Introduction
Under classical economic theory, consumers play an important role in maintaining market discipline because they can use their transactional decisions to ensure that suppliers act in their best interest.Footnote 1 Optimal consumer decision-making is, however, dependent on access to complete and accurate information on the price, characteristics and quality of a commodity (and its substitutes) as well as the terms on which it is sold.Footnote 2 In reality, obtaining information is costly – and in some cases impossible – for consumers.Footnote 3 The consequence is that consumers and suppliers often have an uneven amount of information, which places them in unequal bargaining positions.Footnote 4
Accordingly, regulatory intervention in consumer markets is frequently justified on the need to reduce information gaps to allow consumers make well-informed decisions reflecting their preferences.Footnote 5 To this end, legislators and regulators deploy techniques aimed at improving consumer access to information. A commonly-used technique is mandatory disclosure, which requires that businesses supply specific information to consumers.Footnote 6 Mandatory disclosure aims to give disclosees the information required to make better decisions and to keep disclosers from abusing their superior position.Footnote 7
The use of mandatory disclosure has been criticised because it is thought to rest on ‘false assumptions about how people live, think, and make decisions’.Footnote 8 Mandatory disclosure relies on the neo-classical assumption that consumers are rational and are the best judges of their interests, who maximise their utility with the limited resources available.Footnote 9 This assumption has been questioned by behavioural research which suggests that even when consumers have access to information, there is a tendency not to process such information correctly for several reasons, including the susceptibility to behavioural biases.Footnote 10
Ben-Shahar and Schneider provide a well-articulated criticism of mandatory disclosure in their book More Than You Wanted To Know.Footnote 11 Amongst other things, they assert that
Faced with unfamiliar and complex decisions that disclosures are intended to inform, people don't want to be educated, don't want spreadsheets and don't want scrolls. They want advice. (emphasis added)Footnote 12
They highlight that many markets provide advice in the form of ‘ratings, rankings, scores, grades, labels, warnings and reviews’ and that such advice often does better than disclosures in helping consumers make unfamiliar and complex decisions.Footnote 13 Supporting their assertion, growing evidence suggests that consumers increasingly rely on one type of advice – consumer-generated reviews (CGRs) – when making transactional decisions.Footnote 14 This trend is fuelled by the growth of electronic commerce and online platforms which allow consumers to share their evaluation of products.
Since consumers rely on CGRs in decision-making, this alternative information stream invites closer scrutiny. This paper has two goals. First, it will highlight specific problems which question the reliability of CGRs as a source of consumer information.Footnote 15 Secondly, it considers regulatory responses which may address some of the problems highlighted. To this end, this paper is structured as follows: following this introduction, section 1 considers the growing emergence and merits of CGRs. Section 2 discusses certain problems associated with CGRs, while section 3 considers the appropriate regulatory responses to these problems. Conclusions are then drawn in the final section.
1. Consumer-generated reviews: an alternative?
Consumer-to-consumer information sharing is not novel: word-of-mouth information exchange amongst peers is an ancient communication mechanism.Footnote 16 While word-of-mouth exchanges had a limited reach, the Internet provides a low-cost channel for disseminating information on an unprecedented scale.Footnote 17 With the rise of online transactions, it has become the norm for retail websites to incorporate forums enabling consumers to discuss their experiences of using products and dealing with suppliers.Footnote 18 There are also independent websites solely dedicated to aggregating consumer reviews.Footnote 19 These forums are usually free to access and primarily serve to facilitate information exchange amongst consumers.Footnote 20 CGRs posted on these forums may come in the form of open-ended textual information on a product's perceived quality.Footnote 21 They may also come in the form of ratings where a consumer's evaluation of a product is represented by a specified number on a scale or by the use of stars.Footnote 22 Many online platforms combine both. While ratings give consumers a quick indication of a product's popularity, open-ended textual information provides specific details reflecting each reviewer's experience.
CGRs serve two functions. First, they act as informants because they provide consumers with additional information derived from other consumers’ experiences.Footnote 23 Secondly, they act as recommenders since they provide positive or negative signals about a product's popularity, which reflects the overall positive reviews that the product generates.Footnote 24 These functions place CGRs as one of the most influential non-marketer sources of product information.Footnote 25 Evidence suggests that CGRs affect the decision-making and purchasing behaviour of consumers.Footnote 26 For instance, research carried out by Review Trackers reveals that 63.6% of consumers are likely to check online reviews on Google before transacting with a business,Footnote 27 and 94% admit that a negative online review has convinced them to avoid a business.Footnote 28
The influence of CGRs on consumer decision-making can be explained by certain benefits that they provide, which are discussed below.
(a) Reduces search costs
To make optimal decisions, consumers must gather and process sufficient information on products as well as their substitutes. This process is costly in terms of money, time and effort.Footnote 29 If consumers believe that the cost of searching for information outweighs the benefit, they may settle for products that do not represent their preference or the best bargain that they could have obtained.Footnote 30
The Internet contributes to reducing search costs by facilitating easy and timely access to CGRs at little to no cost.Footnote 31 Consumers can access thousands of reviews on their personal computers or mobile devices, which enables them to compare products and their prices.Footnote 32 One empirical study suggests that CGRs can reduce search time by about 25%Footnote 33 which is valuable to consumers seeking relevant and timely information.Footnote 34
(b) Reduces uncertainty
Consumers face uncertainties online because parties are not physically transacting with each other and because many transactions are one-off events. Consumers’ uncertainty is also fuelled by the inability to verify the identity of suppliers and the attributes/quality of products before purchase.Footnote 35 These uncertainties result from the lack of perfect information.Footnote 36
To overcome uncertainties, consumers read reviews to obtain information.Footnote 37 Information acquired gives consumers some confidence, since CGRs set clearer expectations of a product's performance. In comparison to information provided by businesses, consumers also trust and place a higher value in the opinion of their peers.Footnote 38 This may be because of the perceived non-commercial motives of their peers.Footnote 39 CGRs also emphasise information that consumers may have ignored or thought less-significant.Footnote 40 For instance, reviews may highlight the difficulty in contacting a supplier to secure refunds for defective products – a less salient issue that consumers may not have considered initially.
(c) User-oriented
Business-to-consumer information is often criticised as complex and product-oriented because it focuses on describing product attributes and performance in terms of technical specifications.Footnote 41 It is also usually voluminous, making it difficult for consumers to understand. Consequently, regulators often require that business-to-consumer disclosures are simplified. Simplification may involve drafting disclosures in simple language or streamlining the presentation of disclosed content.Footnote 42 However, Ben-Shahar and Schneider explain that ‘simplifying [disclosures] fail because the complex isn't simple and can't easily be made so’.Footnote 43 Simply put, it is challenging to present complicated information in written language that most people understand.Footnote 44
Because CGRs are written from a consumer perspective, in comparison to disclosures, they may convey information in a less complicated manner.Footnote 45 Being expressed in more relatable language, consumers may find them easier to understand. They will not be as lengthy as disclosures and will provide information that is based on user experience.
(d) Good mechanism for disciplining firms
The Internet provides a relatively cheap medium which allows consumers to reveal the actions of suppliers and to attract other consumers’ attention in the process.Footnote 46 Suppliers care about CGRs because they can affect future profits.Footnote 47 If many CGRs express dissatisfaction with a firm's products, it creates a negative perception of the product and the firm's reputation. Since perception can impact sales, suppliers have an incentive to behave appropriately.Footnote 48
Some commentators believe that the potential to discipline firms can impact the regulation of contracts terms. Taylor argues that if a firm's reputation is attacked because consumers criticise their contract terms, the firm may be motivated to change unfavourable terms.Footnote 49 Echoing this, Becher and Zarsky contend that the information flow between consumers will force suppliers to remove unfair terms, especially where a possibility exists that consumers will refrain from contracting with them.Footnote 50 While this may be true in some cases,Footnote 51 discussions in section 2 will show that this is an optimistic view because it rests on the assumption that many consumers will be motivated enough to read contract terms and to write reviews assessing them.
2. Problems with consumer-generated reviews
The integrity of available information affects the quality of consumer decision-making. While CGRs contribute to reducing information gaps, certain concerns question their role as a reliable source of consumer information. These concerns are considered below.
(a) Bias
CGRs can leave consumers with misleading, inaccurate or false information because a product's aggregated reviews may not reflect the unbiased opinion of other consumers.Footnote 52 Information contained in CGRs may be biased for several reasons, discussed below.
(i) Fake reviews
Since CGRs can impact sales, businesses are incentivised to ensure that reviews remain positive.Footnote 53 Consequently, they may commission fake reviews by paying for them or by planting them on platforms using automated software programs (bots).Footnote 54 Enforcement cases highlight this problem. In 2019, the US Federal Trade Commission (FTC) announced its first case challenging a marketer's use of fake paid reviews.Footnote 55 The defendants allegedly paid a third-party website to create and post Amazon reviews of its weight-loss pill. These sham reviews helped the product attain a 5-star rating on Amazon. The court order settling the FTC's complaint, amongst other things, required that notices detailing the FTC's complaints be emailed to consumers who had purchased the product.
Similarly, in 2014, Bell Canada's employees were encouraged to post positive reviews of the company's apps on the iTunes and Google Play Store.Footnote 56 These reviews were posted without the reviewers disclosing that they were Bell Canada employees. Canada's Competition Bureau determined that the reviews ‘created the general impression that they were made by independent and impartial consumers and temporarily affected the overall star rating for the apps’.Footnote 57 The Bureau required that Bell Canada enhance its compliance program to ensure that employees would not review products.Footnote 58
The practices highlighted in these cases can undermine the reliability of the information contained in CGRs.Footnote 59 Even where consumers are cautious, they are cognitively ill-equipped to identify fake reviews because humans have a truth bias. This means that they are likely to believe information reported as true rather than false.Footnote 60 This raises problems because decisions influenced by sham reviews may not be optimal.
Many CGRs are written anonymously, and the inability to verify a reviewer's identity is associated with a higher possibility of fake reviews. Platforms like Amazon adopt mechanisms which aim to partly resolve this. For instance, verified purchasers have an indicative badge on their reviews. However, Amazon also permits reviews written by unverified purchasers. This means that reviews written by unverified purchasers, who may or may not be fake reviewers, can influence a product's overall rating.
(ii) Contributor integrity and competence
Genuine reviews may be biased due to the reviewer's experience or motivation. Contributor integrity problems exist in online forums because consumers cannot verify a reviewer's knowledge or reliability.Footnote 61 This is significant because such interactions are usually non-recurring.Footnote 62
A reviewer's knowledge and experience affects the quality of information that they provide. For instance, an experienced long-term user of a product may provide better information than a first-time user. Thus, if the reviewer is incompetent or inexperienced, then the advice they provide may be inaccurate and misleading.Footnote 63 It is usually difficult or near impossible for consumers to identify which reviews have been written by experienced or knowledgeable peers. Although some reviewers leave cues indicating their experience, these are not easily verifiable.Footnote 64
Moreover, reviewers do not have strong incentives to post truthful information. This directly contrasts with firms required to provide mandatory disclosures. Such firms will face regulatory sanctions and lawsuits if they fail to provide truthful information.Footnote 65 Should a consumer's transactional decision be induced by false or misleading disclosures, laws like the UK's Consumer Protection from Unfair Trading Regulations 2008Footnote 66 will protect them. However, if an anonymous review induces a transaction, in the absence of a direct contractual link with the reviewer, the consumer may have no remedy.Footnote 67 Neither do anonymous reviewers qualify as traders – if they did, the 2008 Regulations would apply.Footnote 68
Contributor integrity problems can also be linked to the fact that when writing reviews, consumers may be motivated by factors other than the objective evaluation of a product. For instance, evidence suggests that social influence bias affects how consumers write reviews.Footnote 69 Consumers do not write reviews in isolation. They are often exposed to numerous information signals within the review platforms which affect the objective assessment of a product.Footnote 70 Schlosser, for instance, finds that consumers may negatively adjust their review after reading a negative review.Footnote 71 She theorises that this may be because consumers seek to differentiate their reviews and negative reviews stand out because they are perceived as more intelligent.Footnote 72 Confirming the existence of social influence bias, another study shows that consumers who received email invitations and who could not see reviews written by their peers consistently provided higher ratings when compared to self-motivated reviews provided by those who had access to other reviews.Footnote 73
The signals as to product quality provided by aggregated CGRs may also be skewed because many consumers self-select to provide reviews. Hu et al point out that reviews are often written by consumers who are either extremely satisfied or dissatisfied.Footnote 74 The averagely satisfied consumers, who may represent the majority, are less likely to write reviews.Footnote 75 Hence, the information provided in CGRs may reflect only the views of a vocal minority. Such information will be biased since it is written by a non-representative sample of consumers.Footnote 76
(iii) Platform manipulation
Sometimes, review platforms may not be neutral, and this can affect the quality of the information exchanged. Bias on such platforms may be traced to the review platforms’ business structure. Review platforms may be grouped as ‘stand-alone’ platforms or ‘supportive’ platforms.Footnote 77 Stand-alone platforms are solely designed to allow consumers to share reviews. Such platforms often rely on advertising and subscription revenue:Footnote 78 examples include Trustpilot and Yelp. Supportive platforms also allow the sharing of reviews, but they are supplementary to a larger forum for facilitating transactions:Footnote 79 an example would be Amazon.
Friedman explains that because stand-alone platforms need to generate revenue through advertising, promotions, and sales-commission, they are subject to tensions arising from potential conflicts of interest.Footnote 80 This is because they seek to provide a neutral platform for consumers while at the same time providing a worthwhile return for advertisers.Footnote 81 Supportive platforms are less susceptible to such tension since they only constitute a part of a bigger platform.Footnote 82
The difficulty in balancing competing interests suggests that stand-alone platforms are susceptible to abuse. Suppliers can purchase a persuasive presence on such platforms and may pay for their products to be promoted first.Footnote 83 Suppliers may pay for positive reviews to be highlighted, or for negative reviews to be censored, all of which can skew a consumer's assessment of a product.Footnote 84 Instead of encountering unbiased CGRs, consumers will have to deal with information aimed at persuading rather than informing. Decisions induced by a deliberate redirection of information to fit advertising interests will likely yield sub-optimal outcomes.Footnote 85
A class-action suit against a US review website, Angie's List, highlights this problem.Footnote 86 The complaint alleged that Angie's List concealed from consumers the fact that businesses could influence their ratings by paying to either appear higher in rankings or by suppressing unfavourable reviews. Where a business had paid advertising fees, negative reviews would not be counted when compiling ratings, and such reviews were made unreadable.Footnote 87 Angie's List also allegedly altered its rankings by suppressing positive reviews unless a business paid for advertising. Angie's List denied that advertising revenue could affect ratings and stated that it disclosed receipt of revenue from businesses with high ratings. It, however, agreed that in certain situations, revenue accepted from businesses could affect the ranking of search results. Eventually, Angie's List agreed to settle the lawsuit by paying the settlement class.
Platform manipulation places consumers in a vulnerable position because they are given a false sense of empowerment in believing that they are making well-informed decisions when, in fact, their decisions are based on manipulated information. Decisions based on skewed information may not reflect consumers’ preferences, thereby increasing the likelihood of detriment.
(b) Limited coverage
Informed decision-making requires that consumers assess product features and price, as well as the terms on which they will contract. If consumers pay attention to contract terms and shop around for the most favourable ones, then suppliers will be incentivised to offer fair terms.Footnote 88 Conversely, if consumers pay no attention to contract terms, then suppliers will reduce their costs and risks by offering one-sided terms.Footnote 89 To combat this, one aspect of regulating contract terms requires their transparent disclosure, especially if they significantly impact a consumer's interest.Footnote 90
Increased access to contract terms suggests that consumers will evaluate them before transacting. However, this is not always the case. Marrotta-Wurgler points out that consumers do not read the fine print, and increased disclosures do not necessarily translate to increased engagement with contract terms.Footnote 91 Some authors, however, contend that CGRs can contribute to solving this problem. For instance, Peppet argues that a consumer's review is likely to evaluate a product based on multiple variables, one of which is the consumer's experience with the supplier's contract terms.Footnote 92 Becher and Zarsky reason that consumers who encounter negative reviews about a supplier's terms will pay attention to the contract terms when deciding to transact.Footnote 93 Therefore, ex ante consumers will benefit from reviews provided by ex post consumers who have experienced the contract terms.Footnote 94
While these arguments have merit, certain matters must be considered before concluding that CGRs can assist with information on contract terms. First, CGRs will only be a useful source of information on contract terms if there are actual opinions on contract terms.Footnote 95 Unlike mandatory disclosure, where regulation dictates content, CGRs cover several issues, the scope of which is determined by the reviewer. CGRs may focus on a range of issues, and there is no obligation to discuss contract terms. Hence, CGRs may provide no information on contract terms.Footnote 96
Secondly, Becher and Zarsky appear to assume that a sufficient number of ex post consumers can understand contract terms, identify unfair ones and write reviews detailing their experiences with such terms. They also seem to assume that ex ante consumers will easily identify such reviews and act on them. Since literature confirms that few consumers read disclosures about contract terms, it means that only a small pool of consumers can write reviews assessing them.Footnote 97 Even when the terms are read, consumers may not give them considerable weight in their assessment of a product on balance with other features.Footnote 98
Becher and Zarsky's argument is questioned by empirical research suggesting that reviews are unlikely to serve as an effective conduit for passing information on contract terms because many consumers do not take contract biasFootnote 99 into account when writing reviews.Footnote 100 Chari found that highly-rated products on Amazon often have pro-seller terms and that although reviews may contain information on contract terms, they are not conveyed in a way that is useful to ex ante consumers.Footnote 101
Ex-post consumers may not write about their experiences with contract terms unless they have suffered a negative experience linked to the enforcement of such term.Footnote 102 If a negative experience never occurs, it is unlikely that such terms will be addressed in a CGR. Even where consumers have negative experiences, suppliers can adopt mechanisms to discourage them from posting reviews that draw attention to such terms.Footnote 103 These mechanisms could include prompt dispute settlement or voluntary forbearance of contractual rights in the form of refunds and exchanges. If this is the case, CGRs may not discuss the unfair term. Even if a review discusses a dispute arising from a term's application but reports prompt resolution, other consumers may discount the unfavourable term as they may believe that a similar approach (prompt and favourable resolution) will be extended to them even though this is not guaranteed.
In other instances, CGRs on contract terms may not be presented in an explicit manner that is useful to ex ante consumers. As stated above, since consumers do not often read contract term disclosures, they will usually experience controversial terms accidentally when a dispute concerning its application arises.Footnote 104 Many consumers view such disputes in the abstract and do not link them to contractual provisions.Footnote 105 Hence, a review may describe a trader's customer service as lousy/inflexible when the actual source of their discontent is the application of a contract term.Footnote 106 Such a review may not point out the problematic term(s) and, therefore, cannot place an ex ante consumer in a position to make adjustments in response to such terms.
As noted earlier, CGRs may be textual or indicated by a review score. While textual reviews will be more useful for providing information on contract terms, review scores only reveal the overall perception of a product. As consumer preferences differ, they will apportion weight inconsistently to different factors/product features when arriving at a score. If ex ante consumers cannot identify the extent to which reviews are based on the quality of contract terms, then such reviews are difficult to use as a proxy of contract term quality.Footnote 107
Many consumer transactions are based on standard form contracts (SFCs). Terms contained in SFCs are frequently updated to adapt to market, regulatory and technological developments.Footnote 108 If CGRs were to contain relevant assessments of SFCs’ terms, their effect might be negated by these frequent changes. This is because these changes can render the information flow irrelevant since reviews provided by ex post consumers on outdated terms will be unhelpful to ex ante consumers.Footnote 109 The way reviews are compiled can also exacerbate this problem. If platform operators do not set time frames for removing obsolete reviews, there is a likelihood that reviews covering SFCs that have since been updated will continue to influence consumer decision-making. Where CGRs discuss outdated terms, the information they provide will be inconsequential or misleading.
The relevance of CGRs discussing contract terms may also be negated if contract term discrimination exists. Contract term discrimination covers ex ante discrimination, where a seller offers different terms to different consumers at contract formation.Footnote 110 It also includes ex post discrimination where in resolving disputes, sophisticated or assertive consumers obtain remedies that are not available to other consumers.Footnote 111 If consumers purchasing the same product are subject to different terms, this can distort the flow of reliable information. In such situations, CGRs containing information on contract terms may not be accurately representative.Footnote 112
(c) Not suitable for all types of goods
Goods are sometimes classified based on how easily consumers can acquire and evaluate information about them before purchase. To this end, goods can be classified as search, experience, or credence goods.Footnote 113 Search goods are defined as those whose attributes can be observed and verified before purchase.Footnote 114 Experience goods are those whose quality can only be known after use, while credence goods are those whose quality cannot be determined or verified even after purchase and use.Footnote 115
When consumers purchase search goods in a bricks and mortar setting, the information contained in CGRs will only serve to reinforce or confirm judgements held about a product, since the quality is observable before purchase.Footnote 116 In online environments, detailed descriptive CGRs can take on a more prominent role in confirming the product's search attributes.Footnote 117
Information asymmetry problems are more pronounced in transactions involving experience and credence goods because consumers cannot verify their quality before purchase.Footnote 118 One concern linked to this information asymmetry is that in situations where it is impossible to establish goods’ quality in advance, consumers will only be prepared to pay an average price corresponding to the average expected quality. Sellers of high-quality goods will be unwilling to sell at that asking price and will withdraw from the market. The result is that the product quality will decline, as will the price consumers are willing to pay.Footnote 119 Reducing information asymmetry is, therefore, necessary to avoid such market failure.
While disclosures from suppliers may mitigate this problem, concerns remain that the nature of experience and credence goods makes disclosures inadequate in reducing pre-purchase uncertainties.Footnote 120 Since CGRs detail consumer experiences with products, one can assume that they will be more useful to ex ante consumers purchasing experience and credence goods.Footnote 121 In these cases, consumers often attach a higher weight to reviewer agreement than the actual information contained in reviews.Footnote 122 This means that consumers assess the credibility of the information in a review by observing the degree to which other reviewers agree on the quality of the product being reviewed.Footnote 123
Peppet, therefore, reasons that if sufficient reviews about experience and credence goods are aggregated, then such goods can transform into search goods, which are easily evaluated before purchase.Footnote 124 Peppet's argument may find merit with experience goods. Since the quality of experience goods can be evaluated after purchase, consistent reviewer agreement on certain qualities may indicate a product's suitability. However, Peppet appears to assume that information contained in CGRs is a foolproof indication of product quality. This assumption poses certain problems. First, the quality evaluations of many experience and credence products are subjective and dependent on user experience.Footnote 125 Consumer preferences differ, and so their perception of quality will reflect this, which will, in turn, influence how they evaluate products.
Furthermore, products are often complex and, one product may possess different attributes simultaneously.Footnote 126 A product may possess search, experience and credence attributes all at once.Footnote 127 Reviews of such products can provide unintentionally misleading information. This is because consumers will attach varying levels of importance to the different attributes, and their reviews may not indicate this.Footnote 128 Even where consumers can appreciate the complexity of a product's attributes, they may value the wrong features in the short term. This is because, owing to cognitive limitations, consumers do not always weigh the present and future costs and benefits of a decision and can be myopic in the short term.Footnote 129
Moreover, for CGRs to provide a credible proxy for product quality, they must communicate the actual average quality of the product being reviewed.Footnote 130 This is only possible if available CGRs are representative of the consumer population.Footnote 131 As mentioned above, consumers self-select to write reviews, making it difficult to meet this requirement.Footnote 132 Reviews are usually written by extremely satisfied or dissatisfied consumers, and do not represent the average consumer's experience.Footnote 133
Li and Hitt's empirical study of the book market is instructive.Footnote 134 A book's author can be inspected before purchase and qualifies as a search attribute, while the content can only be evaluated after reading, thereby qualifying as an experience attribute.Footnote 135 Consumers attracted to the search attributes (an author's fans) will likely buy the author's book early and will rate the book highly in comparison to the rest of the population.Footnote 136 Their reviews – though truthful – will not be representative of all consumers.
With credence goods, the situation is more complicated. Empirical evidence suggests that most CGRs focus on discussing consumer experiences with search and experience goods.Footnote 137 This is unsurprising, because it is difficult to evaluate the quality of credence products after use. Hence, consumers may be cautious about reviewing such products.Footnote 138 Moreover, because a product's search and experience attributes are easier to evaluate, consumers are more likely to write reviews in the early stage of a product's lifecycle.Footnote 139 The difficulty with assessing credence goods means that consumers will spend more time evaluating them before they write reviewsFootnote 140 (if they eventually write one).
Even where reviews of credence goods are available, they may be an unreliable source of information on product quality or suitability. This is because many complex credence goods, like financial products, are often tailored to the specific circumstances of consumers. For instance, a mortgage plan is tailored to a consumer's financial status, and what is suitable for one may be unsuitable for another. The credence nature of a mortgage plan also means that the precise value of the product may never be known. Owing to these complexities, CGRs of such products often focus on ancillary issues. For instance, a financial product's reviews may comment on a bank representative's professionalism, the ambience of a branch or the waiting time. While this is helpful information, it does not provide a clear assessment of the product's quality or suitability.
(d) Cognitive limitations
As indicated in the introduction, one reason cited for the failure of mandatory disclosure is that it relies on the neo-classical assumption that consumers are rational and are the best judges of their interests who maximise their utility with the limited resources available.Footnote 141 In reality, consumers are not unboundedly rational, and this is exhibited by judgement errors and deviations from the precepts of expected utility theory.Footnote 142 Judgement errors are often linked to the limitations associated with the human capacity to process information. When faced with too much information, consumers rely on mental shortcuts to justify their decisions.Footnote 143 Reliance on these shortcuts can lead to cognitive biases.Footnote 144
As stated above, Ben-Shahar and Schneider assert that when making unfamiliar decisions, consumers prefer advice such as reviews.Footnote 145 They appear to assume that consumers are cognitively better equipped to process information acquired through advice. This assumption cannot stand, because consumer rationality remains limited, notwithstanding the information source. Information contained in CGRs, like any other information, is subject to the limitations of human information processing.Footnote 146 Thus, cognitive biases can also affect how consumers process CGRs.
When deciding whether to purchase a product, consumers will often have access to a large pool of reviews.Footnote 147 For instance, at the time of writing, consumers wishing to buy a set of TaoTronics wireless Bluetooth headphones on Amazon will have access to about 11,193 reviews.Footnote 148 While this provides consumers with multiple perspectives, the large volume can lead to information overload. Assessing these reviews will be daunting because consumers must go through multiple reviews to decide which are credible and relevant.Footnote 149
As stated earlier, information overload will force consumers to settle on mental shortcuts, which can lead to cognitive biases. For example, consumers will be prone to the bandwagon effect when decision-making relies on CGRs. The bandwagon effect indicates that people are more likely to make a choice that is popular amongst an existing majority.Footnote 150 Platforms incorporate mechanisms that allow consumers to signal agreement with a review. For instance, Amazon allows consumers to indicate if a review is helpful. This generates statistical information on the number of consumers endorsing a review.Footnote 151 These endorsement mechanisms affect the perception of particular reviews and can induce a bandwagon effect.Footnote 152 Consumers may process such mechanisms as heuristic cues for evaluating the credibility of information supplied.Footnote 153
The level of reviewer agreement can also trigger the bandwagon heuristic. One study indicates that consumers may purchase a product because most reviews agree on the quality of certain attributes which they consider important.Footnote 154 Focusing on the online purchase of books, another study confirms this.Footnote 155 Results showed that consumers rely on the sales volume and star ratings of a book to evaluate its quality.Footnote 156 A high sales volume or star rating indicates a product's popularity. Basing a decision on these indicators can be tricky because the reviewers may not be a truly representative class, and the information signals about quality may not be accurate.
Consumers may also be subject to confirmation bias when processing information in CGRs. They may justify their decisions by latching onto particular CGRs which confirm their preconceived notions of a product.Footnote 157 Consumers tend to perceive CGRs that confirm their initial impression of a product as more helpful.Footnote 158 Since the initial impression may be influenced by the average rating of the product, one consequence of confirmation bias is that the perceived helpfulness of a review is dependent on a product's average rating.Footnote 159 If a product has high positive ratings with an average of five-stars, negative reviews are perceived as deviating from the average rating and will not be considered helpful.Footnote 160 The opposite would apply to products with low ratings. Thus, consumers may disregard CGRs that contradict their initial perception of a product informed by its ratings.
When processing information, consumers may also be subject to the overconfidence bias. Consumers can be overconfident about their abilities and may believe that in carrying out an action, they will enjoy an outcome better than the average expected outcome.Footnote 161 This will, for instance, cause consumers to overestimate their ability to navigate contract termsFootnote 162 or to ignore generic warnings because they assume that a problem will not affect them.Footnote 163 Peppet argues that online environments may neutralise consumers’ overconfidence because ‘reading online consumer reviews highlights the obvious but important fact that sheer volume of experience uncovers even very low-probability contingencies’.Footnote 164 This implies that CGRs will put consumers on notice about the occurrence of different outcomes, thus lessening the incidents of overconfident decisions.Footnote 165
Although it is tempting to agree with this view, one must note that a consumer's optimism bias can erode the gains Peppet expects. Optimism bias refers to people's tendency to believe that the probability of them facing a bad outcome is lower than it actually is.Footnote 166 Because CGRs are subjective, aggregated reviews can contain conflicting information that makes it challenging to decipher which is credible.Footnote 167 Notwithstanding a significant number of complaints about a product, some exceptional positive experiences may lead consumers to expect that the latter will be their likely outcome.Footnote 168 Thus, even with access to accurate information on possible contingencies, optimism bias may cause consumers to underestimate their personal risks.Footnote 169
Tversky and Kahneman explain that there are situations in which people assess the probability of an event by the ease with which examples can be brought to mind.Footnote 170 While this availability heuristic can be useful when assessing frequency or probability, it can also lead to bias.Footnote 171 This bias can affect consumers relying on CGRs because they may use information from specific reviews as a heuristic for what the broader picture looks like.Footnote 172 Hence the expected outcome of a transactional decision may be based on an easily recalled peer experience.Footnote 173
Consumers may make decisions relying on any of these biases and may even express satisfaction with it, but there remains a possibility that they have forgone a superior transaction.Footnote 174 With the susceptibility to these biases, one must question whether, in comparison to mandatory disclosures, consumers are better placed to process the information obtained from reviews. While CGRs may contribute towards providing consumers with information, they do not solve the problem associated with sub-optimal information processing.Footnote 175
3. Regulatory considerations
The challenge for regulators is to ensure that, where possible, consumers make decisions based on fair information. Some of the issues raised in section 2, such as bias, can be minimised through regulation. Resolving other problems linked with cognitive limitations will be trickier because of inherent human limitations. Nonetheless, since review platforms play an intermediary role in facilitating the exchange of CGRs, this paper takes the view that they should be at the centre of regulatory efforts to address some of the highlighted issues. Accordingly, discussions in this section mainly focus on review platforms and review platform operators (RPOs).
(a) Is regulatory intervention necessary?
Some authors suggest that regulation may be unnecessary because RPOs have an incentive to prevent the abuse of review platforms.Footnote 176 Friedman explains that to keep review platforms relevant, RPOs must attract, retain and engage consumers.Footnote 177 If consumers suspect that a platform is compromised, they will disengage. The competition for users, therefore, inspires RPOs to develop self-regulatory mechanisms to compete on grounds of utility and credibility.Footnote 178 Moreover, market-based solutions like litigationFootnote 179 can keep RPOs and businesses in check and deter the abuse of review platforms.Footnote 180
Friedman's argument is supported by evidence of voluntary self-regulatory efforts adopted by RPOs. For instance, Amazon allows verified purchaser badges on reviews, which helps to confirm reviewer authenticity. Google provides a right of reply, which enables businesses to respond to information contained in reviews. The review platform offered by the UK's health and beauty retailer, Boots, indicates when reviews are written by a staff-customer, whether reviews are part of a promotion and whether reviewers received a free sample. This presumably allows consumers to decide how much they can trust the information supplied. Others allow users to mark reviews as suspicious or helpful and therefore put consumers on notice that a review may be unreliable or worth considering.Footnote 181 Some review platforms use software or sting operations to detect fake reviews.Footnote 182 While commendable, these efforts are not consistent on every platform, are often non-binding, and no guarantee exists that most RPOs will uniformly adopt such measures. Moreover, as pointed out in section 2, many RPOs face conflicts of interests, which indicates that they cannot be trusted to engage neutrally in self-regulation.
Friedman's view is also problematic because it still rests on the assumption that consumers are rational and that upon detecting that a platform has left itself open to abuse, they will abandon it. As highlighted in section 2(d), consumer rationality is not always guaranteed. Secondly, important literature details the weaknesses of private law mechanisms such as litigation.Footnote 183 Plagued with economic, organisational and procedural obstacles, litigation is not always a viable option for regulating economic behaviour. Thirdly, Fink points out that isolated self-regulation lacks transparency and fails to take into account the interests of stakeholders other than the platform.Footnote 184 Hence, to safeguard public interests, external interference may be needed for a counterbalance.Footnote 185 For these reasons, evidence of self-regulation should not sustain calls for regulatory non-intervention.
(b) Relying on existing regulatory frameworks
One can argue that it is more practical to rely on existing frameworks in regulating some of the problems highlighted in section 2. For instance, before the UK House of Lords’ Select Committee on European Union, most witnesses believed that rather than introducing new regulations, the Commission ought to focus on reviewing existing laws and their application to online platforms.Footnote 186 In our context, current regulations, particularly those regulating unfair trading practices, may be of great relevance.
At EU level, the European Commission (the Commission) has made efforts to clarify the Unfair Commercial Practices Directive's (UCPD) application to review platforms.Footnote 187 The revised Guidance on the implementation/application of the UCPD has a section dedicated to ‘user review tools’.Footnote 188 This section provides some insight into how the UCPD may apply. For instance, the Commission clarifies that since the UCPD applies to natural and legal persons who qualify as ‘traders’, it covers consumers who act on behalf of traders when posting reviews. Articles 6(1)(b) and 7(4)(a) require that RPOs provide truthful information on the main characteristics of their services and therefore should not mislead consumers on the origin of reviews.Footnote 189 Platforms will engage in a misleading action contrary to Article 6 or a misleading omission contrary to Article 7 if they suppress reviews without notifying consumers that they are being presented with only selected reviews.Footnote 190 This could also be a practice contrary to the requirements of professional diligence under Article 5(2). Connections between RPOs and traders must be disclosed if the connection could materially affect the credibility of reviews.Footnote 191 Failing to do so would breach Articles 6(1)(c) and 7(2). Fake endorsement of reviews (eg ‘likes’) also breaches Article 6. Where a trader posts fake reviews in the name of consumers, they will be acting contrary to point No 22 of Annex I of the UCPD which prohibits ‘falsely representing oneself as a consumer’.Footnote 192
Relying on the UCPD, however, raises some issues. Like most EU Directives with consumer protection themes, the UCPD is drafted in the context of bilateral transactions completed between two parties: the trader and consumer.Footnote 193 The prohibitions in the UCPD focus on regulating the conduct of traders and will only apply if parties fall within the relevant definitions. CGRs introduce third-party considerations, which do not always fit with the Directive's party designations. The UCPD defines a trader to mean:
any natural or legal person who, in commercial practices covered by this Directive, is acting for purposes relating to his trade, business, craft or profession and anyone acting in the name of or on behalf of a trader.Footnote 194
Commercial practices cover:
any act, omission, course of conduct or representation, commercial communication including advertising and marketing, by a trader, directly connected with the promotion, sale or supply of a product to consumers.Footnote 195
These definitions are broad and will capture online retailers incorporating review platforms (supportive platforms). However, it is unclear whether stand-alone platforms – which only provide a forum for posting reviews – without more, can qualify as a trader. Stand-alone review platforms which do not offer advertising services may not qualify as engaging in a commercial practice. EU case law suggests that the concept of trader is defined by reference to the trader's commercial activity and that the UCPD extends only to commercial practices directly connected with the promotion, sale or supply of a product to consumers.Footnote 196 Applying this, it may be challenging to show that simply providing a review platform establishes a direct connection to the supply of products and services, unless an argument is made that the information supplied qualifies as a product.Footnote 197 Therefore, the current definitions of trader and consumer may require revision to capture transactions falling outside the traditionally defined trader-consumer relationships.Footnote 198
Currently, it appears that the UCPD will only be useful in some situations. First, the UCPD can be relied on where platforms qualifying as traders (or consumers acting on their behalf) are directly involved in practices breaching the Directive. Secondly, the UCPD will apply where platforms collude with traders in manipulating CGRs which mislead consumers (in such situations, RPOs will be seen to act on a trader's behalf). While this may cover many cases, situations falling outside these scenarios may be left outside the UCPD's reach.
Another concern is that the UCPD may be ill-suited for regulating review platforms, which continuously evolve due to changing technology. The Commission acknowledges that many practices highlighted in the UCPD's annex are designed for the offline world.Footnote 199 This may suggest that the UCPD may be ill-equipped to address specific practices that only occur in online settings. Therefore, it seems that a regulatory instrument designed with the peculiarities of the online world in mind may be more appropriate.
(c) Designing a new regulatory landscape: drawing inspiration from co-regulation and standardisation
The Commission has indicated that:
where appropriate, self-regulation and co-regulation can often achieve better outcomes for enabling the development of strong platform ecosystems in Europe and can complement or reinforce the existing legislation that already governs certain activities of online platforms.Footnote 200
As indicated in section 3(a), this paper takes the view that self-regulation will not be appropriate. Alternatively, co-regulation may present a suitable option for regulating review platforms. Co-regulation is a regulatory mechanism whereby:
a Community legislative act entrusts the attainment of the objectives defined by the legislative authority to parties which are recognised in the field (such as economic operators, the social partners, non-governmental organisations, or associations).Footnote 201
Essentially, with co-regulation, a regulator defines an objective in a binding legislative instrument while the implementation details are prepared by stakeholders in the regulated industry. Rather than a separate co-existence, co-regulation combines the use of legislation which is binding and predictable, with self-regulatory mechanisms, which are flexible in a single regulatory framework.Footnote 202 This approach can promote efficiency since it frees up legislative capacity and ensures stakeholder involvement.Footnote 203 It also promises a predictable legal framework while giving ‘sufficient flexibility to industry to shape and decide on their voluntary commitments, respecting the speed of innovation’.Footnote 204
Co-regulatory efforts have been endorsed under the EU's ‘new approach’ to regulation. The ‘new approach’ involves the adoption of a legislative instrument which defines essential requirements concerning the regulation a matter of public interest and a harmonised standard which facilitates the compliance with the essential requirements.Footnote 205 These standards are developed in an institutional setting that supports transparency and stakeholder participation.Footnote 206 Where businesses adopt the harmonised standards, they benefit from a presumption of conformity with the essential requirements set in the legislation.Footnote 207
Busch argues that a harmonised regulatory framework for review platforms can benefit from adopting this ‘new approach’ to regulation.Footnote 208 Hence, Busch suggests that the regulation of review platforms may be achieved through a Directive that defines the general regulatory principles, combined with a European standard formulated by the European Committee for Standardization.Footnote 209 The role of standardisation as a viable regulatory option has been endorsed by the International Standards Organisation, which in 2018 published the ISO 20488 (the standard) applicable to online consumer reviews.Footnote 210 The standard provides ‘requirements and recommendations for the principles and methods for review administrators to apply to their collection, moderation and publication of reviews’.Footnote 211 They apply to all organisations which publish reviews, including suppliers who collect reviews from their customers and third-party platforms, whether independent of the supplier or not.Footnote 212 Though non-binding, the standard provides an international template for further concrete regulatory efforts in the area.
(i) Imposing a harmonised duty of care
Growing evidence suggests that regulators will demand more accountability from RPOs in the near future. For instance, the Commission acknowledges that ‘in respect of access to information and content for many parts of society, platforms are increasingly taking centre stage. This role, necessarily, brings with it a wider responsibility’.Footnote 213 In jurisdictions like the UK, there are discussions about imposing a duty of care on platforms to prevent online harms.Footnote 214 The UK's Competition and Markets Authority recognises that online platforms should supervise their users and ensure they can comply with consumer protection law.Footnote 215 The ISO standards also acknowledge that RPOs should adopt best practices.
Accordingly, this paper proposes a new regulatory framework that imposes a duty of care on RPOs. This duty of care will require that RPOs reasonably ensure that mechanisms are put in place to discourage the abuse of review platforms. Broadly drafted, such duty will apply to all RPOs who provide a forum accessible to consumers located in the EU for the sharing and publication of reviews.Footnote 216 The duty will be owed to all platform users located in the EU and will cover consumers whose transactional decisions may be affected by the abuse of review platforms.
As the intermediaries central to the information exchange facilitated through CGRs, RPOs wield significant influence. As seen in section 2(a)(iii), they control the platforms and can re-engineer information by highlighting or suppressing positive/negative CGRs. Hence, it is justifiable that obligations be imposed on them to support the provision of fair information.
(ii) How will the proposal work?
Owing to the cross-border reach of review platforms, regulatory efforts at EU level will be more appropriate, as this will provide a coherent and harmonised approach to regulation. Fragmented national regimesFootnote 217 will create uncertainty and confusion for consumers and businesses, and this will, in turn, create barriers in the European Single Market.Footnote 218
Building on Busch's suggestion, the proposed regime will adopt a co-regulatory approach. Hence, to avoid a disproportionate response, a legislative instrument can define the scope and essential elements of the duty of care while guidelines and industry codes can elaborate on how RPOs may meet the duty.
Instead of a Directive, as suggested by Busch, this paper proposes the use of a Regulation as the legislative instrument of choice. There are several reasons for this. First, it addresses concerns about regulatory coherence in the Single Market, since Regulations introduce mandatory rules at Union level.Footnote 219 Secondly, a Regulation will not be subject to transposition delays associated with Directives. This will promote certainty, which results from the uniform and immediate application of adopted rules. Thirdly, it conforms with recent regulatory initiatives in the EU focusing on online platforms. For instance, to promote fairness and transparency for business users of online platforms, the EU's instrument of choice was a Regulation (ie the Online Platform Regulation).Footnote 220
To clarify how RPOs may meet the duty of care, the Commission can develop guidelines which endorse the ISO standard.Footnote 221 In line with co-regulatory principles, it can also encourage and endorse industry codes of practice. An RPO can demonstrate that it has met its duty of care if it conforms with standards/codes of practice developed by industry.
Such industry codes may build on the ISO standards in articulating more specific measures for meeting the duty of care. The ISO standards require that RPOs take certain principles into account when planning, designing and delivering review platforms. These principles include integrity, accuracy, privacy, security, transparency, accessibility and responsiveness.Footnote 222 These principles reflect common themes in existing regulatory efforts and academic work in the area.Footnote 223 These themes cover the collection, processing, moderation and publication of reviews. It also includes matters relating to transparency in the use of consolidated reviews and the disclosure of existing commercial relationships. The common themes suggest that convergence is occurring, and the proposed intervention will provide a framework for codifying these developments.
The advantages of this approach are manifold. First, these efforts can place RPOs in a position to reduce some of the problems identified in section 2. For instance, requiring that RPOs adopt best practices in processing reviews can address the prevalence of fake reviews. Reliance on misleading information may be reduced if RPOs are required to disclose paid reviews and commercial relationships with businesses. Time limits imposed on the visibility of published reviews can address concerns about consumers relying on outdated reviews where, for instance, terms and conditions have changed. Requiring that RPOs include prominent notices advising consumers to consider contract terms along with CGRs may nudge consumers to review the fine print.
Secondly, a broadly drafted duty of care whose implementation is supported by industry codes allows for future-proofing. Rapidly changing technology will affect how review platforms evolve, and an overly prescriptive framework will be ineffective. Industry codes are flexible and can be regularly updated to reflect changing technology and consumer behaviour. Thirdly, a duty of care focuses on an objective – in this context, preventing the abuse of review platforms – and leaves the detail of the means to those best placed to come up with solutions.Footnote 224 This fits with the co-regulatory approach favoured by the EU. Fourthly, a duty of care does not exclude the possibility of other measures and can provide a foundation for more detailed interventions to target specific circumstances.Footnote 225 Finally, industry efforts which are designed within a binding regulatory framework overseen by external regulators are more likely to engender consumer trust.
(iii) Compatibility with existing liability regimes
The current liability regime applying to online platforms is set under the E-Commerce Directive (ECD).Footnote 226 The ECD harmonises limitations on liability for online intermediaries that provide cachingFootnote 227 or hosting servicesFootnote 228 or which act as a mere conduit of information.Footnote 229 Relevant to our discussion is the fact that online platforms will not be liable for any illegal content hosted (or stored) on their platform if they are not aware of such information and if, upon obtaining knowledge, they act expeditiously in removing or disabling access to such information.Footnote 230
The liability exemptions cover civil, criminal and administrative liability regarding all illegal activities initiated by third parties online, including unfair commercial practices and misleading advertising.Footnote 231 In our context, this means that RPOs will benefit from the existing exemptions and will not be liable for (potentially) defamatory or false third-party reviews. EU case law supports this conclusion. For instance, Busch et al report that in 2015, in an action brought by a hotel business against a hotel review site operator, a German federal court held that the operator was not liable for a third-party review which suggested that the hotel had bedbugs.Footnote 232
One pressing question is how the proposed framework in this paper will fit with the ECD's liability regime. It is argued that the proposal is not incompatible with the ECD's regime for several reasons. First, recital 48 of the ECD states that the Directive:
does not affect the possibility for Member States of requiring service providers, who host information provided by recipients of their service, to apply duties of care, which can reasonably be expected from them and which are specified by national law, in order to detect and prevent certain types of illegal activities.
This suggests that the ECD permits other forms of liability arising from imposed duties of care as long as they are not incompatible with the exemptions in the Directive. Secondly, the ECD emphasises that Member States should not impose a general obligation on Internet intermediaries to monitor information which they transmit or store; neither should there be an obligation to actively seek out facts or circumstances indicating illegality.Footnote 233 The proposal complies with this because RPOs are not required to police review content or to ensure absolute veracity. Instead, it encourages RPOs to adopt proactive measures which minimise the abuse of review platforms.
Conclusion
Informed decision-making is crucial to protecting the interest of consumers. Traditionally, consumers have mainly relied on information emanating from businesses to bridge information gaps. However, with the growth of the Internet, consumers increasingly rely on other sources of information, such as CGRs. CGRs affect consumer decision-making and therefore warrant closer inquiry. This paper has focused on highlighting the problems associated with CGRs that justify regulatory scrutiny. The paper has also laid out possible regulatory responses to some of these problems. Owing to their role as communication intermediaries, this paper suggests that RPOs are central to regulatory efforts aimed at minimising the abuse of review platforms.
