This article finds that online reviews submitted during the weekend tend to have lower rating scores than reviews submitted during the week. Analyzing 400 million reviews across 33 e-commerce, hospitality, entertainment, and employer platforms, the authors find that weekend reviews have a 3% lower relative share of 5-star ratings and a 6% higher relative share of 1-, 2-, or 3-star ratings compared with weekday reviews. The pattern emerges even when controlling for quality of reviewed items. This weekend effect is surprising given that studies usually report higher happiness levels and a better mood on weekends. The authors discuss several explanations related to where the review is submitted (platform characteristics), what the review is about (listing characteristics), and who submits the review (reviewer characteristics). They present evidence that temporal self-selection of reviewers is a dominant driver of the weekend effect. During the weekend, a different set of users-those more prone to write negative reviews-is more likely to leave a review. These findings complement extant research on review self-selection by adding a temporal layer to the self-selection processes inherent in online reviews. This article also highlights managerial implications by demonstrating that solicitations sent during the weekend (vs. weekday solicitation) lead to collecting more negative reviews.

What constitutes good research, and how should it be evaluated and communicated in scholarly publishing? Drawing on nearly 25 years of editorial experience at leading marketing journals, we develop a unified perspective on the research and publication process, inspired by the legacy of Don Lehmann. We conceptualize good research as work that is interesting, important, and not wrong, and examine how trade-offs among these dimensions shape scholarly output. We contend that prevailing publication norms often overemphasize methodological sophistication and formal metrics, at the expense of interestingness and importance. In contrast, impactful research prioritizes meaningful questions, aligns methods with the problem, and communicates insights clearly. We want to stimulate a reorientation toward research that truly advances knowledge and practice. For researchers, we provide actionable guidance on how to craft work that is both publishable and influential. For editors and reviewers, we offer principles for decision-making and reducing type II errors-rejecting potentially high-impact work. For institutions, we challenge the reliance on imperfect metrics and call for better evaluation systems. Ultimately, we argue that the future of marketing scholarship depends not on maximizing publication counts, but on generating insights that are worth knowing, robust enough to trust, and consequential enough to matter.

Our faces are said to be windows into the soul. But can they also reflect who we are as consumers? Can facial images predict brand preferences? To answer these questions, we analyze a unique dataset of over 100,000 single-face Twitter profile pictures linked with brand followership data for 444 brands across categories and brand personality metrics. Using advanced machine learning for automated face analysis, we demonstrate that consumers' social media profile faces can reveal their preferences between rival brands (study 1). We further show that consumers who follow brands with similar brand personality traits (e.g., glamorous, rugged) tend to look alike (study 2). Finally, we identify facial appearance characteristics associated with brand personality traits; for example, followers of "smart" brands like Barnes & Noble are likely to wear reading glasses, while "edgy" brands like Diesel attract bearded men (study 3). Together, our findings show that brands can extract actionable insights from consumers' facial images, both to address the "cold start" problem by predicting preferences before any consumption actions are observed and to better understand the brand's identity through the aggregated traits of its followers. This work is among the first to demonstrate large-scale associations between faces and brand preferences.

Maximilian Beichert, Andreas Bayerl, Jacob Goldenberg, and Andreas Lanz show that influencers with small followings, despite their limited reach, produce a higher return on investment in direct sales campaigns than their larger counterparts. By linking the strength of viewers’ engagement to actual revenue rather than vanity metrics, they provide the first large-scale evidence that, in influencer marketing, bigger is not always better.

Andreas Lanz, Jacob Goldenberg, Daniel Shapira, and Florian Stahl introduce a forward-looking influencer marketing framework that allows managers identify and engage still-unknown prospective influencers in order to buy future endorsements.

Jacob Goldenberg, Andreas Lanz, Florian Stahl, and Daniel Shapira show that managers and creators who build ties with nearby, low-status influencers outperform those who cultivate high-status influencers. This efficient means of expanding audiences uses the existing dynamics of social networks.

LLM-based digital twins represent individuals through text alone, capturing what people report in surveys while leaving physical appearance unrepresented despite its central role in human identity. We introduce a validated approach for generating synthetic persona facial images from survey data alone, without real photographs. Building on the Twin-2K-500 framework of Toubia et al. (2025), we extract portrait-relevant visual cues from each respondent's survey answers and use a large image model to generate photorealistic facial images per persona. Three analyses validate that the generated faces represent the underlying persona profiles: (1) faces match their source persona above chance; (2) more similar survey profiles generate more similar faces, even after controlling for demographic similarity; and (3) survey questions unrelated to appearance predict similarity in appearance-related visual cues. An exploratory analysis indicates that depression and Big Five scales — and, less consistently, anxiety — are most associated with the personality–face link. Our contribution is representational: a visual layer can be constructed from non-visual survey data alone, providing a reusable complement to the textual twin. Because the visual layer is defined by persona-level visual-cue prompts rather than fixed images, it can be re-rendered by any current or future image model as the technology improves.

Journal Article Improving Our Scientific Understanding of Consumer Behavior Get access Oleg Urminsky, Oleg Urminsky Email: [email protected] https://orcid.org/0000-0003-4390-386X Search for other works by this author on: Oxford Academic PubMed Google Scholar Giana M Eckhardt, Giana M Eckhardt Search for other works by this author on: Oxford Academic PubMed Google Scholar Jacob Goldenberg, Jacob Goldenberg Search for other works by this author on: Oxford Academic PubMed Google Scholar Margaret G Meloy, Margaret G Meloy Search for other works by this author on: Oxford Academic PubMed Google Scholar Stephen A Spiller Stephen A Spiller Search for other works by this author on: Oxford Academic PubMed Google Scholar Journal of Consumer Research, Volume 51, Issue 5, February 2025, Pages 867–870, https://doi.org/10.1093/jcr/ucae074 Published: 21 January 2025

Excessive monetary compensation and existing contractual agreements of influencers limit the ability of many firms to engage in effective influencer seeding. The authors suggest a forward-looking approach of targeting prospective influencers—while they are still largely unknown (e.g., a few months after their platform registration)—and signing them to endorse the firm in the future (e.g., more than a year later). This approach has the potential to significantly reduce costs. However, as only rarely do newly registered users ultimately become influencers (and as signals are weak), the authors propose a novel framework to cope with this rare-event problem. For empirical demonstration and application, the authors conduct data-based simulations using a data set from a worldwide leading audio platform. Every wave of newly registered users is associated with a profit potential stemming from future endorsements by prospective influencers. With knowledge about the order of magnitude of the return on successful influencer spend, managers applying the framework can extract around 20% of this profit potential (if the return is around three times the spend).

Our given name is a social tag associated with us early in life. This study investigates the possibility of a self-fulfilling prophecy effect wherein individuals' facial appearance develops over time to resemble the social stereotypes associated with given names. Leveraging the face-name matching effect, which demonstrates an ability to match adults' names to their faces, we hypothesized that individuals would resemble their social stereotype (name) in adulthood but not in childhood. To test this hypothesis, children and adults were asked to match faces and names of children and adults. Results revealed that both adults and children correctly matched adult faces to their corresponding names, significantly above the chance level. However, when it came to children's faces and names, participants were unable to make accurate associations. Complementing our lab studies, we employed a machine-learning framework to process facial image data and found that facial representations of adults with the same name were more similar to each other than to those of adults with different names. This pattern of similarity was absent among the facial representations of children, thereby strengthening the case for the self-fulfilling prophecy hypothesis. Furthermore, the face-name matching effect was evident for adults but not for children's faces that were artificially aged to resemble adults, supporting the conjectured role of social development in this effect. Together, these findings suggest that even our facial appearance can be influenced by a social factor such as our name, confirming the potent impact of social expectations.

On user-generated content platforms, individuals and firms alike seek to build and expand their follower base to eventually increase the reach of the content they upload. The bulk of the seeding literature in marketing suggests targeting users with a large follower base, that is, high-status influencers. In contrast, some recent studies find targeting lower-status influencers to be a more effective seeding policy. This multimethod article shifts the focus from the follower base of the seeding target to the focal content creator. The authors propose accelerating natural triadic closure by leveraging first-degree followers as interconnectors to target second-degree followers, that is, the nearby (low-status) influencers (who are interconnected with the focal content creator). Empirical studies document that this seeding target is much more effective for building and expanding the follower base, compared with targeting influencers who are not interconnected with the focal content creator-that is, the remote (both high- and low-status) influencers-by 2,300% and 46%, respectively. These studies on the acceleration of natural triadic closure are augmented by a preregistered field experiment to obtain convergent validity of the findings.

Direct-to-consumer firms increasingly believe that influencer marketing is an effective option for seeding. However, the current managerially relevant question for direct-to-consumer firms of whether to target low- or high-followership influencers to generate immediate revenue is still unresolved. In this article, the authors’ goal is to answer this question by considering for the first time the whole influencer-marketing funnel, that is, from followers on user-generated content networks (e.g., on Instagram), to reached followers, to engagement, to actual revenue, while accounting for the cost of paid endorsements. The authors find that low-followership targeting outperforms high-followership targeting by order of magnitude across three performance (return on investment) metrics. A mediation analysis reveals that engagement can explain the negative relationship between the influencer followership levels and return on investment. This is in line with the rationale based on social capital theory that with higher followership levels of an influencer, the engagement between an influencer and their followers decreases. These two findings are derived from secondary sales data of 1,881,533 purchases and results of three full-fledged field studies with hundreds of paid influencer endorsements, establishing the robustness of the findings.

Direct-to-consumer (DTC) firms increasingly believe that influencer marketing is an effective option for seeding. However, the current managerially relevant question for DTC firms of whether to target low- or high-followership influencers to generate immediate revenue is still unresolved. In this article, the authors’ goal is to answer this question by considering for the first time the whole influencer-marketing funnel, i.e., from followers on user-generated content networks (e.g., on Instagram), to reached followers, to engagement, to actual revenue, while accounting for the cost of paid endorsements. The authors find that low-followership targeting outperforms high-followership targeting by order of magnitude across three performance (ROI) metrics. A mediation analysis reveals that engagement can explain the negative relationship between the influencer followership levels and ROI. This is in line with the rationale based on social capital theory that with higher followership levels of an influencer, the engagement between an influencer and his/her followers decreases. These two findings are derived from secondary sales data of 1,881,533 purchases and results of three full-fledged field studies with hundreds of paid influencer endorsements, establishing the robustness of the findings.

Consumers tend to have negative perceptions of service providers that limit their freedom. People might therefore be expected to respond particularly negatively to service providers that physically limit their freedom of movement. Yet, we suggest that physical constraints that a service provider unapologetically imposes with no obvious logical justification (e.g., closing a door and restricting consumers to stay inside a room) may, in fact, boost consumers' evaluations of the service provider. We propose that this effect occurs because consumers perceive such constraints as creating a structured environment, which they inherently value. Six studies lend converging support to these propositions, while ruling out alternative accounts (cognitive dissonance, self-attribution theory). We further show that the positive effect of physical constraints on evaluations is reversed when consumers perceive the constraints as excessively restrictive (rather than mild). These findings suggest that service providers may benefit from creating consumption conditions that mildly restrict consumers' freedom of movement.

Research (JCR) (2022) and a recent

Can the sensation of moving fast versus slow systematically influence consumer behavior? With recent technological innovations, people increasingly experience speed during decision making. They can be physically on the move with their devices or virtually immersed in speed simulated through their devices. Through seven experiments, we provide evidence for a speed-abstraction effect, where the perception of moving faster (vs. slower) leads people to rely on more abstract (vs. concrete) mental representations during decision making. This effect manifests for virtually simulated (experiment 1) and physically experienced (experiment 2) movement on moving trains. We suggest that it stems from an underlying speed-abstraction schema where people associate faster speed with abstraction and slower speed with concreteness (experiments 3a-3c). Weakening this schema attenuates the effect (experiment 4). Through a field study, experiment 5 demonstrates that video ads placed on Facebook are more engaging when virtually simulated speed matches the linguistic abstraction level of the message. Dimensions of psychological distance (time, space) and factors influencing mental representation (affect, fluency, spatial- orientation) are addressed as possible alternative explanations that cannot account for the effect. We propose a framework for understanding how experiencing speed—both physical and virtual—can influence decision making.

We show that a decision of potential shamers to take part in (“share” and “retweet”) an online shaming campaign against alleged wrongdoers is shaped by two factors: the potential shamer’s level of adherence to the nonmaleficence principle (i.e., do no harm) and the wrongdoer identifiability (the extent to which a wrongdoer’s details are exposed). Each shaming campaign may promote social norms and prevention of similar harm (i.e., positive consequences) and yet at the same time create harm for the individual wrongdoer being shamed (i.e., negative consequences). We suggest that potential shamers with high adherence to the nonmaleficence principle are more likely to join a shaming campaign with a low‐identifiability wrongdoer compared to potential shamers with a low endorsement of the nonmaleficence principle. We term this phenomenon the Nonmaleficence in Shaming effect. Five studies consistently demonstrate this effect and its attenuation in the case of a shaming campaign with a high‐identifiability wrongdoer. We further show that what drives the effect is advancing the positive over the negative consequences of a shaming campaign. Our findings contribute to a better understanding of norm‐enforcement behavior in digital communications and the social media space.

Social support is known to reduce stress and increase quality of life among patients undergoing IVF. Increasing social media use introduces a social support mechanism, yet data regarding the effect of this support on IVF outcomes are scarce. This observational, retrospective cohort study included women undergoing their first IVF cycle at an academic tertiary medical center. Fertility outcomes were compared between 82 women who were active users of social media (posting on Facebook at least 3 times a week) and 83 women who did not use Facebook or any other social media platform (the control group). For the social media group, we coded all Facebook Feed activities (Posts, Comments, Likes) for each participant up to 8 weeks prior to beta hCG test. Social support was measured by average Likes and Comments per post, on fertility outcomes. The social media group included more single women than the control group (17% vs. 5%, respectively, p = 0.012) and had a shorter infertility duration (1.6 ± 0.9 years vs. 2.3 ± 1.4, respectively, p = 0.001(. We found a trend in fertilization rates between groups (social media group 58% vs. controls 50%, p = 0.07). No difference was found regarding pregnancy rate between groups (p = 0.587). The social media group had a lower miscarriage rate compared to the controls (6% vs. 25%, p = 0.042). These results were also validated in the multivariant regression analysis. Social support (via Facebook) may have a positive effect on IVF outcomes, especially regarding miscarriages rate, with minor effect regrading fertilization rate and no effect regarding pregnancy rate. Therefore, encouraging women to be active on Facebook during treatment, including OPU day, may impact treatment results.

A general conjecture is that successful products attain their popularity through influence of adopters on their peers and product information disseminating over the social network. Indeed, many studies have confirmed the existence of local peer effects and contagion. But others have shown that peer influence has a marginal, if any, effect on cascades of adoptions. In this work, we study this discrepancy by analyzing video games propagating over the social network of gamers on Steam, the world's largest video game platform. A major identification problem – distinguishing homophily from peer influence – is a challenge in any peer influence study based on observational data. To overcome it, we introduce a novel method, Revealed Preference-based Matching Estimation, that estimates the impact of peer influence on adoption by using unsupervised machine-learning algorithm to match product adopters to users based solely on similarity of their past adoption. This procedure is applied to thousands of products and reveals how peer influence changes over their lifecycle, thus allowing us to draw general conclusions about the entire ecosystem. Results show that most products belong to one of two distinct groups, each exhibiting a characteristic temporal pattern of adoption: products that exhibit substantial peer influence; and products that do not, for which adoption is driven by preferences. Considering the reach of products in each group, surprisingly, we found that local peer effects are stronger in less popular products. Even more surprising is the fact that almost all blockbusters (products adopted by millions of users) did not exhibit substantial peer influence at any stage of their lifecycle. These results shed light on the discrepancy between observed local peer effects and the lack of peer influence in large adoption cascades that are characteristic of successful products.

As humans, we are uniquely competent at incorporating ourselves into groups that scale up from a few members to millions of individuals to engage in joint activities in social circles of varying sizes. Yet, the question of how a group's survival depends on its social structure is not well understood. In an analysis of more than 10 122 real-life online communities (with a total of 134 147 members) hosted by a leading platform over periods of more than a decade, we observe a prominent structural difference between stable and unstable communities, enabling the prediction of sustainability up to a decade ahead. We find that communities that fail to maintain a typical hierarchical social structure that preserves cohesiveness across size scales do not survive, while communities that exhibit such balance prevail. This difference is observable in as early as the first 30 days of a community's lifetime, enabling prediction of community sustainability up to 10 years in the future. We theorize that communities comprising distinct social structures that balance global and local factors across scales of sizes are more likely to maintain sustainability.