The Illusion of Authenticity in Online Reviews: Truth Bias and the Role of Valence

As consumer awareness of fake online reviews grows, platforms face increasing challenges in maintaining trust. Although skepticism toward reviews is rising, our research finds that consumers still exhibit a “truth bias,” meaning that they tend to accept individual reviews as genuine—even when fake review detection rates are low. This highlights the need for platforms to proactively identify and address fraudulent content rather than relying on user reporting of suspected fakes, which is largely ineffective. Platforms might also consider labeling suspected fake reviews with warning badges or fact-check indicators. Additionally, we find that truth bias is stronger for negative reviews, making fake negative reviews particularly impactful and damaging. Consequently, platforms should prioritize detecting and mitigating fake negative reviews over fake positive ones. Our findings also suggest that structuring reviews into separate positive and negative sections or allowing (or defaulting to) valence-based review sorting might reduce consumer likelihood of being fooled by fake negative reviews. These insights inform platform policy by emphasizing the importance of proactive fraud detection, transparent labeling, and interface design in safeguarding consumer trust and lowering fraud. Despite a growing stream of research documenting the prevalence of “fake” online reviews and improving their automated detection, little is known about how consumers make real or fake judgments of reviews with unknown veracity. Integrating literature on truth-default theory and deception motives, we propose that consumers have a general tendency to view reviews as real rather than fake (a truth bias) and to be more accurate at detecting real reviews than fake reviews (a veracity effect). Moreover, we argue that the truth bias is weaker for positive reviews than negative reviews (a valence effect) because of a largely automatic process in which consumers project deception motives onto reviewers. To test these proposals, we conducted five experiments in which participants classified sets of reviews as real or fake. Results provided broad support for our theorizing, and they have important implications for firms and platforms as they establish priorities for combating fake reviews. History: Choon-Ling Sia, Senior Editor; David (Jingjun) Xu, Associate Editor. Supplemental Material: The online appendix is available at https://doi.org/10.1287/isre.2023.0339 .