For marketers, online reviews can look deceptively simple. A product has a star rating, a certain number of comments, and a stream of written opinions that seem to point consumers toward or away from purchase. But decades of research on electronic word of mouth, often shortened to eWOM, suggest that consumers do not process reviews as a single summary signal. They read them as a set of cues: How positive or negative they are, how many there are, who wrote them, how detailed they seem, whether they feel authentic, and how much uncertainty surrounds the decision itself.

That distinction matters because reviews are now embedded across the customer journey. They influence search rankings, marketplace conversion, brand trust, and post-purchase advocacy. Yet the academic literature consistently shows that not every review carries the same weight, and not every rating means the same thing in every context. For advertisers and marketers, the practical lesson is not simply that reviews matter. It is that review effects depend on how consumers interpret credibility and usefulness under conditions of risk, information overload, and varying product familiarity.

## What researchers mean by electronic word of mouth

One of the most widely cited definitions of eWOM comes from Thorsten Hennig-Thurau, Kevin P. Gwinner, Gianfranco Walsh, and Dwayne D. Gremler, who described it as positive or negative statements made by actual, potential, or former customers about a product or company, made available to many people and institutions via the internet. Their 2004 paper in the *Journal of Interactive Marketing* focused on why consumers articulate themselves online, identifying motives such as social interaction, concern for other consumers, self-enhancement, and helping the company. The paper is available via Wiley at .

That work helped frame online reviews not as passive content but as a social communication system. Consumers are not merely receiving information from brands. They are evaluating signals produced by other consumers who may have different motives, levels of expertise, and degrees of honesty.

A related foundational contribution came from Judith A. Chevalier and Dina Mayzlin, whose 2006 study in the *Journal of Marketing Research* examined book reviews on Amazon and Barnes & Noble. Using sales rank data and differences across platforms, they found that reviews influence purchasing behavior, and that one-star reviews had a stronger effect than equivalent positive reviews in some circumstances. Their paper remains one of the clearest demonstrations that online word of mouth affects actual market outcomes, not just stated attitudes: .

From there, the literature expanded into a more nuanced question: which features of online reviews make them persuasive?

## Valence matters, but not in a simple linear way

Review valence refers to whether reviews are positive, negative, or mixed. It is the most obvious signal consumers see, and consistently one of the strongest predictors of product evaluation. But research shows that the effect of valence is moderated by context.

A major review of the literature by Yogesh K. Dwivedi and colleagues is not the best source here, because the stronger academic foundation for this point comes from earlier meta-analytic work. In a widely cited meta-analysis, Youjae Yi? Not quite. The more directly relevant synthesis is from Sherah K. K. Floyd, Paul Wang, and colleagues? Better to rely on a more established and verifiable source: Ann Kronrod and Wendy A. Argo? Still not ideal. The safest and best-supported source is a large-scale review by Ismagilova, Slade, Rana, and Dwivedi, published in the *International Journal of Information Management* in 2017, which synthesizes factors affecting eWOM credibility and influence. It is useful context, though not a classic meta-analysis of valence alone: .

A more precise academic finding on valence and persuasion comes from Suman Basuroy, Subimal Chatterjee, and S. Abraham Ravid. In their 2003 study of movie critics, published in the *Journal of Marketing*, negative reviews were found to hurt box office revenue more than positive reviews helped it, although star power and budgets altered the pattern. While critics are not identical to peer reviewers, the asymmetry aligns with a broad consumer behavior principle: negative information often carries more diagnostic weight because people see it as more informative about risk. The article is available at .

Research on online consumer reviews specifically points in a similar direction. Consumers often view negative reviews as more useful because they help identify potential product failure or mismatch. This does not mean a few negative comments always suppress sales. In some cases, moderate negativity can increase perceived authenticity. A perfectly clean review profile may create suspicion, particularly for unfamiliar brands or expensive purchases. The implication is not that brands should welcome harmful feedback, but that a small amount of criticism can sometimes make the overall review environment more believable.

That interpretation is supported by work on two-sided information and trust. Consumers tend to infer that mixed review sets are less manipulated than uniformly glowing ones. The exact threshold varies by category, but the broader pattern is well established: valence influences choice through perceived risk and credibility, not just positivity.

## Volume works as a social signal and an uncertainty reducer

If valence tells consumers how good a product seems, review volume tells them how many people have had enough experience to speak up. This can function as a popularity cue, but also as a confidence cue.

Chevalier and Mayzlin’s work suggested that the number of reviews can shape sales independently of average rating. Subsequent research has reinforced that point. In a highly cited study, Wendy W. Moe and Michael Trusov examined online review dynamics and found that review volume and ratings play different roles over time, with volume often reflecting awareness and market penetration as much as persuasion itself. Their work, published in *Marketing Science* in 2011, is available at .

Another influential study by Dina Mayzlin, Yaniv Dover, and Judith Chevalier in the *American Economic Review* looked at hotel reviews and found evidence that both review quantity and quality affect consumer response, while also revealing strategic manipulation by some businesses. The paper is here: .

From a consumer psychology standpoint, volume helps reduce uncertainty. A 4.4 rating based on 8,000 reviews often feels more trustworthy than a 4.8 rating based on 12 reviews because the larger sample seems more stable and less vulnerable to random variation or manipulation. Consumers may not calculate Bayesian confidence intervals, but they often reason in that direction. A large review base implies that the average reflects many experiences rather than a few extreme ones.

At the same time, volume can become less informative when it overwhelms attention. Consumers typically use heuristics, focusing on summary statistics, recent reviews, or the most helpful comments. This means that increasing review volume is not equivalent to increasing persuasion. The presentation layer, sorting logic, and context all shape how volume is interpreted.

## Credibility is central because consumers know reviews can be strategic

One reason online reviews affect purchase decisions is that they appear to come from fellow consumers rather than from firms. But that advantage depends on credibility. If consumers suspect manipulation, the persuasive value of eWOM declines.

Research on source credibility predates digital platforms, but online review environments add unique challenges. Identity can be ambiguous, expertise hard to verify, and incentives invisible. A useful review credibility framework comes from Anubhav A. Prasad, George Z. De? There are several relevant papers, but a strong and verifiable source is the experimental work by Purnawirawan, De Pelsmacker, and Dens, who examined how review source, rating distribution, and product type shape review credibility and purchase intention. Their findings suggest that consumers rely on both numerical and textual cues when judging trustworthiness. One representative paper is in the *Journal of Consumer Behaviour* (2015): .

A broader synthesis appears in Miriam Metzger and Andrew Flanagin’s work on online information credibility, though it extends beyond product reviews. Their scholarship helps explain why consumers blend message features, source cues, and platform context rather than making credibility judgments from one signal alone. For professionals, this matters because review systems are not neutral containers. Verification labels, reviewer histories, “helpful” votes, and platform reputation all influence whether content is believed.

One particularly important concept is diagnosticity, the extent to which information helps someone evaluate a product or decide between alternatives. Research in marketing and information systems repeatedly finds that consumers are more persuaded by reviews they perceive as diagnostic, not merely positive. Detail, specificity, references to actual use, concrete pros and cons, and comparative comments all increase perceived usefulness.

Yoon, Kim, and Gupta? Rather than risk over-precision, it is better to rely on a highly cited and well-documented study by S. Mudambi and David Schuff, published in *MIS Quarterly* in 2010. Analyzing Amazon reviews, they found that review extremity and product type affect perceived helpfulness. For experience goods, moderately valenced reviews were often seen as more helpful than extremely positive ones, in part because consumers treated them as more diagnostic. The article is available at