Quality Verification in the Presence of Review Bias

Abstract

Over time, firms across industries have adopted quality verification as a costly yet credible tool to signal product excellence and convey reliable information to consumers. With the prosperity of E-commerce, consumers increasingly rely on online reviews—an accessible but inherently biased source of product information—to guide their purchase decisions. In practice, companies now frequently combine quality verification with consumer reviews to influence the perception of product quality. This study examines how firms can strategically determine their optimal quality verification approach to mitigate the negative effects of review bias, and how the timing of verification affects outcomes. We consider two quality verification formats: pre-release verification, conducted before reviews are available, and responsive verification, conducted after reviews are observed. Pre-release verification helps shape early consumers’ expectations and improves late consumers’ interpretation of biased reviews. As the magnitude of review bias increases, firms are more likely to proactively adopt pre-release verification, although this leads to a decline in expected profits due to higher upfront costs. By contrast, responsive verification allows firms to selectively react to negatively biased reviews, preserving the upside potential of favorable reviews. When review volatility is high, this reactive strategy can lead to higher profits. Our results show that each verification format can enhance firm profitability under specific conditions, depending on the cost of verification and the degree of review bias. These findings offer actionable insights into how firms can manage information flows and strategically balance third-party verification with user-generated content in dynamic market environments.

1. Introduction

Quality verification is a hallmark of product excellence, and is a longstanding method of disseminating product information to consumers (Brach et al., 2018; Pandanwangi et al., 2023). The nature of quality verification that requires hard evidence or third-party testimony ensures its objectivity. However, because quality verification is initiated and controlled by the firm, it often involves a significant cost. Meanwhile, with the proliferation of online shopping, contemporary consumers also rely on online reviews for product information before making their purchase decisions (Elwalda and Lu, 2016; Maslowska et al., 2017). Compared to the traditional methods of verifying product quality, consumer reviews are free and easy to access, making them one of the most useful information channels for consumers to resolve their quality concerns (Chen and Xie, 2005; Chevalier and Mayzlin, 2006; Forman et al., 2008; Moe and Schweidel, 2012). As early as 2012, a Nielsen survey reported that online reviews were the second most trusted source of brand information globally, following recommendations from friends and family (Nielsen, 2012). In the UK, review aggregator sites have become the most trusted source of honest product opinions, with usage increasing steadily from 2018 to 2020 (Trustpilot, 2024).

Therefore, one might think that consumer reviews will ultimately render traditional verification methods obsolete. However, this paper offers a different perspective. We find that consumer reviews generated by word-of-mouth and quality verification are actually highly complementary to each other, so that the firm’s best practice is to rely on both methods for new product promotion. The key reason is that consumer reviews—while valuable—can be biased and unreliable due to subjective experiences and a firm’s limited control over them (Cabral and Li, 2015; Hu et al., 2009; Joshi and Musalem, 2017; Li and Hitt, 2008, 2010). To mitigate the risk of negative quality inferences stemming from biased reviews, firms can adopt quality verification as a counterbalance. In one real-world example, Agoda.com, one of the largest online third-party travel intermediaries, uses both aggregated consumer reviews for hotels and quality verification from the hotel itself to offer more balanced and substantial information to undecided consumers. As Figure 1(a) illustrates, the homepage of the Gwangalli Gray 193 Hotel displays a “Korea Quality” (KQ) verification label in addition to the hotel’s aggregate consumer rating according to Agoda.com. ¹ In another example, the landlords in Airbnb can decide whether to obtain “Airbnb Plus” verification label to align with consumer reviews to highlight high-quality homes, see Figure 1(b) for illustration. ²

Figure 1.

Examples of revealing quality information via quality verification and consumer reviews together. (a) “Korea Quality” verification and consumer reviews on Agoda.com and (b) “Airbnb Plus” verification and consumer reviews on Airbnb.

The first research question addressed in this paper is how can firms design quality verification strategies that can mitigate the negative impacts of consumer review bias. We begin by assuming that bias in consumer reviews mainly arises from the consumer’s idiosyncratic preferences toward the product/service, a trend especially prevalent in categories such as hotel accommodation, snack food, and other experience goods/service (Clemons et al., 2006; Hu et al., 2009; Li and Hitt, 2008). Even though the consumers in these markets are rational, their preferences can be highly diversified by product type or brand image, evolving quickly over time. As a result, the comments left for products are usually based on these subjective suitability criteria that do not accurately reflect the product’s actual quality. This unique characteristic of consumer reviews—informative yet biased—motivates our exploration of its strategic interaction with the firm’s traditional quality verification strategy.

We are also interested in what are the implications of different verification options when consumer reviews are biased? This interest is motivated by the anecdotes regarding the different timing of quality verification and the realization of consumer reviews. Usually, a brand offers verification of product quality upfront as part of its new product promotion process, because doing so can not only attract early arriving consumers but also mitigate the potential for negative inference from biased consumer reviews after they are posted. It is also plausible, however, that a firm obtains quality verification only after observing consumer reviews, as was the case when customers complained of gastrointestinal discomfort after consuming Soylent Foods snack bars containing algal flour supplied by TerraVia. In order to mitigate the potentially reputation-damaging effects of the negative perception of the reliability of its products, TerraVia responded by presenting a “no objection” letter from the U.S. Food & Drug Administration verifying the Generally Recognized as Safe designation for the implicated flour. ³ Similarly, when consumers posted online that their cats got sick after eating Canadian Go! brand cat food, with some even reporting their cats had died, the manufacturer had an authoritative third-party inspection agency to test the products involved and provide an official report validating the good quality of the product in question. ⁴ Although the ubiquity of online reviews and their influence on the marketplace is making responsive verification strategies such as these, more common, there is still scant academic research addressing the topic and its implications for firms and consumers, making this study so timely and original.

To address these research questions, we design a two-period model for quality information revelation involving a monopoly firm with private knowledge of product quality and two groups of consumers arriving in sequence. The initial period serves as the phase for aggregating reviews, wherein early arriving consumers make purchase decisions based on prior beliefs about quality, and those consumers who purchase the product leave reviews based on their usage experiences. Period 2 can be considered as the review revelation period, in which all consumer reviews are aggregated into a weighted rating, with late-arriving consumers making purchase decisions predicted on the realized review rating. In this paper, we define two quality verification formats according to the timing of quality verification. With pre-release quality verification, the firm obtains quality verification at the beginning of the first stage, while with responsive quality verification, the firm adopts quality verification to rectify negatively biased consumer reviews after they are released. In our game setting, quality verification is a truthful information revelation method that requires hard evidence to demonstrate, while it is also imperfect in that the revealed quality information is restricted to a range of quality levels. Due to this limitation, consumers cannot solely rely on a single information source, whether quality verification or consumer reviews, to accurately infer product quality.

We focus on two distinct scenarios where the firm is restricted to exclusively using either pre-release or responsive verification to regulate the consumers’ quality inference. We show that under the pre-release scenario, review bias plays a pivotal role in determining the firm’s optimal quality verification strategy. When the review variance is low, consumer reviews serve as a free, credible information source, such that the firm relies more on them than on quality verification to promote its product. The firm’s verification strategy remains the same as that when the consumer reviews are objective, whose functionality is to enhance the early adopters’ quality expectation. Nonetheless, when the review variance is high, consumer reviews may deviate greatly from the true product quality, in which case the firm becomes more proactive to obtain quality verification than in the scenario of objective consumer reviews. By doing so, the firm not only improves the early adopters’ quality expectation, but more importantly, it helps late-arriving consumers better infer the product quality from the biased reviews. In equilibrium, the firm’s expected profit monotonically decreases in the review variance, as when consumer reviews are less accurate, their ability to signal quality information diminishes, making it more challenging for the followers to accurately infer true product quality.

Under the scenario of responsive verification, although the firm cannot use verification to enhance the early arriving consumers’ quality expectations, it is able to rectify the late-arriving consumers’ negative quality inference when they encounter negatively biased reviews. This endows the firm with more flexibility to adjust its verification strategy in response to the realized reviews. In particular, we show that in equilibrium the firm obtains responsive verification only if all three conditions hold: (1) the verification cost is low, (2) the aggregated consumer review is negative and contains a high magnitude of variance, and (3) the product exceeds the quality standard. More interestingly, different from the scenario of pre-release verification, in which the firm is consistently hurt by the review variance, the firm may benefit from a higher review variance under the scenario of responsive format. This is because with a higher level of review variance, the consumers can generate a quite high-quality expectation once the review is biased toward the positive; if the review is biased toward the negative, the firm can use responsive verification to mitigate such a downside on the consumers’ quality inference process.

Although both verification formats can effectively mitigate the variance in consumer reviews, their impacts on the firm’s profitability react differently. The main functionality of pre-release verification is to enhance the early arriving consumers’ quality expectation and to improve the late-arriving consumers’ quality inference from biased consumer reviews. However, its drawback is the potential for unnecessary verification costs if initial consumer reviews turn out to be positive. Conversely, the responsive verification strategy proves more cost-effective, as it enables the firm to initiate quality verification solely in response to negatively biased consumer reviews, though this comes at the expense of potentially lower first-period profits. Combining these conflicting effects together, we show that when the verification cost is low that the impact for cost saving is negligible, pre-release format is the dominant option. Otherwise, the firm’s expected profit is higher with responsive verification. We further find that the firm’s preference for responsive verification rises when the magnitude of review bias is high. This is because the volatility of consumer reviews can amplify the benefit of a responsive format, in which the firm can use it to mitigate the downside of negative reviews but retain the benefit of positive reviews. For the pre-release verification format, the firm has to undertake it before the reviews are realized, so that the impact of review variance is mitigated by the ex ante uncertainty of consumer reviews.

We also extend our baseline model to examine the robustness of our main results. We consider five scenarios: the firm endogenously determines quality verification timing; the consumers may be more likely to share their positive (respectively, negative) consumption experiences than negative (respectively, positive) ones; the followers are naive in a way that simply take the aggregated consumer review as the true product quality; the followers are able to intuit the true product quality level when the number of reviews is sufficiently large; and the consumers may exhibit reference-dependent preference in their reviews. We find that our main results remain qualitatively similar across all these scenarios. Overall, our comparative results are informative to firms deciding how to design the appropriate quality verification mechanism to mitigate the negative effects of variance and bias in the consumer reviews.

The remaining sections of this paper are structured as follows. Section 2 provides a comprehensive review of the existing related literature. In Section 3, we present the model setup and benchmark. Section 4 compares the firm’s pre-release and responsive quality verification strategies. Section 5 examines five model extensions, with concluding remarks provided in Section 6. All proofs are provided in the Supplemental Appendix.

2. Literature Review

Our study draws from the current body of research to contribute to three primary streams of literature: (i) marketing strategy with consumer reviews, (ii) voluntary information disclosure, and (iii) quality verification.

Our paper first contributes to the existing marketing literature devoted to studying how firms make strategic use of consumer reviews to increase revenue. We proceed from the work of scholars such as Chen and Xie (2008), who examine how companies should align their release of information, such as regarding product attributes, in response to early consumers’ product reviews, and Kuksov and Xie (2010), who consider the firm’s optimal pricing and promotional strategy of offering bonus frills. Papanastasiou and Savva (2017) study pricing strategy in response to strategic consumers who learn about quality from their peers, demonstrating that social learning can exacerbate strategic delay and benefit the firm. Feng et al. (2019) investigate a firm’s dynamic pricing strategy with consumer reviews and shows the firm can not only influence online product reviews through pricing, but also adjust pricing dynamically in reply to those reviews. In a closely related study, Joshi and Musalem (2017) consider a setting where a firm can signal product quality through advertising under the word of mouth (WOM) effect. Other scholars have also studied the influence of reviews on firms’ pricing and selling strategies with market competition, such as Kwark et al. (2014), Liu et al. (2017), Li et al. (2018), and Ye et al. (2025).

This paper is distinct from the above-mentioned studies for its focus on a firm’s quality verification strategy in response to the bias present in consumer reviews. Unlike a traditional quality signaling tool, such as paid advertising, quality verification represents a credible method for revealing information that can directly alter a consumer’s perception of product quality. Although a few recent studies have explored management response strategies with online reviews (Chen et al., 2019; Kumar et al., 2018), they are from an empirical perspective with mechanisms entirely different from this paper’s focus on the firm’s responsive verification.

Our paper also contributes to the literature on voluntary information disclosure. This literature concentrates mainly on the optimal disclosure strategy adopted by firms and its impact on profits under various market conditions (Grossman and Hart, 1980; Guan and Chen, 2017; Guan et al., 2020; Guo, 2009; Jovanovic, 1982; Matthews and Postlewaite, 1985; Milgrom, 1981; Shavell, 1994; Xu et al., 2024). In this stream of literature, quality disclosure reveals information in a truthful and definitive way. However, our paper takes the unique perspective of identifying the firm’s optimal quality verification strategy of engaging a third-party verification agency to systematically confirm whether the product quality is higher than a certain standard set in advance. The nature of quality verification means that it only reveals partial quality information to consumers. Guan et al. (2020) are among the first to consider a firm’s voluntary disclosure strategy with consumer reviews, which is most related to ours. However, review bias in their paper is driven by consumers’ reference-dependent preference, and so the sophisticated followers are able to strategically rectify the biased quality review to its original level. They mainly investigate how consumers’ reference-dependent preferences affect a firm’s quality disclosure decision. In our study, we assume that review bias is driven by consumers’ idiosyncratic preferences and fit with the product, which makes it hard for followers to infer the true product quality. Moreover, the main focus of our paper is to investigate how the firm can design its quality verification strategy accordingly to reduce review bias. We consider two quality verification formats, pre-release and responsive, and show that under the scenario of pre-release verification, the firm’s quality verification strategy is determined according to its expectations for any possible outcomes of review bias, either positive or negative. When the magnitude of review bias increases, the firm becomes more proactive in obtaining pre-release quality verification. In contrast, by adopting responsive verification, the firm has more flexibility to devise a strategy after observing the biased consumer reviews, and so would only need to pursue quality verification if the reviews are negatively distorted.

The third stream of literature this paper draws from and contributes to is quality verification. In some leading studies in this field, Zapechelnyuk (2020) construct a quality verification model in a moral hazard setting and indicate that simple verification systems such as quality assurance and pass–fail rules are optimal. Iyer and Singh (2018) explore the impact of the consumer moral hazard on the firms’ verification incentive and show that its impact depends highly on the relationship between the inherent product safety and the firm’s effort. Marinovic et al. (2018) find that when product quality is persistent, and verification is costly, firms may fall into the oververification trap in order to build and maintain their quality reputation. Bian et al. (2022) discover that the impact of quality verification on the farming cooperative’s profit hinges on three factors: the quality standard, the cost of verification, and the precision rate of government information, and its profit exhibits a non-monotonic relationship with any of these factors. Some other scholars conduct empirical studies to explore the impact of quality verification on consumer demand varies (Elfenbein et al., 2015) and firm performance (Dewan et al., 2023). Hui et al. (2023) further investigate the impact of the number of verification tiers on market outcomes. Our work differs from the aforementioned studies by investigating the implications of verification timing in the presence of review bias, which has generated a series of non-trivial results that complement the prior literature. We show that the benefit of pre-release verification format is to enhance the early arriving consumers’ quality expectation and to improve the late-arriving consumers’ quality inference from biased consumer reviews. However, its drawback is the potential for unnecessary verification costs if initial consumer reviews turn out to be positive. Conversely, the responsive verification strategy proves more cost-effective, as it enables the firm to initiate quality verification solely in response to negatively biased consumer reviews, though this comes at the expense of potentially higher first-period profits.

3. Model and Benchmark

3.1. Model Setup

The main difference between quality verification and consumer reviews as promotional information channels is that quality verification is sought after proactively by the firm and requires hard evidence to demonstrate its objectivity, while an online review is generated spontaneously by the consumers and thus reflects personal preferences that are usually biased.

To capture these characteristics, our model considers a firm selling a new experience product to consumers over two periods. In each period $i$ , $i=1,2$ , a group of consumers arrive and make the purchase decision (Tan, 2024). Period 1 can be understood as the review aggregation period, in which early arriving consumers make purchase decisions based on their prior quality belief, and the consumers who acquire the product then give their reviews based on their usage experiences. Period 2 can be viewed as the review revelation period, in which all consumer reviews are aggregated into a weighted rating, and late-arriving consumers make a purchase decision based on the realized review rating. In line with practice, we focus on two quality verification formats in this paper: (1) Pre-release Quality Verification, in which the firm obtains quality verification in period 1, which is before consumers generating reviews, and (2) Responsive Quality Verification, in which the firm adopts quality verification in period 2 that is after consumer reviews are published. To understand and isolate the impacts of two verification formats, in our primary analysis, we consider two distinct scenarios where the firm is restricted to exclusively using either pre-release or responsive verification. In Section 5.1, we further consider a scenario in which the firm endogenously determines quality verification timing.

Consumer Purchase Decisions. Without loss of generality, we term the mass of consumers in period 1 as the early adopters, and in period 2 as the followers (Joshi and Musalem, 2017). We also normalize the market size in period 1 to $1$ and in period 2 to $m$ , which captures the market expansion or contraction in period 2. We assume that $m<M$ ( $M>1$ ) to ensure the impact of the firm’s profit in period 1 to be non-trivial. In each period, consumers make their purchasing decisions based on the comparison between their purchasing utilities from the firm and the outside option. That is,

$$U_i={\overline{q}}_i-p_i\ge \theta ,i=1,2.$$

(1)

In (1), ${\overline{q}}_i$ and $p_i$ represent the consumers’ quality expectation and the retail price of period $i$ , respectively. $\theta$ represents consumers’ outside option, whose value follows a uniform distribution between $[0,1]$ , which denotes consumer heterogeneity in the outside option (Zapechelnyuk, 2020). As such, a consumer would purchase the product only when they derive a higher utility than with the outside option, $\theta \le {\overline{q}}_i-p_i$ , which leads to the inverse demand function for the firm in each period, or, $D_i=P_r(\theta \le {\overline{q}}_i-p_i)={\overline{q}}_i-p_i$ .

As it is used in our paper, the term “quality” is a mixed concept encapsulating all the vertical attributes of a product affecting its perceived value (i.e., performance, reliability, and functionality). In line with a wide range of literature (Grossman and Hart, 1980; Guan et al., 2020; Guo, 2009), we consider product quality, denoted as $q$ , as a naturally determined random variable, whose value follows a uniform distribution between zero (lowest quality) and one (highest quality), $q\sim U[0,1]$ . The firm privately observes this quality information, while consumers cannot ascertain its value immediately before consumption. This is especially applicable to experienced products, for which consumers still have to rely on the experienced consumers’ reviews and the firm’s quality verification strategy to update their beliefs about product quality.

Consumer Reviews. To capture the personal bias inherent in consumer reviews, we assume that after using product, the early adopters’ aggregated review is given by $\mathbb{R}=q+\epsilon$ , where $q$ is the product quality and $\epsilon$ is a random variable that captures the review variance (caused by the consumers’ idiosyncratic preferences toward the product) with respect to the exact product quality. ⁵ $\epsilon$ follows a uniform distribution that $\epsilon \sim U[-v,v]$ , and a higher $v$ indicates a higher magnitude of review variance. $\mathbb{R}$ can be understood as an aggregated review from all early arriving consumers, with $\mathbb{R}\in [0,1]$ , indicating that consumer reviews should fall between the lowest product quality and the highest product quality. The aggregated review is quite common in practice, with most review systems including some form of ratings boundary, so that even if a consumer feels extremely good or bad about their experience, they cannot give a score outside the review boundary. In our baseline model, we assume the followers are sophisticated enough to rationally infer the product quality from the biased reviews (Tan et al., 2024). Later in Section 5.3, we consider an alternative case with naive consumers who are not sensitive to the review bias, and in a scenario in Section 5.4, the followers are able to infer the exact product quality once there is a sufficient number of reviews.

Figure 2.

Timeline of pre-release verification scenario.

Figure 3.

Timeline of responsive verification scenario.

Quality Verification. In this study, we adopt the most common verification rule, pass–fail verification (Zapechelnyuk, 2020), in which there exists a quality threshold value $b$ such that the consumer can be informed whether the product quality is higher or lower than $b$ (pass or fail). Typical examples include national standards, such as “Korea Quality” and “Airbnb Plus” badges in our motivating example. Here, $b$ can be regarded as the quality requirement for obtaining quality verification, which is exogenously determined by authorities or the verification agency. Obtaining quality verification indicates that a firm’s product quality is equal to or above $b$ , or $q\ge b$ , in response to which consumers will update their belief that the product quality slips into another uniform distribution between $b$ and $1$ , or $q\sim U[b,1]$ . ⁶ We use $c$ to denote the cost of efforts exerted by the firm to obtain quality verification from a third-party organization. Keeping in line with the related literature (Bian et al., 2022; Iyer and Singh, 2018) and practice, we assume both verification standard $b$ and cost $c$ are common knowledge ⁷ and the verification cost $c$ is constant and independent of the quality standard $b$ . ⁸

Due to the nature of quality verification, it only partially reveals the product information to consumers, improving quality expectation and reducing the product quality variance. This assumption departs from the existing literature’s view that voluntary disclosure perfectly reveals quality information (Grossman and Hart, 1980; Milgrom, 1981) and so better captures the practice. If the firm does not achieve quality verification, consumers infer product quality information in a process qualitatively similar to what is found in the literature on voluntary disclosure. This process will be explained in detail later in the paper.

Sequence of Events. Under the scenario of pre-release quality verification, the firm with private quality information first decides whether to reveal its high product quality by obtaining costly quality verification at the beginning of period 1 and then offers retail price $p_1$ to the early adopters. Then, after observing the firm’s quality verification decision, the early adopters update their quality belief about the product and make the purchase decision. Next, after consumption, the early adopters who buy the product generate an aggregated quality review observable to both the firm and the followers. And finally, the firm decides its retail price $p_2$ in period 2, and the followers update their quality belief and make the purchase decision. The game sequences of this scenario are illustrated in Figure 2. The timeline under the scenario of responsive quality verification is similar, with the only difference that the firm cannot obtain quality verification at the beginning of period 1, instead of deciding whether to do it after the consumer reviews are released. Please refer to Figure 3 for illustration.

Both the firm and consumers are risk neutral with a view to maximizing their own surplus. Without loss of generality, the firm’s marginal operational cost is normalized to zero. To avoid some trivial discussions, we assume that $b\ge 1/2$ and $v\le 1-b$ , indicating that the prior quality expectation should be lower than the quality verification standard and the review variance is not too significant. Table 1 summarizes the main notations in our paper. We apply backward induction to ensure subgame perfection.

Table 1.

Notations in our main analysis.

Notation	Description
$\theta$	Consumers’ outside option, $\theta \sim U[0,1]$
$m$	Market size in period 2
$b$	Quality verification standard
$c$	Quality verification cost
$q$	Product quality, $q\sim U[0,1]$
${\overline{q}}_i$	Consumer quality expectation in period $i$ , $i=1,2$
$\hat{q}$	Consumers’ updated quality belief without quality verification ( $\hat{q}\in \{b,1\}$ )
$p_i$	Retail price in period $i$
$D_i$	Demand in period $i$
${◻}^{M,N}$	Equilibrium outcomes with consumer review type $M$ ( $M\in \{O,S\}$ ) and under the verification policy
	$N$ ( $N\in \{P,R\}$ ), where superscript $O$ (respectively, $S$ ) represents
	objective (respectively, subjective) consumer reviews and superscript $P$ (respectively, $R$ )
	represents pre-release (responsive) quality verification
$c^{M,N}$	Quality verification cost threshold
$\epsilon$	Review variance with respect to the exact product quality, $\epsilon \sim U[-v,v]$
$\mathbb{R}$	Reviews by early adopter, $\mathbb{R}\in [0,1]$
${\pi }_{i,d/nd}^{M,N}$	Firm’s profit in period $i$ with ( $d$ ) or without ( $nd$ ) quality verification
${\mathrm{\Pi }}^{M,N}$	Firm’s expected profit
${\text{CS}}^{M,N}$	Consumer surplus
Extension
$k$	Demand threshold between review bias and no bias
$\delta$	Magnitude of consumer reference-dependent preferences

3.2. Benchmark: Objective Consumer Reviews

We first consider a benchmark scenario wherein the early adopter generates a post-consumption quality review that truthfully reflects the product quality, that is, $\mathbb{R}=q$ . While this is an idealistic situation, it isolates the impact of review bias. We begin with the scenario of pre-release quality verification, where the firm decides whether to obtain quality verification at the beginning of period 1. Given that the followers can infer the product quality based on consumer reviews, the firm sets $p_2=q/2$ to maximize its profit ${\pi }_2^{O,P}=m{q}^2/4$ in period 2. However, the firm still needs to determine whether to seek quality verification in period 1 in order to shape the early adopters’ perception of quality. In equilibrium, the firm will adopt quality verification to inform consumers of its high-quality information if doing so is profitable. If the firm decides to obtain the quality verification, the early adopters will infer the product quality is above $b$ , that is, $q\sim U[b,1]$ . The firm sets the retail price to maximize its first-period profit, that is, $max$ $p_1({\overline{q}}_1-p_1)$ , wherein ${\overline{q}}_1=(1+b)/2$ , making it evident that the firm’s optimal retail price becomes $p_1=(1+b)/4$ and subsequently leading to the first-period profit $\pi$ ${}_{1,d}^{O,P}$ with quality verification,

$$\pi {}_{1,d}^{O,P}=\frac{(1+b{)}^2}{16}-c,$$

in which subscript $d$ denotes the subcase of adopting quality verification, superscript $O$ denotes the scenario of objective consumer reviews and superscript $P$ denotes the scenario of pre-release quality verification. If the firm does not obtain quality verification, the early adopters will believe this to be caused by one of two reasons: either $q$ is lower than the quality verification standard $b$ so that the firm is unable to obtain quality verification; or, as the expected profit from obtaining the quality verification falls short of covering its cost, the firm has made the decision to abandon obtaining the quality verification. In the first case, the early adopters update their prior belief to $q\sim U[0,b]$ ; while in the second case, the early adopter continue to hold their previous belief that $q\sim U[0,1]$ . Given the early adopter’s updated quality belief, the firm’s first-period profit without quality verification (denoted by $nd$ in the superscript) ${\pi }_{1,nd}^{O,P}$ is

$$\pi {}_{1,nd}^{O,P}=\frac{{\hat{q}}^2}{16},$$

wherein $\hat{q}$ is the early adopters’ updated quality belief without quality verification, that is, $\hat{q}=b$ or $1$ depending on the firm’s quality verification strategy.

We use a rational equilibrium concept to resolve this question. Note that in equilibrium, the firm will adopt quality verification if and only if it is profitable, that is, ${\pi }_{1,d}^{O,P}>{\pi }_{1,nd}^{O,P}$ , and the consumers’ rational inference should be consistent with the firm’s verification strategy. Thus, the equilibrium condition for the firm to adopt quality verification under the scenario of pre-release verification is

$$\begin{array}{rl}{\pi }_{1,d}^{O,P}& =\frac{(1+b{)}^2}{16}-c>{\pi }_{1,nd}^{O,P}=\frac{1}{16}\to c<c^{O,P}=\frac{b(b+2)}{16},\end{array}$$

(2)

wherein $c^{O,P}$ is the quality verification cost threshold with objective consumer review under the pre-release quality verification format. Otherwise, if $c\ge c^{O,P}$ , the firm will forgo verification.

Under the scenario of responsive verification, it is evident that the firm never obtains quality verification when the consumer reviews are objective, which highlights the different functionality of the two verification formats. Note that the main function of responsive quality verification is to mitigate the followers’ negative perception formed from the early adopters’ quality reviews. However, given that now the reviews are unbiased, the firm certainly has no incentive to obtain quality verification either. We then conclude the firm’s equilibrium verification strategies with objective reviews under two scenarios in the following lemma, whose results will be compared with the core scenarios with review bias.

Lemma 1

When consumer reviews are objective, in equilibrium,

Under pre-release verification, the firm obtains quality verification when $c<c^{O,P}$ .

Under responsive verification, the firm never obtains quality verification.

4. Analysis

In this section, we examine the impacts of different quality verification formats on the firm’s profit and consumer surplus when the consumer reviews are subjective (i.e., in the presence of review bias). We first identify the firm’s equilibrium quality verification strategies under each scenario, and then compare the firm’s expected profits under the two scenarios. In the rest of the paper, we use superscript “ $S$ ” to denote the solution with subjective consumer reviews, superscript “ $P$ ” to denote the solution under the scenario of pre-release verification, and superscript “ $R$ ” to denote the solution under the scenario of responsive verification.

4.1. Scenario of Pre-Release Quality Verification

We begin with the scenario of pre-release verification where the firm decides whether to obtain quality verification at the beginning of period 1, in anticipation of the possible bias in the early adopters’ reviews, see Figure 2 for illustration. To derive the equilibrium quality verification strategy, we need to compare the firm’s profits with and without verification, respectively. In the following, we first discuss the subgame in which the firm does not pursue quality verification. If in equilibrium the firm does not obtain verification, neither consumer segment updates their prior beliefs about product quality upon observing this absence. With backward induction, the followers observe an aggregated consumer review from the early adopters. We assume the followers are sophisticated enough to rationally infer precise product quality based on the review as $\mathbb{R}=q+\epsilon$ . That is, given the quality review $\mathbb{R}$ , the followers infer the quality $q$ should follow a uniform distribution that

$$q\sim U[max\{0,\mathbb{R}-v\},min\{\mathbb{R}+v,1\}].$$

(3)

The above equation characterizes the followers’ de-bias process, the rationale of which is as follows. When consumer reviews fall into an intermediate range, that is, $max\{b-v,v\}<\mathbb{R}<1-v$ , given the possible value of review bias $\epsilon \sim U[-v,v]$ , the followers can directly infer the range of accurate product quality as $q\sim [\mathbb{R}-v,\mathbb{R}+v]$ . However, when the consumer reviews are negatively biased so that $b-v<\mathbb{R}<max\{b-v,v\}$ , the followers infer that the review bias follows the uniform distribution between $-v$ and $\mathbb{R}$ , so that $q\sim U[0,\mathbb{R}+v]$ . If the consumer reviews are quite positive, such that $\mathbb{R}>1-v$ , the followers can then infer the true quality $q\sim U[\mathbb{R}-v,1]$ . It should also be noted that given $q\ge b$ , it is impossible for the early adopters to produce a negative review below $b-v$ (i.e., $\mathbb{R}<b-v$ ), even without verification.

The firm also observes the early adopters’ aggregated quality review $\mathbb{R}$ and adjusts its second-period price $p_2$ in accordance with the followers’ updated quality expectation ${\overline{q}}_2$ , that is, $p_2={\overline{q}}_2/2$ . As in period 1, since the early adopters would not update the quality expectation upon no verification, the firm’s retail price equals half of the early adopters’ prior quality expectation, so that $p_1={\overline{q}}_1/2=1/4$ . We can then derive the firm’s profit without quality verification, with the derivation process provided in detail in Supplemental Appendix A.

We next discuss the subgame where, in equilibrium, the firm obtains quality verification. Given this verification, both types of consumers first update quality belief to $q\sim U[b,1]$ . Unlike early adopters who must rely on the firm’s verification to form their perceptions of product quality, the followers are able to infer more information after observing the quality review $\mathbb{R}$ . In the updating process, if the followers observe consumer reviews $\mathbb{R}$ so that $\mathbb{R}\subset [b-v,b+v]$ , then they will infer that the true quality $q$ is uniformly distributed between $b$ and $min\{\mathbb{R}+v,1\}$ , where

$$q\sim U[b,min\{\mathbb{R}+v,1\}].$$

(4)

If the consumer review is quite positive, where $\mathbb{R}>b+v$ , the followers’ updated quality belief of $q$ follows a uniform distribution such that $q\sim U[\mathbb{R}-v,min\{\mathbb{R}+v,1\}]$ . In this case, quality verification does not influence the followers’ inference process. Similarly, given $q\ge b$ , it is impossible for the early adopter to generate a negative review below $b-v$ , or $\mathbb{R}<b-v$ .

It can be inferred from this consumer’s quality belief updating process that the firm’s retail prices over two periods are $p_1=(1+b)/4$ and $p_2={\overline{q}}_2/2$ , and we can easily derive the firm’s profit with quality verification. The specific value is provided in Supplemental Appendix A. The firm’s equilibrium pre-release verification strategy with subjective consumer reviews is derived by comparing the firm’s profits with and without quality verification.

Proposition 1

When consumer reviews are subjective and under the scenario of pre-release quality verification, in equilibrium,

The firm obtains quality verification when $c<c^{S,P}$ , where

$$c^{S,P}=\{\begin{array}{ll}{\displaystyle \frac{b(b+2)}{16},}& {\displaystyle \text{if }v<\frac{(1-b)}{2},}\\ {\displaystyle \frac{b(b+2)}{16}+\frac{m(b-v+2){(b+2v-1)}^2}{48v},}& {\displaystyle \text{if}v\ge \frac{(1-b)}{2}.}\end{array}$$

Otherwise, the firm will not obtain quality verification.

The firm’s incentive for obtaining verification is (weakly) higher than with objective reviews, that is, $c^{O,P}\le c^{S,P}$ .

In Proposition 1, it is clear that review bias exerts a non-trivial effect on the firm’s optimal quality verification strategy, that is, the firm becomes more likely to obtain quality verification with subjective consumer reviews. If the magnitude of review bias is low ( $v<(1-b)/2$ ), consumer reviews of high-quality products will not be highly distorted even if the review bias is negative. Here, the main function of quality verification is to enhance the early adopters’ quality expectations. In other words, quality verification does not change the followers’ updated quality level, so the firm disseminates information about the product quality to the followers through consumer reviews. Under such circumstances, the firm’s verification strategy remains the same as objective consumer reviews.

However, when the magnitude of review bias increases ( $v\ge (1-b)/2$ ), compared to the scenario of objective consumer reviews, the firm becomes more proactive about obtaining quality verification, whose incentive monotonically increases with the magnitude of review bias, or $\mathrm{\partial }{c}^{S,P}/\mathrm{\partial }v>0$ . Intuitively, when the volatility of reviews increases, followers would reduce their quality expectations more significantly after exposure to the negative reviews. In this case, firms with high product quality are more vulnerable to negative reviews than low-quality firms. Therefore, it is necessary for the high-quality firm to obtain quality verification upfront, both to prevent a negative quality inference and to lead the rational followers to infer that the negative reviews are driven by the early adopters’ review bias rather than poor intrinsic product quality. The above results lead us to the following managerial insight. In the presence of review bias, if the new product quality is high and relatively niche (corresponding to a large $\epsilon$ ), the firm should pursue quality verification more proactively once entering the market to alleviate the potential consequences of negative post-entry feedback.

We proceed to analyze the influence of review bias on the firm’s profitability. Building upon the firm’s equilibrium verification strategy outlined in Proposition 1, we then derive the firm’s expected profit ${\mathrm{\Pi }}^{S,P}$ under pre-release quality verification format, where ⁹

$${\mathrm{\Pi }}^{S,P}=\{\begin{array}{ll}{\displaystyle \frac{(1+b-b^2)}{16}+\frac{m(16-20v^2+35v^3)}{192}-c(1-b),}& {\displaystyle \text{if}v<\frac{(1-b)}{2}\text{ and }c<c^{S,P},}\\ {\displaystyle \frac{(1+b-b^2)}{16}+G_1(m,b,v)-c(1-b),}& {\displaystyle \text{if}v\ge \frac{(1-b)}{2}\text{ and }c<c^{S,P},}\\ {\displaystyle \frac{1}{16}+\frac{m(16-20v^2+19v^3)}{192},}& \text{ otherwise.}\end{array}$$

(5)

Corollary 1

When consumer reviews are subjective, and under the scenario of pre-release quality verification, the firm’s expected profit monotonically decreases in review variance, $\mathrm{\partial }{\mathrm{\Pi }}^{S,P}/\mathrm{\partial }v<0$ .

Corollary 1 identifies the impact of review bias on the firm’s expected profit, which monotonically decreases in the review variance, no matter whether the firm adopts quality verification or not in equilibrium. Intuitively, the firm’s expected profit hinges on the information transparency level over two periods, in which the firm can obtain a higher profit once the quality information is more transparent. When the review variance increases, the consumer reviews are less accurate; their ability to signal quality information diminishes, making it more challenging for the followers to accurately infer true product quality. In extreme cases where the review variance is exceptionally high, the reviews become uninformative. Thus, the firm’s expected profit decreases.

4.2. Scenario of Responsive Quality Verification

In this subsection, we study an alternative scenario where the firm decides whether to obtain quality verification after observing consumer reviews, see Figure 3 for illustration. We will investigate how the firm can use this strategy to encourage followers to purchase in the presence of review bias and how it affects the firm’s profitability.

When the firm can only make its verification decision after the reviews are realized, the early adopters decide whether to purchase only according to their prior beliefs about product quality (i.e., $q\sim U[0,1]$ ), and then generate an aggregated review $R$ based on personal experience, $\mathbb{R}\subset (q-v,q+v)$ . After the reviews are released, the followers first update their quality beliefs, taking into account that the firm has the option to verify quality to influence their perception. If the firm obtains quality verification, when combined with the consumer reviews, the followers’ updated quality inference is given by

$$q\sim U[max\{\mathbb{R}-v,b\},min\{\mathbb{R}+v,1\}].$$

(6)

Should no verification be obtained, then the followers’ updated quality inference is

$$\begin{array}{r}q\sim U[max\{0,\mathbb{R}-v\},min\{\mathbb{R}+v,\hat{q}\}],\text{ wherein }\hat{q}=b\text{ or }1.\end{array}$$

(7)

Note that in (7), the followers infer the lack of verification to be caused by either low product quality ( $q<b$ ) or high verification cost. This then leads to two possible outcomes for product quality shown in (7), $\hat{q}=b$ or $1$ .

Given the followers’ quality updating process, it is evident that the firm will never obtain quality verification when (1) $\mathbb{R}\ge b+v$ , and (2) $\mathbb{R}\le b-v$ . In the first case, consumer reviews are sufficiently positive so that there is no need for the costly verification, which will not further improve the followers’ quality expectation. In the second case, the reviews are very negative, reflecting a product of low enough quality that the firm is unable to obtain verification. Excluding these two cases, the firm might pursue quality verification if the consumer reviews fall into an intermediate range of $b-v<\mathbb{R}<b+v$ , and it is only in this range that the followers’ quality inference can be affected by the firm’s verification strategy. The following proposition gives the firm’s optimal responsive quality verification strategy, for which the derivation process is detailed in Supplemental Appendix A.

Proposition 2

When consumer reviews are subjective, and under the scenario of responsive quality verification, in equilibrium, the firm obtains quality verification when

$$(1)c<c^{S,R}=\{\begin{array}{ll}\frac{mv(2b-v)}{4}\text{,}& ifv<\frac{b}{2},\\ \frac{bm(b+4v)}{16}\text{,}& ifv\ge \frac{b}{2},\end{array}$$

and

$$(2)max\{b-v,\frac{16c-m{b}^2-2bmv}{2mb}\}<\mathbb{R}<{\mathbb{R}}^{S,R},$$

where

$$\begin{array}{rl}{\mathbb{R}}^{S,R}& =min\{\frac{2\sqrt{m^2(b+v{)}^2-12cm}+m(b+v)}{3m},\\ & \frac{m(v-1)+\sqrt{m^2(1+b{)}^2-16mc}}{m}\}.\end{array}$$

Proposition 2 shows that in the scenario of responsive quality verification, the firm has the incentive to obtain quality verification when the reviews are subjective. The intuition is that the consumer reviews may be negatively biased, which diminishes the followers’ quality expectation of a firm with high product quality; therefore, the firm has to obtain quality verification to correct the followers’ quality expectation. In equilibrium, the firm will obtain quality verification in response to the consumer reviews only when the following conditions all hold. The first condition is that the cost of responsive verification $c$ must be relatively low or the magnitude of review variance must be relatively high, under which case the benefit of obtaining verification can cover its corresponding cost. When the review variance is high, the followers’ quality inference is inaccurate enough that the firm has the incentive to mitigate the variance and improve product quality expectations for its products through verification. Otherwise, if the review variance is low, the followers’ quality inference is close to the true quality level, leaving little space for the firm to influence consumer perceptions about product quality. We can thus infer that the firm’s verification incentive increases monotonically in the review variance, or $\mathrm{\partial }{c}^{S,R}/\mathrm{\partial }v>0$ . The second condition is that the consumer reviews must fall into an intermediate–low rating range. As mentioned earlier, if the aggregated consumer rating is quite low ( $\mathbb{R}<b-v$ ), the product quality is insufficient for the firm to attain quality verification. The consumer ratings also need to be lower than an upper bound ${\mathbb{R}}^{S,R}$ to ensure the benefits of quality verification can cover the firm’s verification cost.

Recall Proposition 1, the equilibrium verification strategy is markedly different from that under the scenario of pre-release verification, where the firm’s quality verification strategy is determined according to its expectations for any possible outcomes of review bias, either positive or negative. In contrast, by adopting responsive verification, the firm has more flexibility to devise a strategy after observing the biased consumer reviews, and so would only need to pursue quality verification if the reviews are negatively distorted.

Corollary 2

When consumer reviews are subjective and under the scenario of responsive quality verification, in equilibrium, the firm’s expected profit first decreases and then increases in review variance $v$ when $b<b_1$ or $c<c_1$ ; it always decreases in $v$ when $b\ge b_1$ and $c>c_2$ ; otherwise, it first decreases and then increases and finally decreases in $v$ .

Corollary 2 demonstrates that in strict contrast to the scenario of pre-release verification, the firm may benefit from a higher review variance under the scenario of responsive verification, see Figure 4 for illustration. Recall Corollary 1, where the firm has to adopt quality verification before observing the consumer reviews, the main functionality of quality verification is to enhance the transparency of quality information by providing an additional information source (i.e., quality standard) for consumers to better inform the product quality. Therefore, a higher review variance obscures the quality information and thus undermines the firm’s expected profit. Differently, when the firm makes the verification decision after the reviews are realized, it would take it to correct the consumers’ negative inference when they encounter negative reviews, but omit it when they encounter positive reviews. In other words, by designing a responsive verification strategy, the downside of negative reviews can be mitigated by quality verification while the bright side of positive reviews is still retained. When both the quality standard and the verification cost are high, it makes it harder for the firm to use verification to mitigate the negative effect of review bias; the firm’s profit still decreases in the review variance due to the reduced level of information transparency. Otherwise, the firm’s profit may increase in the review variance once its level is sufficiently high, as the firm can actively use verification to mitigate the negative bias but enjoy the benefit from the positive bias.

Figure 4.

The impact of review variance on profitability under the scenario of responsive verification. (a) $b=2/3$ and $c=0.02$ and (b)  $b=0.85$ and $c=0.014$ . Note. In the shaded region, the firm would never obtain quality verification.

4.3. Comparison Between Pre-Release and Responsive Verifications

A close look at the equilibrium outcomes in Propositions 1 and 2 reveals that the review bias exerts a marked influence on the firm’s equilibrium quality verification strategy under two scenarios with different verification formats. In this subsection, we further compare the firm’s expected profit as well as consumer surplus based on the assumption that the firm is restricted to undertake only one verification format and incurs the same verification cost.

Proposition 3

Compared to the scenario of pre-release quality verification, the firm’s expected profit is lower under the scenario of responsive quality verification when the verification cost $c\le c_3$ , is higher when $c_3<c\le max\{c^{S,R},c^{S,P}\}$ , and is the same otherwise, where $c_3$ first increases and then decreases in review bias $v$ . ¹⁰

Proposition 3 shows that when reviews are biased, the firm’s expected profit can be higher under either verification scenario, which highlights the adverse effects of review bias on profitability under each verification format. See Figure 5(a) for illustration. The main functionality of pre-release verification is to enhance the early arriving consumers’ quality expectation and to improve the late-arriving consumers’ quality inference from biased consumer reviews. However, its drawback is the potential for unnecessary verification costs if initial consumer reviews turn out to be positive. Conversely, the responsive verification strategy proves more cost-effective, as it enables the firm to initiate quality verification solely in response to negatively biased consumer reviews, though this comes at the expense of potentially higher first-period profits.

Figure 5.

(a) Firm’s expected profit and (b) consumer surplus comparison under two quality verification format ( $b=0.6$ , $m=1.5$ ).

Specifically, when the verification cost is lower than a threshold, that is, $c\le max\{c^{S,R},c^{S,P}\}$ , the firm pursues verification under the scenario of pre-release verification when $c<c^{S,P}$ , while doing so under the scenario of responsive format only if the consumer reviews are negatively biased (see conditions in Proposition 2). In this case, if the verification cost is sufficiently low ( $c\le c_3$ ), the impact of verification cost saving is negligible, and the firm earns more under the pre-release verification format due to the increased expected profit in period 1. However, within a moderate cost range ( $c_3<c\le max\{c^{S,R},c^{S,P}\}$ ), if the firm still commits to the pre-release format, the increased expenditure of obtaining verification outweighs the benefit of a higher expected profit in period 1. If the firm chooses a responsive verification format, it can enjoy the flexibility of adopting a verification strategy in response to the realized consumer reviews. When the cost is sufficiently high ( $c>max\{c^{S,R},c^{S,P}\}$ ), the firm forgoes quality verification under both formats, the early adopters retain their priors, and the followers rely on consumer reviews to infer product quality under both formats. As a result, the firm sees no difference in its expected profits between the two quality verification formats.

Figure 5(a) illustrates that when the review variance increases, the firm’s preference over pre-release verification format first increases and then decreases, that is, the threshold cost $c_3$ exhibits an inverted U-shape. This result is consistent with those observations in Corollaries 1 and 2, where we have shown that the firm’s expected profit monotonically decreases in the review variance under the pre-release scenario but might increase in the variance once it is sufficiently high under the responsive scenario. The rationale is that the benefit of a responsive verification format is more pronounced when the review variance is higher, in which case the firm can use verification to mitigate the downside of negative reviews more effectively. This flexibility is absent under the pre-release verification format, which is designed to enhance the followers’ overall quality inference accuracy from both negatively and positively biased reviews.

Proposition 4

Compared to the scenario of pre-release quality verification, consumer surplus under the scenario of responsive quality verification is higher when $b<2/3$ , $v>v_1$ and $c<max\{c_4,0\}$ , is the same when $c>max\{c^{S,R},c^{S,P}\}$ , and is lower otherwise. ¹¹

Proposition 4 demonstrates that review bias also exerts a non-trivial effect on the consumers’ preference toward two verification formats. Pre-release verification format can help both early adopters and followers to better infer the product quality over two periods, while the responsive verification format can only rectify the followers’ negative quality inference in the second period. Intuitively, one might expect that consumer surplus under the scenario of pre-release verification should not be any lower than under the scenario of responsive verification. While this is true in most cases, as illustrated in Figure 5(b), consumer surplus can be higher under the responsive verification when the level of review bias is high and the verification cost is low. This happens because when consumer reviews are highly biased toward the positive, followers generate a higher quality expectation and thus derive greater surplus. Conversely, if the reviews are negatively biased, the firm can easily obtain quality verification to bolster consumer quality expectations when the verification cost is low. Here, the benefits of improving the followers’ quality expectations via responsive verification prevail and lead to higher consumer surplus under this scenario.

5. Extensions

In this section, we will investigate five additional extensions to demonstrate the boundary conditions for our main results, as well as their robustness. First, we investigate a scenario wherein the firm endogenously determines quality verification timing. Second, we assume that consumer reviews exhibit positive (respectively, negative) bias, where consumers may be more likely to share their positive (respectively, negative) consumption experiences than negative (respectively, positive) ones. Third, we assume that the followers are naive, meaning they simply take the aggregated consumer review as the true product quality. Fourth, we examine a scenario wherein the followers are able to intuit the true product quality level given a sufficient number of consumer reviews. Fifth, we assume that the consumers may exhibit reference-dependent preferences in their reviews. Last, we discuss managerial implications.

5.1. When the Firm Endogenously Determines Quality Verification Timing

In the baseline model, we consider two independent scenarios wherein the firm either makes a quality verification decision at the beginning of period 1 (pre-release quality verification format) or after the consumer reviews are released (responsive quality verification format). In this subsection, we consider an alternative case wherein the firm can endogenously determine the timing of quality verification. That is, the firm first decides whether to obtain quality verification (at a cost) at the beginning of period 1; if it does not, after the consumer reviews are released, the firm can still decide whether to obtain quality verification to influence the followers’ quality expectation in the second period. Such a sequential signaling game requires a more dedicated belief updating process for the consumers. For example, although both early adopters and followers would update their quality belief in the first period after observing the firm’s verification action, they would take the firm’s possible responsive verification in the second period into account. This differs from the scenario of pre-release verification, where consumers update their quality belief in the first-period solely based on the firm’s pre-release verification decision. Also, unlike the scenario of respective verification where all consumers hold a prior quality belief in the first period, in this scenario, if the firm obtains quality verification in period 1, both early adopters and followers update their belief that product quality is uniformly distributed on $[b,1]$ . Then, after observing consumer reviews, the followers further infer that the true quality follows $q\sim U[max\{\mathbb{R}-v,b\},min\{\mathbb{R}+v,1\}]$ . If the firm does not obtain quality verification in period 1, the early adopters and followers still update their quality belief as $q\sim [0,\hat{q}]$ , where $\hat{q}=b$ or $1$ . Then, in period 2, the followers further update their quality beliefs according to the realized consumer reviews and the firm’s quality verification strategy: if the firm obtains quality verification, they infer $q\sim [max\{\mathbb{R}-v,b\},min\{\mathbb{R}+v,1\}]$ , otherwise they update their quality belief as $q\sim [max\{\mathbb{R}-v,0\},min\{\mathbb{R}+v,\hat{q}\}]$ . Based on the consumers’ updated quality beliefs in each period, we can derive the firm’s equilibrium verification strategy in the following proposition.

Proposition 1 Propositin 5

When the firm endogenously determines whether and when to obtain quality verification, in equilibrium,

The firm obtains quality verification at the beginning of period 1 when $c<c_5$ ;

The firm obtains quality verification after consumer reviews are released when

$$c_5\le c<max\{c_5,c^{S,R}\},$$

and

$$max\{b-v,\frac{16c-m{b}^2-2bmv}{2mb}\}<\mathbb{R}<{\mathbb{R}}^{S,R};$$

Otherwise, the firm never obtains quality verification. ¹²

Proposition 5 derives the firm’s optimal quality verification strategy over two periods, when it is able to adopt quality verification in either period. Recall Propositions 1 and 2, it shows that the main functionalities of pre-release and responsive verification formats still hold in this scenario with endogenous verification timing. Adopting quality verification in the first period (pre-release verification) can enhance the early adopters’ quality expectation and help followers to better infer the product quality; therefore, the firm always has the incentive to undertake it as long as the verification cost is limited, that is, $c<c_5$ . If the firm does not adopt verification upfront, it would undertake the verification in the second period after encountering negatively biased reviews, as long as the verification cost is not too high, that is, $c_5\le c<max\{c_5,c^{S,R}\}$ . Here, the main impact of quality verification is the same as the response verification format to correct the followers’ negative quality inference.

Compared to the scenario of pre-release verification, the firm becomes less active to obtain quality verification in the first period when it can endogenously determine the timing for quality verification, that is, $c_5\le c^{S,P}$ . The intuition is that, anticipating there exists another chance to rectify the followers’ quality expectation after the reviews are released the firm’s incentive for taking quality verification upfront diminishes. This also alters the consumers’ quality updating process. For example, when the verification cost is between $c_5$ and $c^{S,P}$ , under the scenario of pre-release verification, if the firm does not obtain quality verification in the first period, the consumers would form a low-quality belief that $q\sim [0,b]$ . While under the scenario of endogenous verification timing, the consumers would attribute such a non-verification behavior to the high verification cost and keep their prior quality belief that $q\sim [0,1]$ . On the other hand, compared to the scenario of responsive verification, the firm’s incentive to adopt quality verification in the second period also decreases. This is not only attributed to the fact that the firm would now undertake verification in the first period once the verification cost is sufficiently low, but also to the fact that the consumers can now further update their quality belief in the absence of first-period quality verification. In either case, the followers can better infer the product quality, and the impact of taking verification in the second period diminishes.

5.2. When Reviews Exhibit Negative/Positive Bias

In the baseline model, we assume that the review bias caused by the consumers’ idiosyncratic preferences is symmetric, such that the consumer reviews tend to be positive or negative with equal degree. While in practice, consumers sometimes systematically encounter negative idiosyncratic preferences, such as the product may not align well with their preferences (e.g., a hotel during renovations where every guest encounters some disruption, but the severity varies) or systematically experiences positive shocks, as shown in our motivating example of Airbnb. In either case, the consumer review is no longer symmetric but exhibits either negative or positive bias. In this subsection, we will examine whether the results still hold in the presence of negative and positive bias, respectively.

For the negative bias, let us assume that given the product quality $q$ , the aggregated consumer review ${\mathbb{R}}_{neg}=q+\epsilon$ , in which the review bias $\epsilon$ now follows a uniform distribution between $-v$ and $0$ , or $\epsilon \sim U[-v,0]$ . By a similar token, we assume that for the positive bias, the review bias $\epsilon$ now follows a uniform distribution between $0$ and $v$ , that is, $\epsilon \sim U[0,v]$ . It is worth mentioning that we only consider two extreme cases to illustrate the impacts of negative and positive bias, the impacts of which can be applied to more general cases where consumers generate both positive and negative reviews while the degrees are different.

We can show that whenever reviews exhibit negative or positive bias, one can see that the firm’s verification strategy remains qualitatively similar to that in the baseline model. Moreover, compared to the baseline model, the firm’s incentive for adopting quality verification under both scenarios of pre-release and responsive verification decreases, no matter whether the reviews exhibit positive or negative bias. The intuition is that when the reviews exhibit negative or positive bias, the variance of review bias is actually reduced, so that rational consumers can better infer the product quality from these reviews, and the reviews become a more reliable information source. Therefore, the firm becomes less active in using verification to mitigate the impact of review bias while relying more on consumer reviews to disseminate the quality information.

Another interesting observation is that under the pre-release scenario, the firm’s quality verification incentive remains the same with negative and positive bias, while under the responsive scenario, the firm is more active at obtaining quality verification when the reviews exhibit positive bias. This result further highlights the different functionality of two verification formats, wherein the pre-release format is designed to enhance the consumers’ overall quality inference accuracy, so that it is independent of the bias type but only relates to the magnitude of variance. While the responsive format is designed to correct the followers’ negative quality inference. Specifically, when the reviews exhibit positive bias, the followers know that the reviews are higher than the true quality level, so as to negatively rectify their quality belief based on the reviews, leading to a lower updated quality expectation. While the opposite is true for the negative bias, where consumers generate a higher updated quality expectation from the reviews. Therefore, the reviews are more likely to be negatively biased with positive bias, and the firm has to take verification more frequently.

5.3. When Consumers are Naive at Review Bias

In the base model, we assume consumers are sophisticated enough to discern the bias present in reviews. In this subsection, we consider an alternative scenario wherein the consumers are naive toward review bias, meaning they consider the aggregated consumer reviews as a true reflection of product quality. Despite this misapprehension among consumers, the firm’s quality verification strategy still works, and when there is a contradiction between consumer reviews and verification, the consumers trust the verification (Guan et al., 2020). If the firm does not obtain verification, the followers can ensure the true product quality is lower than the quality standard $b$ , then, after observing the aggregated consumer reviews $R$ , the followers believe the product quality level is $\mathbb{R}$ if $\mathbb{R}<b$ while otherwise updating perceived quality as $b/2$ . Should the firm obtain verification, the followers believe product quality level is $\mathbb{R}$ if $\mathbb{R}>b$ while otherwise updating perceived quality as $(1+b)/2$ . All other settings remain unchanged.

When the followers are naive, in comparison to the baseline model, we show that the firm has less incentive to obtain quality verification under the pre-release scenario but a stronger incentive to obtain quality under the responsive scenario. Such a contrast can be explained as follows. Under the pre-release verification scenario, in the baseline model, the main functionality of verification is to help rationalize followers better infer the product quality, as they would infer the quality to fall into a certain quality range after observing the reviews. However, when consumers are naive, they simply accept the reviews as accurate reflections of quality and have no uncertainty about the product quality. This mitigates the impact of quality verification, and the firm’s incentive for pre-release verification diminishes. In contrast, under the scenario of responsive verification, the firm mainly uses verification to correct the consumers’ negative inference of product quality. When consumers are naive, and there is a contradiction between consumer reviews and verification, the consumers trust more in verification, which helps the firm to significantly enhance the consumers’ quality expectation via verification (Guan et al., 2020). For example, if a high-quality product encounters consumer reviews below the threshold $b$ (i.e., $\mathbb{R}<b$ ), the firm can use verification to induce the followers’ quality belief to the range of $q\sim U[b,1]$ . This magnifies the benefit of quality verification and also the firm’s incentive for verification under the scenario of a responsive format.

5.4. When Sufficient Reviews Mitigate Review Variance

In the base model, we assume the consumer reviews inevitably contain some variance, and the followers can only infer the true product quality range. Arguably, this assumption is restrictive because, given a sufficient number of consumer reviews, the followers are able to accurately infer the true product quality level through advanced data analysis technology. In this section, we extend our base model by accounting for the reduction of review variance caused by a sufficient number of consumer reviews.

The setup of the model is the same as the base model, with the only exception that the specific functional form of review bias $\epsilon$ is as follows:

$$\{\begin{array}{ll}{\displaystyle \epsilon \sim U[-v,v],}& {\displaystyle \text{if }{D}_1<k.}\\ {\displaystyle \epsilon =0,}& {\displaystyle \text{if}{D}_1\ge k.}\end{array}$$

That is, if the demand in period 1 $D_1$ exceeds the exogenous threshold $k$ , the followers are able to accurately infer the true product quality level. Otherwise, they can only infer based on a range of true product quality. To ensure the discussion is non-trivial, we assume $k<K$ and $m=1$ , or else the analysis would degenerate to the base model.

The use of such a step function has been adopted by related literature (Valletti and Wu, 2020) and is common in statistical analysis. One example is the classical hypothesis testing. In our results, the followers’ product quality inferences can be considered a variation of hypothesis testing problems, where the null hypothesis is that a consumer’s valuation is equal to a pre-specified value. A more accurate estimate is obtained by having a larger sample.

Proposition 5proposition 6

When sufficient reviews mitigate review variance, in equilibrium,

Under the pre-release scenario, the firm obtains quality verification when $c<c_k^P$ , in which the firm’s incentive for verification first decreases and then increases with $k$ . ¹³

Under the responsive scenario, the firm never obtains the quality verification when $k\le 1/4$ and degenerates to the base model when $k>1/4$ .

Proposition 6 illustrates how the firm’s verification strategy changes when the review bias can be mitigated by the number of reviews. Under the pre-release scenario, conditional on the magnitude of $k$ , the firm becomes less active in obtaining quality verification when $k$ is small. Because now the first-period demand can easily exceed $k$ , the followers can infer the true quality from the unbiased reviews, so that the impact of verification on the second-period profit vanishes. Since now the firm can use verification to only influence the early adopters’ quality inference, its incentive diminishes. When $k$ is large, the firm reversely becomes more active in adopting quality verification. By doing so, the firm can use pre-release verification to induce a higher demand in the first period that exceeds $k$ , so as to better explore the benefit of consumer reviews in the second period. Under the responsive scenario, the demand in the first period is independent of quality verification and remains constant at $D_1=1/4$ . Thus, when $k\le 1/4$ , the followers can observe the true quality via consumer reviews, and the firm will never adopt quality verification. If $k>1/4$ , the model degenerates to the base model.

5.5. When Consumers Exhibit Reference-Dependent Preference

In the baseline model, the review bias is driven by the gap between consumers’ perceived quality and the product’s true quality. In this subsection, we assume that the review bias is further influenced by the consumers’ reference-dependent preference. That is, if the perceived quality is higher (lower) than the quality reference point, the consumers would experience a sense of enjoyment (disappointment) and will comment more positively (negatively) on the product. Incorporating the reference effect, we assume that the aggregated consumer reviews are given by $\mathbb{R}=q+\epsilon +\delta (q-{\overline{q}}_1)$ , where $\epsilon$ is still the variance term, $\delta$ is the magnitude of consumer reference effect, and ${\overline{q}}_1$ is the reference point that equals the early adopters’ quality expectation in period 1. When $\delta =0$ , the game degenerates to our baseline model. When $\delta$ becomes higher, the reviews are more influenced by the consumers’ reference effect. All other settings remain unchanged.

It can be verified that when the consumers exhibit reference-dependent preference, the firm’s equilibrium verification strategy remains qualitatively similar as a baseline model, while the reference effect influences the firm’s incentive for taking quality verification in opposite ways under two scenarios. Specifically, when the magnitude of consumer reference effect $\delta$ increases, the firm’s incentive for verification increases under the pre-release verification scenario, while decreases under the responsive verification scenario. The intuitions are as follows. Under a pre-release scenario, for the high-quality product, if the firm does not adopt quality verification upfront, the early adopters hold an expected quality ${\overline{q}}_1=1/2$ and would be more likely to generate positive reviews given that now the reference effect is positive, that is, $b>1/2$ and $\delta (q-(1/2))>0$ . However, in anticipating this, the followers would negatively rectify the positive reviews and generate lower quality expectations for the high-quality product. Differently, if the firm adopts quality verification upfront, it raises the consumers’ reference point from $1/2$ to $(1+b)/2$ and thus mitigates the consumers’ negative reference to quality caused by reference preference. When $\delta$ increases, the negative impact of the reference effect strengthens, which reversely incentivizes the firm to adopt pre-release verification more frequently. In contrast, under the scenario of responsive verification, the firm uses verification to enhance the consumers’ quality expectation when they encounter negative reviews that are below the true product quality. However, since now the consumers are more likely to generate positive reviews for the high-quality product due to the existence of positive reference effect, that is, $\delta (q-(1/2))>0$ , the firm accordingly becomes less active at adopting quality verification, whose incentive further decreases in the magnitude of reference effect.

5.6. Managerial Implications

It is increasingly common for companies to combine quality verification with consumer reviews to influence the consumers’ perception of product quality. Building a game-theoretical model with private quality information, our analysis provides a number of managerial insights regarding how firms can strategically determine their optimal quality verification approach to mitigate the negative effects of review bias. Below, we discuss in detail these managerial insights derived directly from the propositions.

Proposition 1Insight 1

Firms should proactively seek quality verification when the review variance increases.

In practice, firms gauge review variance by monitoring the polarization of consumer ratings (Sun, 2012). Specifically, high review variance typically manifests as a bimodal distribution where consumer opinions show a sharp divergence between extreme positive and negative feedback rather than forming a consensus. Furthermore, sentiment analysis measures the intensity of opinions by analyzing emotional language. Review variance is considered high when opinions are polarized into strong positives or negatives with few neutral comments. Our model demonstrates that when firms identify the high review variance, they should rely more heavily on quality verification to mitigate the negative influence of dispersed reviews on consumer quality inferences.

Proposition 2Insight 2

The firm should choose a pre-release format when the verification cost is low and a responsive format when the verification cost is intermediate. Otherwise, it is better for the firm to forgo verification.

Our analysis also provides useful guidance on how to choose the right verification format. If the verification cost is low, the impact of verification cost saving is negligible, and the firm should use a pre-release verification format to enhance the consumers’ quality expectation in the first period. This is typical for standardized safety tests where technologies are mature. For example, manufacturers in the electronics industry often obtain UL or CE certifications before product launch to signal safety and reliability. However, if the verification cost falls within a moderate range, the firm should choose the responsive verification format. This scenario is common for products requiring customized expert audits where the process is more resource-intensive. Finally, when the verification cost increases further than the benefit of verification cannot cover its cost, the firm should forgo verification.

Proposition 3Insight 3

When the followers are easier to accurately infer the product quality based on a sufficient number of consumer reviews, the firm can rely more on consumer reviews to reveal quality information.

As data analysis technology matures and professional product reviews become more prevalent, consumers become easier to obtain true product quality information based on reviews. Our model suggests that, amidst this trend, the firms should rely more on reviews to reveal quality information as the impact of verification on the second-period profit vanishes. This provides a theoretical explanation for why Lululemon leverages community-driven reviews to strengthen quality perception, for example, Lululemon prominently features user-uploaded photos of products in fitness scenarios alongside detailed reviews on its official website and e-commerce platforms. ¹⁴

6. Conclusion

Quality verification is a longstanding, credible yet costly method of disseminating reliable product information to consumers. Nowadays, consumers also rely on accessible online reviews for product quality information to guide purchase decisions. However, reviews inevitably contain the consumers’ idiosyncratic preference toward the product and are thus never an objective source of product quality. Therefore, in practice, firms frequently combine quality verification with consumer reviews to reveal quality information. This paper aims to investigate how the firm could determine its optimal quality verification strategy to mitigate the negative effect of review bias, and what are the implications of verification timing. In this paper, we define two quality verification formats according to the timing of quality verification. With pre-release quality verification, the firm obtains quality verification at the beginning of the first stage, while with responsive quality verification, the firm adopts quality verification to correct negatively biased consumer reviews after they are released. We build a two-period dynamic model, wherein the firm privately observes its product quality, and two groups of consumers arrive sequentially and make their rational quality inference according to the firm’s verification strategy and reviews.

We show that the main functionality of pre-release verification is to enhance the early arriving consumers’ quality expectation and to improve the late-arriving consumers’ quality inference from biased reviews. However, its drawback is the potential for unnecessary verification costs if initial consumer reviews turn out to be positive. When the review variance is low, consumer reviews serve as a reliable, cost-free information source, such that the firm’s verification strategy remains the same as objective consumer reviews. Nonetheless, when the magnitude of bias is high, consumer reviews may diverge considerably from the true quality, prompting the firm to take a more proactive approach to quality verification. In equilibrium, the firm’s expected profit monotonically decreases in the review variance. The responsive verification strategy proves more cost-effective, as it enables the firm to initiate quality verification solely in response to negatively biased consumer reviews, though this comes at the expense of potentially higher first-period profits. In equilibrium, the firm obtains responsive verification only when the verification cost is low, the aggregated consumer reviews are negatively biased, and the product exceeds the quality standard. Different from the scenario of pre-release verification, when the volatility of consumer reviews increases, the firm can use responsive verification to mitigate the downside of negative reviews but retain the benefit of positive reviews, so that its profit may become higher. Due to the conflicting functionality of two verification formats, we show that either verification format can result in a higher profitability for the firm, conditional on the verification cost and the magnitude of review bias.

We conclude the paper by suggesting possible future research directions. First, in this study, we consider a monopoly firm’s quality verification strategy in the presence of review bias. It would be worthwhile to examine the impacts of consumer reviews and the firm’s preference toward two verification formats in a competitive environment. This, however, requires a very different model setup, which is beyond the scope of this study. Second, this study considers consumer reviews as a summary statistic for the average rating. It would be interesting to see what happens when the followers infer quality information according to the actual text of reviews and how such a form of reviews influences the firm’s profit and quality verification timing. Third, while we assume quality verification mainly focuses on the vertical attribute-product quality; consumers may have different preferences toward these vertical attributes (i.e., health and taste). Another potential direction could thus be to explore the impacts of review bias and verification formats by taking horizontal attributes into consideration.

Supplemental Material

sj-pdf-1-pao-10.1177_10591478261433279 - Supplemental material for Quality Verification in the Presence of Review Bias

Supplemental material, sj-pdf-1-pao-10.1177_10591478261433279 for Quality Verification in the Presence of Review Bias by Xu Guan, Yaxu Guo, Yuan Jiang, Guangrui Ma and Yinliang(Ricky) Tan in Production and Operations Management