms intersect with technological advancements in hyper-personalization.

Personalization is defined as the ability to proactively tailor products […] to tastes of individual consumers based upon their personal and preference information (Chellappa and Sin, 2005, p. 181). It mostly manifests in two forms: personalized advertising and personalized services (Awad and Krishnan, 2006). For the purpose of this study, the focus is on personalized services, emphasizing the effectiveness of targeting individual needs. Unlike traditional approaches, personalization represents a transformative shift in marketing, where genetic science is seamlessly integrated with marketing strategies while blending both personalization and customization elements to create highly individualized experiences (Rosenbaum et al., 2021).

However, consumers may resist hyper-personalization, perceiving the collection and utilization of personal data that underpin hyper-personalization as too invasive (e.g. Moore et al., 2015). Known as the “privacy paradox”, first coined by Barnes (2006), it underscores the incongruity between consumer concerns about personal privacy and their behaviors. Privacy concerns potentially threaten consumers’ sense of autonomy by implying that preferences might be genetically predetermined, thereby undermining the perception of free choice in consumption decisions (Wertenbroch et al., 2020). In this study, this manifestation of the privacy paradox evolves into a more complex paradox, the hyper-personalization-privacy paradox. As a new phenomenon, it reflects growing consumer concerns towards privacy and genetic data acceptance with third-party services (Gil and Guerreiro, 2025). In this paper, privacy concerns and personalization are emphasized as two salient facets.

Privacy calculus theory (PCT), originally proposed by Laufer and Wolfe (1977), has been extensively studied as one of the most foundational frameworks to examine privacy-related decision-making (e.g. Wang et al., 2024) as a cost-benefit analysis where individuals weigh the benefits of data sharing against privacy risks (Chellappa and Sin, 2005). Despite its extensive application in other contexts, PCT’s use in DTC-GT consumer privacy remains scarce, with only Gil and Guerreiro (2025) exploring this area. Whether the benefits in this context outweigh the privacy concerns remains in question (Daviet et al., 2022).

Consumer perceptions often lean more favorably towards the benefits from DTC-GT, with risk playing a lesser role in their decision-making process (Grandhi and Plotnick, 2022). These tests, ranging from uncovering forgotten family histories to identifying genetic predispositions towards certain health conditions, appeal to cognitive-driven consumers’ views (Daviet et al., 2022). Nill and Laczniak (2022) classify DTC-GT into eight distinct types: Ancestry, relatedness (e.g. Baig et al., 2020), nutrigenetic, talent and athletic ability, prenatal tests, diagnostic tests, personalized medicine, carrier testing (e.g. King, 2019). Specifically, Toussaint et al. (2022) streamline these into three broader groups: Health, Relationship (or Ancestry), and Lifestyle. Such classifications not only segment the primary consumer bases, health enthusiasts, specific genetic information seekers and consumers with concerns over chronic health conditions or genetic risks but also enhance consumers’ understanding of the diverse offerings of the proposed service, underscoring the critical, sometimes life-altering, benefits of personalized health services (Bol et al., 2018).

Additionally, DNA testing stands out as an exponential advancement over traditional information-seeking methods. Traditional methods, such as historical records and parish registers to compile names and dates of birth and death, often provide limited information and may encounter missing or conflicting records (Darby and Clough, 2013; Duff and Johnson, 2003). There are, at the very least, two key benefits to opting for DNA testing. First, it provides a rock-solid “peace of mind” by confirming suspicions or clarifying uncertainties about one’s ancestry, allowing individuals to understand their roots more clearly (Hazel et al., 2021). Second, it helps overcome “brick wall”, points at which traditional methods fail due to missing or conflicting records. By uploading DNA results into an ecosystem, DNA testing offers new leads and connections that traditional documentation might not uncover.

As consumers increasingly seek personalized benefits, privacy costs simultaneously surface. Notwithstanding, the question remains as to which kind of privacy cost is most pertinent to consumers’ willingness-to-share their DNA data for such purposes. To date, privacy studies have characterized these costs in terms of such as privacy risk beliefs, perceived privacy risk or privacy concerns (Bol et al., 2018). In an attempt to be as straightforward as possible within this framework, the term privacy concerns will consistently refer to the cost factor within the PCT throughout the rest of this paper (Chen, 2018). From the standpoint of PCT, the term privacy concerns might be rooted in fear of opportunistic behavior (Carlsson Hauff and Nilsson, 2023) where personal information collectors may act in ways contrary to the consumer’s interests in sharing, a concern that can sometimes overshadow the benefits of personalization (Awad and Krishnan, 2006).

Unlike other forms of personal data (e.g. credit card numbers or email addresses), genetic data is immutable and cannot be changed. This permanence magnifies the fear of discrimination, stigmatization, or misuse among consumers’ minds (Daviet et al., 2022). For example, Aeroméxico, a major Mexican airline, ran a campaign offering flight discounts to US residents based on their percentage of Mexican DNA, as determined through genetic testing. This marketing strategy, which rewarded genetic proximity to Mexico, created genetic-based price discrimination. While novel, such practices raise ethical and practical concerns about whether these promotions deliver real value to consumers or merely serve as attention-grabbing tactics (Daviet et al., 2022).

To complicate matters further, even when DNA data are labeled as anonymized, it might still enable reidentification attacks (Gymrek et al., 2013). This identifiable and predictive nature of DNA data not only pertains to one’s genomic data but also, to some degree, to one’s nongenotyped relatives (e.g. Daviet et al., 2022). By its nature, DNA data, inherently familial, has the potential to reveal data about consumers beyond those who consented to testing, effectively serving as a lifelong identifier for both consumers tested and their biological relatives (Nill and Laczniak, 2022). Once leaked, DNA data becomes an irreversible action, with no feasible means to retract or shield this deeply personal information from being accessed and used by others, often overlooking regulations or ethical norms.

Trust plays a crucial role in privacy calculus models; however, its integration needs to be more present in DTC-GT research. Prior research has incorporated the concept of trust, in its diverse manifestations, alongside the personalization-privacy paradox (e.g. Aguirre et al., 2015; Guo et al., 2016; Cloarec et al., 2024), but not in the specific context of DTC-GT. It assumes a complex and critical role, acting as a key mediator between personalized benefits and consumer privacy concerns (Guo et al., 2016; Wu and Xu, 2023).

In the DTC-GT market, privacy concerns may undermine trust by raising doubts about organizations’ ability to protect DNA data. High concerns about monetization, in particular, might lead to skepticism and distrust toward commercial DTC-GT companies often perceived as profit-driven (Critchley et al., 2021; Schaper et al., 2019). Low concerns, however, may foster implicit trust, as some consumers equate DNA data with digital data, assuming anonymity and extending confidence to third-party sharing (Baig et al., 2020).

Personalized benefits may counterbalance privacy concerns, positively influencing trust when consumers perceive tangible value from sharing their data (Cloarec et al., 2024). However, trust erodes when privacy risks outweigh these benefits (Bol et al., 2018). As a mediator between privacy concerns and perceived benefits, trust determines whether consumers feel confident enough to engage in data sharing for hyper-personalization (Martin and Murphy, 2017). Without trust, even significant personalized benefits may fail to overcome privacy concerns, as trust provides the confidence needed to engage in data-sharing activities (Cloarec et al., 2024). In this way, trust serves as a bridge between risk and reward, functioning as the linchpin of the privacy calculus framework.

Raeiro et al. (2025) found that trust in regulations had no significant effect on consumers’ decisions to undergo DTC-GT. Instead, both privacy concerns and perceived benefits were significant drivers of decision-making, with benefits being the stronger predictor. While regulatory trust was non-significant, this study explores trust more broadly as a key mechanism in DNA data-sharing decisions.

Extensive literature has been focused on consumer data-sharing willingness, either for altruistic scientific or commercial purposes. Research have consistently shown that consumers are aware of the secondary uses of their DNA data (Baig et al., 2020; Mladucky et al., 2021). This willingness-to-share, however, is not uniform and varies significantly with the intended purpose behind data utilization. Altruistically, previous studies have revealed that consumers feel comfortable with the use of DNA data for research purposes (Mladucky et al., 2021), especially when it contributes to advancements in science and medicine (Haeusermann et al., 2018). In contrast, other studies observed low willingness among individuals to donate DNA for research, attributing this hesitance to low trust levels in data-sharing practices (Middleton et al., 2020). Looking at it commercially, consumers seem to be resistant to the idea of third parties profiting from personal DNA data (Baig et al., 2020; Mladucky et al., 2021). The core of this centers on the key question:

Awad and Krishnan (2006) highlighted a notable reluctance among consumers to be subjected to online profiling for personalization, presenting a paradox for companies that invest heavily in personalized services. Trust can positively influence consumers’ willingness-to-share information for personalization purposes (Cloarec et al., 2024).

In all, trust is seen as a central mediator influencing consumers’ willingness-to-share DNA data with companies for such purposes, rendering the following hypotheses:

On the basis of these hypotheses, the following schematic research model was developed to serve as a summary of the proposed pathways’ predictions (Figure 1).

An online survey was conducted using the Qualtrics platform in two phases. In the first phase, a pilot study was conducted with 40 Swedish participants to establish the reliability and clarity of the questionnaire, focusing on the concept of hyper-personalization, referred to herein as high tailoring. In the second phase, the primary survey was rolled out in collaboration with a Swedish genealogy organization that targeted Swedes interested in ancestry, facilitated through announcements in newsletters and emails.

Recruitment took place between March and April 2024. Of the 2,031 individuals invited to participate, 582 provided valid responses after excluding 31 (∼5%) for careless answering, resulting in a 29.1% valid response rate (Meade and Craig, 2012). To ensure accuracy and cultural relevance, the survey was translated into Swedish, leveraging the country’s strong legacy in global genetic research (Swede et al., 2007). Participation was entirely voluntary, and informed consent was obtained before respondents to proceed.

This study broke new ground in its ability to reach the right sample. Genealogy is a rare interest, with less than 0.5% of the Swedish population estimated to have undergone DTC-GT (Swedish National Forensic Centre, 2021). Most DTC-GT services are still primarily used for discovering ancestry, even as new services emerge offering other applications. Unlike previous studies that relied on social media platforms such as Facebook to recruit participants (e.g. Gil and Guerreiro, 2025), this research leveraged a strategic partnership the genealogy organization to access a group of highly knowledgeable consumers about the details and options of DNA testing, making their insights uniquely valuable to understand consumer behavior in this domain. Accessing such a specialized audience is inherently difficult, which explains why many previous studies rely on small, qualitative samples (<25 participants). By partnering with the Swedish organization, this study successfully addresses this gap, gathering data from a hard-to-reach group and achieving a substantial sample of consumers with firsthand experience in DNA testing.

A non-probability sampling method was used to target genealogy enthusiasts who had experience with DNA testing. Random sampling was not feasible due to the highly specialized nature of the target group. To protect confidentiality and integrity, the survey omitted identifying details (e.g. names, email addresses). Only aggregated data were analyzed, and no individual-level information was disclosed at any stage. Participants were provided with clear information about the study’s purpose, data collection process and assurances of confidentiality, with the option to withdraw at any stage. To respect participants’ autonomy and avoid over-solicitation, only one follow-up reminder was sent, reflecting the Swedish principle of “lagom” (just the right amount) (Kittleson, 1997).

The demographic profile of the sample revealed a skew toward older adults, with approximately 91% of respondents aged 60 or above, reflecting the typical demographic of individuals with an active interest in genealogy. While this age distribution may limit generalizability, it could be mitigated by the fact that 66.5% of participants had already undergone DNA testing, indicating that concerns about privacy did not prevent them from sharing their genetic information with private entities (Christofides and O’Doherty, 2016; Grandhi and Plotnick, 2022). The survey also achieved a balanced gender distribution, with approximately 50.5% males and 47.9% females, as detailed in Table 1. Among those who had taken a DNA test, the primary motivation was to gain insights into their ancestry, with FamilyTreeDNA (45.9%) and AncestryDNA (21.6%) emerging as the most popular providers. Entry-level tests from these providers offer ancestry testing at an accessible cost, suggesting this suggests that socio-economic factors were not deemed key barriers for participants interested in genealogy. This aligns with Kirkpatrick and Rashkin (2017), who state these providers as key players in DTC ancestry testing.

Behavioral patterns in the sample highlight that 40.9% of participants underwent only one DNA test, and 25.3% had not updated their testing in over five years, pointing to an early adoption phase followed by a plateau. Interestingly, retracting DNA data from databases was uncommon (65.3%). In contrast to actual sharing behavior, perceived willingness-to-share behavior, representing participants who had not yet undergone a DNA test, stood at 33.5%, with only 10.3% expressing an aversion, revealing a clear divide in attitudes toward genetic data exchange.

For this study, a structured questionnaire was crafted, consisting of three parts. Categorically, the first part introduced the purpose and the context of the research, setting the stage for respondents to understand the study’s objectives. The second part contained a screening section determining participants’ prior experience with a DTC-GT and statement-like items rated on a five-point Likert scale anchored by (1) = strongly disagree and (5) = strongly agree. To the extent possible, all constructs were established based on prior research and were reframed to suit the present research context, except for personalized benefits (PB). This original construct was developed from prior qualitative studies (Grandhi and Plotnick, 2022; Baig et al., 2020; King, 2019) owing to the absence of pre-existing scales. Privacy concerns (PC) was operationalized by adapting a four-item scale from Xu et al. (2011). Trust (TR) was posited as the central mediator of the hyper-personalization-privacy paradox for this study, measured using a five-item scale adapted from Malhotra et al. (2004). As an outcome, hyper-personalization willingness-to-share (HPW) was innovatively adapted from Culnan and Armstrong’s (1999) two-item scale (for review, see also Awad and Krishnan, 2006), refined to emphasize hyper-personalization. This adaptation sets it apart from broader willingness-to-share measures (e.g. Xu et al., 2011) by explicitly giving emphasis on the sharing of highly personal data. Unlike generic data-sharing contexts, DNA testing makes the decision to share uniquely consequential. In such cases, a focused two-item construct avoids dilution and maintains clarity. Prior research supports this approach in exploratory contexts. Finally, the third part covers standard demographic data through closed-ended, multiple-choice questions on gender and educational background and uniquely asked participants’ birth years rather than age intervals to improve data accuracy. Notably, respondents were allowed to leave comments in an open-ended question section for additional comments to deepen the understanding of the collected data. All items are listed in Table 2.

To examine the research model of this study, a three-step approach was utilized, using SPSS 25.0. and STATA 18.0. Initially, descriptive analysis was used to summarize the data set’s characteristics. Subsequently, exploratory factor analysis was performed to assess the adequacy of the measurement scales, particularly for the newly developed scale (PB). Based on the prior information from EFA, CFA was then applied to validate the construct validity of the scales, further refining the factor structure identified by the EFA. The final stage entailed assessing and analyzing the hypothesized relationship model through structural equation modeling (SEM).

In the first step, descriptive analysis was conducted for all variables. As summarized in Table 3, each variable presents the means and standard deviations for each study variable. Notably, PB was generally perceived positively by consumers regarding DNA testing, which highlights their value of tangible advantages such as ancestry insights (M = 3.81, SD = 1.052). However, PC showed moderate levels, suggesting that while concerns exist, they may not entirely overshadow the perceived benefits (M = 3.28, SD = 1.241), TR exhibited lower scores, reflecting consumers’ cautious attitudes toward the organizations managing their data (M = 2.86, SD = 1.166). As expected, HPW revealed a reluctance among consumers to share their genetic data, with attitudes toward sharing behavior showing considerable variability (M = 1.71, SD = 1.034). Such low willingness-to-share is particularly striking, as it comes from a sample of consumers who have already shared their genetic data. One possible explanation might be that even among a group predisposed to engage in data-sharing practices, systemic resistance to hyper-personalization persists. Therefore, the hyper-personalization paradigm may not align with consumer preferences, particularly when TR and PC act as key barriers.

In the second step, exploratory factor analysis (EFA) was used to uncover the underlying structure of the instrument. Principal component analysis with varimax rotation, a commonly used method that simplifies the factor structure by maximizing the variance of the squared loadings, initially extracted 14 observed variables across four reflective theoretical constructs.

Factor loadings below 0.4 were discarded to simplify a factor structure, exceeding the general rule for over 350 observatories (n = 582). All items are loaded onto the expected factors. After inspecting the communality, however, PB3 was excluded due to its communality value falling below the <0.5 threshold (Hair et al., 2013), ensuring sufficient shared variance within the factor. By adopting the eigenvalue of 1, the revised factors, including PB, PC, TR and HPW, accounted for an increased total variance explained by 80.340%. As such, the data demonstrated robust construct validity.

Bartlett’s test of sphericity demonstrated a significant correlation between the original variables (chi-square = 10869.323, p < 0.001), whereas the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy is close to 1 (KMO = 0.839); together they indicate that the data are suitable for factor analysis (Hair et al., 2013).

To address potential common method variance (CMV), procedural and statistical precautions were taken. Procedurally, participants were assured anonymity, a pilot study refined item clarity, attention-check filtered inattentive responses, mixed wording (e.g. hyper-personalization = high tailoring) reduced priming effects and randomization of question order with varied response formats (Likert, multiple-choice, open-ended) reduced response biases. Statistically, Harman’s Single-Factor Test indicated no single factor explained over 50% (only 31%) of the variance (Podsakoff et al., 2003) and a full collinearity check confirmed all variance inflation factors (VIF) were below 3.3 (Kock, 2015), suggesting minimal CMV concerns.

Reliability check, with Cronbach’s alpha (α) values, ranged from 0.844 to 0.961 All items scored well above the recommended cut-off of 0.70 (Hair et al., 2013), confirming the constructs’ reliability and their ability to capture consumer attitudes. These results validate the use of EFA.

Following EFA, the factor structure was further tested through confirmatory factor analysis (CFA). As shown in Table 4, the results of goodness-of-fit indices (GFI) for the SEM of hyper-personalization are displayed. The model yielded a chi-square CMIN/df = 3.37. Although this value exceeds the more stringent threshold of 3.00 (>3.00), it remains below the permissible cut-off point of 5.00. This is considered acceptable, given the Chi-square test’s sensitivity to large sample sizes (Byrne, 2013). To assess the overall model fit, four widely recommended fit indices were used (Hu and Bentler, 1999): Root mean square error of Approximation (RMSEA) = 0.077 (<0.10) (Browne and Cudek, 1993), Comparative fit index (CFI) = 0.932 (>0.90) (Hu and Bentler, 1995), Standardized Root Mean Square Residual (SRMR) = 0.04 (<0.10) (Bentler, 1995), Tucker–Lewis index (TLI) = 0.919 (>0.90) (Bentler and Bonett, 1980). As a result, the indices demonstrated the model of goodness-of-fit in which the final model adequately fits the data.

Once fit validity was established, construct validity was assessed through evaluations of both convergent and discriminant validity. Convergent validity was ascertained by inspecting the factor loadings (lambdas), which were found to be satisfactory (>0.50) and highly significant across all constructs (Jöreskog and Sörbom, 1993) criteria. Additionally, the Composite Reliability of all constructs exceeded the 0.80 threshold, signifying their acceptability. The average variance extracted (AVE) was also tested for all constructs. By computing AVE, all constructs surpass the cut-off value of 0.50 (Hair et al., 2013). Consequently, convergent validity was assured (Table 5). Additionally, discriminant validity was assessed. On the basis of Fornell and Larcker (1981), the analysis involved comparing the AVE with the squared estimated correlations between each construct. As none of the squared correlations among the constructs surpassed the square root of the AVE from the constructs, it is evident that discriminant validity was successfully established.

For the structural model, the coefficient of determination (R²) was used to evaluate the predictive power of the consumer’s willingness-to-share DNA data with third parties for hyper-personalization purposes. The R², which is close to one, can signify a strong predictive ability; the complexity of consumer behavior in privacy-sensitive data-sharing contexts often results in lower R² values. As expected, the results revealed a moderate predictive power of the independent variables on the dependent variables: personalized benefits (R²_PB = 13.26%) and privacy concerns (R²_PC =16.57%) on trust (R²_TR =6.55%), but lower predictive power of the moderation on hyper-personalization willingness-to-share (R²_HPW = 4.7%). In (hyper-)personalization privacy paradoxes, it is expected to observe lower R² values due to the myriad influences on behavior. Regarding the effect size (f²), the tested model is depicted in Figure 2 with detailed SEM results presented in Table 6. The model includes two latent-independent (exogenous) variables: PB and PC; one latent-dependent (endogenous) variable: HPW; one latent-mediating variable: TR. As a mediating variable, TR was examined through indirect effects, calculated as the product of two paths:

1. the effect of the independent variable (PC or PB) on trust; and

2. the effect of trust on HPW.

For PC, the indirect effect was 0.012, reflecting that while privacy concerns undermine trust, the presence of trust slightly softens its negative impact on willingness-to-share genetic data for hyper-personalized purposes. For PB, however, the indirect effect was −0.022, revealing that while personalized benefits strengthen trust, this pathway paradoxically dampens consumer’s willingness-to-share.

To test the hypotheses and deepen understanding of the hyper-personalization-privacy paradox and the mediation role of trust, both direct and indirect relationships were analyzed. Results revealed that personalized benefits were shown to positively influence trust (t = 4.90, p < 0.01), supporting H1. Conversely, privacy concerns had a negative effect on trust (t = −2.75, p < 0.01), which confirmed H2. As the central mediator in this study, trust also positively influenced such hyper-personalization willingness-to-share (t = 2.48, p <0.05), affirming H3. As summarized in Table 6, all hypotheses were well supported, and the proposed model was empirically validated.

The paper illuminates the concept of hyper-personalization, examining it through the lens of genetic science. DNA-based applications in marketing, remain largely unexplored, leaving significant gaps, particularly in understanding the consumer’s willingness-to-share information (Gil and Guerreiro, 2025) for (hyper-)personalized products and services (Awad and Krishnan, 2006; Cloarec et al., 2024; Culnan and Armstrong, 1999). Toward this end, a novel research model was developed to decode the hyper-personalization-privacy paradox. The model, viewed through a privacy calculus lens, evaluates consumers by rationalizing and weighing the benefits of genetically tailored offerings against the potential risks to their privacy.

The first key finding brings to light the central role of trust in the calculus. Similar to Wu and Xu (2023) and Guo et al. (2016), trust acts as a mediator between personalized benefits and privacy concerns toward sharing behavior. Importantly, personalized benefits were found to have a positive influence on trust among consumers, particularly when it comes to the prospect of sharing data (Cloarec et al., 2024). This suggests that individuals who perceive personalized benefits from DNA testing are more likely to place trust in the companies handling their DNA data. Conversely, privacy concerns were observed to act as a stumbling block, eroding trust and fueling fears about potential data misuse or unauthorized access can diminish consumer trust in these companies. In turn, trust emerged as the gateway to sharing behavior in exchange for hyper-personalized experiences. While personalized benefits can encourage data sharing, there remains an aversion among consumers toward hyper-personalized experiences. This aversion may stem from concerns about over-profiling or intrusive marketing tactics. Beneath this willingness-to-share lies a cautious consumer stance, with trust tempering their willingness to engage in hyper-personalized data sharing. This finding lends support to Awad and Krishnan (2006), who highlighted a reluctance among consumers to undergo online profiling for personalization.

Interestingly, most consumers in this study engaged with these services more than five years ago and typically opted for a one-time testing service. This occasional engagement suggests that once consumers obtain their DNA results, they feel free of the need for additional tests. This behavior suggests a “set it and forget it” mentality, where consumers see DNA testing as a one-off transactional rather than ongoing relationship. Furthermore, it may also be reflected in the rarity with which consumers retract their DNA data from databases, an indication of a cautious yet passive attitude towards the management of stored data. Alternatively, it might reflect a cautious approach to the use of such technologies, driven by an instinctive mistrust.

These findings suggest that, moving forward, the future of hyper-personalization may face obstacles if it continues to operate under assumptions of ongoing consumer engagement. If hyper-personalization is to evolve, it must overcome not only skepticism about trust and privacy but also the passive disengagement that seems embedded in consumer behavior toward genetic data.

This study provides both theoretical and practical implications. Theoretically, it contributes to the literature body by bridging the gap between genetic science and marketing. As one of the first empirical studies to explore hyper-personalization via DNA sequencing, this research addresses a notable gap in the existing literature, offering fresh insights into consumer sharing behavior through DNA data (Rosenbaum et al., 2021). In the wake of Daviet et al. (2022), this study empirically answered the unresolved question concerning whether the potential benefits of DTC-GT indeed outweigh its privacy concerns. Finally, the study extends the conventional personalization-privacy paradox, decoding the hyper-personalization privacy paradox. This new phenomenon arises from the use of highly sensitive and personal genetic data, contributing to an advanced understanding of consumer privacy in hyper-personalized contexts.

Practically, this study advances knowledge for marketers and public policy decision-makers. Marketers have already begun translating consumers’ fascination with DNA to develop hyper-personalized advancements in genetic and genomic technologies raise privacy concerns, affecting willingness to disclose DNA data. One recent incident that illustrates the gravity of these risks is the 23andMe data breach. What started as a breach of just 0.1% of user accounts quickly snowballed into the large-scale exposure of nearly 7 million consumers’ personal information. As a result, curated lists specifically targeting Chinese and Ashkenazi Jewish descendants were shared on the dark web (Carballo, 2023; Carballo et al., 2024).

Given the cautious mindset of consumers toward sharing hyper-personalized data, marketers should consider implementing targeted educational initiatives. These programs could enhance digital literacy, helping to bridge the technology gap and making consumers more comfortable with, and receptive to, online platforms. Education should address data privacy concerns and reassure consumers about protective measures. Such education efforts are especially important in the DNA data marketplace, where similar educational materials on data sharing are lacking (Ahmed and Shabani, 2019).

As trust mediates the effects of privacy concerns and personalized benefits, marketers must prioritize trust-building measures. These measures, when implemented effectively, can help alleviate ethical considerations and skepticism about the actual benefits of such practices. Bowen et al. (2005, p. 676) liken its today use to “modern snake oil”, where deceitful con artists exploit the public with fraudulent remedies, fueling mistrust (Rosenbaum et al., 2017). Hyper-personalization faces similar skepticism, with uncertainties surrounding whether genomics truly offers more physiological benefits over standard products and services (Rosenbaum et al., 2021). Patsiaouras (2017) raises ethical concerns about promoting “'fault’ genes and creating false needs?”, which fuels public skepticism. Educating consumers about the legitimate benefits of genomics, alongside transparent communication about these aspects (Toussaint et al., 2022), can help mitigate the view of hyper-personalization as modern “snake oil” (Rosenbaum et al., 2021).

On the policy front, the proliferation of third-party genetic interpretation services continues to operate in a regulatory grey area. In Europe, the DTC-GT landscape remains highly fragmented. While some countries (e.g. France and Germany) have effectively banned DTC-GT outside clinical setting, others (e.g. Austria) impose partial restrictions. More liberal markets (e.g. Poland, Romania, Luxembourg and Sweden – where this study was conducted) allow greater market freedom. This regulatory disparity complicates universal policymaking and conceals a hidden ethical crisis: one-size-fits-all regulations are unlikely to succeed across such diverse legal and cultural contexts. Sweden’s Genetic Integrity Act (2006) ensures consent and data protection but does not explicitly regulate DNA-based marketing practices, limiting its scope in the commercial DTC-GT. Some European countries (e.g. Portugal, Italy, Hungary, Lithuania and Spain) mandate medical oversight or counseling before testing, while others (e.g. Denmark and The Netherlands) have no specific rules. These differences suggest that generalization of this study’s findings must be approached with sensitivity to national regulatory frameworks and cultural attitudes toward privacy, personalization, and risk. As previously discussed, one feasible policy remedy involves strengthening consumer education through “genetic counseling”. Although not explicitly required in Sweden, the legal obligation for informed consent implies a level of understanding typically supported by counseling, yet this support is often missing in commercial DTC-GT contexts. In more regulated environments, trust in healthcare intermediaries may shape consumer decision-making differently than in liberal markets where consumers engage with DTC-GT independently (Hoxhaj et al., 2020; Kalokairinou et al., 2018; Van Steijvoort et al., 2024).

This study carries several limitations, which can guide future research avenues. First, its focus on older consumers with a demonstrated interest in genealogy may limit its generalizability to the broader population. Examining age differences in consumer engagement with hyper-personalization technology could yield valuable insights. Young consumers, who are often more tech-savvy and open to hyper-personalized experiences, may exhibit different behaviors and preferences compared to elderly generations. Second, the uneven distribution of DNA-tested versus non-tested consumers could bias risk perceptions and cost evaluations. This imbalance may influence perceptions of risk and skew results (Christofides and O’Doherty, 2016; Grandhi and Plotnick, 2022). Future research should use balanced sampling or cluster analyses to address this. Third, this study did not examine consumer awareness of DNA-sharing risks, highlighting a need for future studies to examine how risk awareness affects disclosure decisions. Finally, as this study was conducted within Sweden’s specific cultural context, known for stringent privacy regulations (e.g. GDPR), generalizing findings to culturally and regulatorily distinct contexts like Mexico, Vietnam or the USA requires caution. Future cross-cultural comparative research is warranted to explore these regional differences in privacy attitudes and acceptance of genetic testing.