The effect of the impugned practices
162 I now consider whether the ACCC has established that the impugned practices had the effect alleged in [20] of the concise statement (set out at [20] above). The issue is whether the MSA-masking practice and the bulk withholding practice had the effect, or likely effect, of reducing the number of recent negative reviews of the Meriton properties submitted to TripAdvisor and whether this: (a) improved, or was likely to improve, the relative number of favourable reviews compared to unfavourable reviews of Meriton properties that would have otherwise been posted; and/or (b) improved or maintained, or was likely to improve or maintain, TripAdvisor ratings or rankings of Meriton properties, and thereby created, or was likely to create, a more positive or favourable impression of the standard, quality of suitability of accommodation services provided by Meriton on consumers who used the TripAdvisor website to find suitable accommodation.
163 Before directly considering whether the impugned practices had the effect of reducing the number of negative reviews, I deal with some general and contextual matters:
(a) It would seem to be uncontroversial, and in any event the evidence establishes, that participation in Review Express substantially increased the number of online reviews of Meriton properties posted on the TripAdvisor website. Professor Malthouse gave evidence in his first report that sending automated emails to customers would increase the number of reviews posted over what he called self-motivated reviews. He described a self-motivated reviewer as one who decides to write a review for a product or service without any prompting. He outlined some of the steps that a self-motivated reviewer may have to undertake, including finding the website (such as TripAdvisor), finding the relevant page on the website, and creating a log in. He accepted in cross-examination that some self-motivated reviewers would not need to undertake all these steps (eg, if they already had a TripAdvisor account). Nevertheless, the point made in his first report remains valid: "Email solicitations make the process of writing a review substantially simpler, since the consumer often only must click on a link in the email. This makes the process more frictionless and will increase the number of reviews." That participation in Review Express substantially increased the number of reviews of Meriton properties on TripAdvisor is also supported by the information about Review Express on the TripAdvisor website (see [67] above) and the PowerPoint presentation given by TripAdvisor to Meriton in November 2014 (see [72] above). I note also that Mr Emmins (at [17.16]-[17.17] of his first report) broadly agreed with Professor Malthouse's evidence on this matter.
(b) I would also accept that participation in an automated email solicitation system, such as Review Express, is likely to increase the overall proportion of reviews that are positive. (I use the expression "positive" in the same way that Meriton used it, namely four-star or five-star reviews.) Professor Malthouse gave evidence in his first report that his research showed that the number and fraction of positive reviews would increase after automated email solicitations were introduced. He explained that self-motivated reviews have a selection bias towards negative customers. He continued: "When a customer must find the website, create a log in, etc, dissatisfied customers are more likely to invest the substantial effort to write a review. When the process is frictionless, as with email solicitations, a broader and more representative cross section of customers will write reviews. Assuming that a substantial fraction of customers are, in fact, satisfied with the product or service, having a broader group of reviewers will attract more positive reviews." This evidence was not challenged in cross-examination and I accept it. It is also consistent with the information provided by TripAdvisor to Meriton in November 2014 (see [72]-[74] above). I note also that Mr Emmins (at [17.18] of his first report) indicated his agreement with Professor Malthouse on this aspect.
164 I now turn to address the other issues outlined in [162] above, in particular whether the impugned practices had the effect of reducing the number of negative reviews. This issue is squarely addressed in the expert evidence called by each side. (Some of the lay evidence is also relevant to this issue.) In brief terms, Professor Malthouse expressed the opinion that the impugned practices had the effect of reducing the number of negative reviews. On the other hand, Mr Emmins expressed the opinion that the impugned practices did not have this effect.
165 In section 7 of his first report, Professor Malthouse addressed, first, the bulk withholding practice and then, second, the MSA-masking practice. (I note that the word "not" should read "now" in the third last line of p 15 of this report.) His evidence can be summarised as follows:
(a) In relation to the bulk withholding practice, Professor Malthouse expressed the view that if Meriton had followed the policies of TripAdvisor and sent email solicitations during the periods with service disruptions, then:
(i) the total number of positive reviews would have remained about the same (as before the service disruption), since customers who experienced the service disruption would likely not have given positive reviews, and hence there would have been few new positive reviews;
(ii) the number of negative reviews would have been greater and, moreover, those negative reviews would have been recent, and thus displayed at the top of the list of reviews, and affected the property's rank within its geographical region; and
(iii) the average rating for a particular Meriton property would have remained about the same as long as the property had many existing reviews (noting that the Meriton properties currently have thousands of reviews).
(b) Professor Malthouse expressed the view that: by suppressing email solicitations during periods of service disruptions, Meriton likely avoided having negative reviews about those service disruptions appear at the top of the list of reviews; and these negative reviews would likely have been submitted if Meriton had followed TripAdvisor's policies. Professor Malthouse accepted during cross-examination that he did not have empirical evidence for his opinion that reviews submitted during a period of service disruption would be disproportionately negative. However, he maintained that this was his opinion.
(c) Professor Malthouse accepted during cross-examination that self-motivated reviews are still possible during a period of service disruption. He also accepted that if (as he expects) more negative reviews are likely to arise during a period of service disruption, then it is likely that there would be a higher number of self-motivated reviews during that period.
(d) In relation to the MSA-masking practice, Professor Malthouse expressed the view that whether or not Meriton resolved the complaint to the customer's satisfaction was relevant. He explained that: the effects of complaint resolution have long been studied in the services literature; satisfaction with complaint resolution has a positive impact on customer loyalty; and complaint resolution is thus an important element in a company's customer retention strategy. He addressed two distinct scenarios:
(i) The first scenario was where Meriton did not resolve complaints to its customers' satisfaction. In this scenario, masking the email addresses of those customers who complained would not affect the number of positive reviews, but would reduce the number of negative reviews. In other words, in his opinion, without masking, there would be more negative reviews.
(ii) The second scenario was where Meriton did resolve complaints to its customers' satisfaction. In this scenario, the number of positive reviews would be higher without masking than with masking. He stated: "Under the scenario of positive complaint resolution, Meriton is hurting itself with the masking process." In the course of cross-examination, Professor Malthouse accepted that the expression "service recovery" was used to describe an intervention by the retailer or supplier whereby it was able to repair the harm that might otherwise have been done by a negative feature of a product or experience. He also accepted that the literature he had reviewed showed that successful service recovery can increase customer satisfaction over the customer who never had the problem in the first place. Professor Malthouse also accepted that this analysis would apply where emails had not been forwarded "en bloc" if the complaints of the affected group of customers had been resolved to their entire satisfaction. However, the evidence generally suggests that positive service recovery (in the sense discussed by Professor Malthouse) occurred in only a small proportion of the cases in which the impugned practices were implemented. There are only a few examples in the evidence of such positive service recovery. If Meriton was generally achieving positive service recovery, which would have been likely to generate positive reviews, it is unlikely that it would have adopted the impugned practices. While it is theoretically possible that it did not realise that it was depriving itself of a large number of positive reviews, it seems unlikely that Meriton, a large and sophisticated business, would not have appreciated this. For completeness, I note that I do not think the fact that Ms Nazha placed a response on TripAdvisor to every review (whether positive of negative) detracts from these conclusions. That response was posted after the customer had already placed a review (whether positive or negative). It did not generally constitute positive service recovery as described by Professor Malthouse.
166 I note for completeness that in his first report, Professor Malthouse referred to "snippets" of the reviews that appeared on the TripAdvisor website, as illustrated in Fig 3 to the report. A snippet appears to be a short one-line extract taken from the text of a customer review. Professor Malthouse accepted in cross-examination that whether the snippet was visible may depend on the "particular method of interrogation" of the website that was adopted. I therefore put this aspect to one side.
167 I also note for completeness that in the course of cross-examination, Professor Malthouse was asked some questions about an article he co-authored with E Maslowska and S Bernritter entitled "Too good to be true: the role of online reviews' features in probability to buy" (2017) 36(1) International Journal of Advertising 142. In the article, the authors concluded (at p 154) that, for the categories of products studied, products with an average star rating of 4.5 to 5 were less likely to be purchased than those between 4 and 4.5 stars. The authors stated that the reason for this may be that consumers perceive such positive reviews (ie, of 4.5 to 5 stars) as too good to be true. The authors stated that this finding suggests that it is important to have some fraction of non-perfect reviews, as a few negative messages can increase perceived credibility. The authors agreed with previous researchers who had expressed the view that companies should not censor negative reviews.
168 Mr Emmins approached the question of whether the impugned practices had an effect on the number of negative reviews, broadly, in three ways. (It is important to note that Mr Emmins used the expression "negative review" to refer to a one or two-star review, whereas Meriton also considered a three-star review to be a negative review.) First, Mr Emmins looked at the cases of service disruption identified by the ACCC in the concise statement and reviewed the TripAdvisor website at about the times of these disruptions and counted how many negative reviews there were. Secondly, Mr Emmins relied on the TrustYou Reports to compare the rating and volume of reviews of the Meriton properties on Booking.com and TripAdvisor. Thirdly, Mr Emmins personally looked at reviews on the Booking.com website, particularly at the times of the service disruptions referred to by the ACCC. I deal with each of these approaches in turn.
169 In relation to Mr Emmins's first approach, I note the following:
(a) In sections 2-7 of his first report, Mr Emmins set out data and conclusions based on his analysis of the TripAdvisor website at about the times of (and immediately after) the instances of the bulk withholding practice referred to in [16] of the concise statement. That paragraph of the concise statement alleges four instances of the bulk withholding practice: (a) in relation to Meriton's Bondi Junction property in the period 16 April 2015 to 29 April 2015 (when phone lines were down, there was no hot water available and a lift was not operating); (b) in relation to Meriton's Kent Street property between 1 and 8 June 2015 (when there were lift delays); (c) in relation to the Kent Street property on 15 June 2015 (when the hot water was unavailable); and (d) in relation to Meriton's Pitt Street property on 10 August 2015 (when there was an evacuation of guests on one level of the apartments due to a damaged gas line, which also caused a hot water failure for guests on other levels). Mr Emmins analysed the reviews on the TripAdvisor website for these properties at about the times of (and immediately after) these service disruptions and noted the number of reviews that were one, two, three, four or five-star reviews. He concluded that: there was a total of 72 self-motivated reviews (ie, unprompted reviews) posted about the relevant properties during the specified periods, of which only three were negative (ie, one or two-star reviews); and within the content of these negative reviews, there was no mention of the problems (ie, specific service disruptions) cited by the ACCC in [16] of the concise statement. On the basis of these conclusions, Mr Emmins expressed the view that he disagreed with [20] of the concise statement "as my research and analysis of the TripAdvisor data pertinent to the cited periods shows no evidence that the masking practice described had any discernible effects, improvements or maintaining of the positions on TripAdvisor or was likely to have had any discernible effects, improvements or maintaining of the position on TripAdvisor as described".
(b) One immediate difficulty with Mr Emmins's first approach is that he adopted a different definition of negative review to that used by Meriton. This point was raised with Mr Emmins during cross-examination and he maintained his position that only a one or two-star review should be considered negative. However, in the particular circumstances of Meriton's business, its own view of what constitutes a negative review would appear to be relevant. There are examples in the evidence of three-star reviews that may aptly be described as negative. In any event, Mr Emmins did not address the difference between his characterisation and that of Meriton, or discuss whether it affected his conclusions. Further, as a result of the approach he took, Mr Emmins did not look for references to the service disruptions in three-star reviews.
(c) Another difficulty is that Mr Emmins did not keep proper records of the reviews he analysed, making it difficult to test some of his conclusions.
(d) Mr Emmins gave oral evidence-in-chief that he had carried out a similar exercise to that described above in relation to the service disruptions identified in certain paragraphs of the updated schedule to the concise statement. He did not produce any documentary record of this work, including in his two reports. He said that he had not seen specific references to the service disruption issues on the TripAdvisor pages for the relevant Meriton properties and, on this basis, he presumed there were no negative reviews on the subject. In cross-examination, Mr Emmins said he did not keep a record of this work in his notebooks. In the absence of any documents to demonstrate that this work was carried out, and the results of this work, I do not place any weight on Mr Emmins's analysis of the TripAdvisor website in respect of any incidents alleged in the updated schedule to the concise statement, as he described in his oral evidence-in-chief.
170 In relation to Mr Emmins's second approach, I note the following:
(a) In sections 8-11 of his first report, Mr Emmins relied on the TrustYou Reports as the basis for comparing reviews on Booking.com and reviews on TripAdvisor about certain properties. Specifically, Mr Emmins examined the three Meriton properties referred to in [16] of the concise statement, namely Bondi Junction, Kent Street and Pitt Street. The premise of comparing reviews on Booking.com with those on the TripAdvisor website was that the Booking.com reviews were not affected by the impugned practices. This premise is established by Mr Chan's evidence and is not disputed by the ACCC. The TrustYou Reports comprise the dashboard reports for several of the Meriton properties for the period 1 November 2014 to 31 October 2015. Each dashboard report contains an Overall Score Graph and a New Reviews Graph, as described in [42] above. Mr Emmins's analysis was substantially the same for each of the three properties. It is therefore sufficient to refer, by way of example, to his discussion of the Bondi Junction property. Mr Emmins referred to the Overall Score Graph and stated: "This graph clearly shows a very consistent 'score' across the entire period - including the period cited by the ACCC [in [16] of the concise statement] - and across the reviews websites. There are no discernible fluctuations." After referring to the New Reviews Graph, Mr Emmins expressed the opinion: "I note that there is significant consistency between the figures relating to Booking.com and TripAdvisor, including during the period cited by the ACCC. Given that there could have been no interference with the Booking.com results, this indicates, that actions by Meriton employees during the period cited by the ACCC did not discernibly affect review volume." After referring to the number of pages in the TrustYou dashboard report for the Bondi Junction property and that it has a "number of different breakdowns of data", Mr Emmins expressed the view: "There appear to be no significant fluctuations in consistency of the volume of posting of reviews or the ratings scores being awarded by consumers during the period cited by the ACCC. I am of the opinion that if the actions of Meriton employees had significantly affected the volumes or ratings scores, then there would have been evidence of this in the TrustYou data." Mr Emmins also stated that he believed that this (ie, the impact of the impugned practices) would have been made apparent from: fluctuations in the data relating to TripAdvisor concerning volumes during the period cited by the ACCC; a deterioration in the satisfaction performance of the relevant property on Booking.com during the period cited by the ACCC; and fluctuations in the comparative satisfaction performance data between TripAdvisor and Booking.com during the period cited by the ACCC. In his supplementary report, Mr Emmins added that the fluctuations in the data that he would have expected to see if the masking practice had significantly affected the volume of reviews or rating scores were not present, and therefore he concluded that the masking practice described in the concise statement had not had the effect or likely effect of "changing the message given to consumers" who visited the identified pages of the TripAdvisor website.
(b) The fundamental difficulty in relying on the New Reviews Graph is that, as Professor Malthouse pointed out in his second report, the data is presented on a monthly basis. It is therefore not possible to determine from this data whether there were fluctuations in the volume of new reviews on Booking.com or TripAdvisor in the days of and immediately following the four service disruptions examined by Mr Emmins (namely the service disruptions described in [16] of the concise statement). The data simply is not sufficient granular to enable any conclusions to be drawn about whether there were fluctuations as between the two websites in respect of these specific service disruptions. There may well have been fluctuations between them during the course of the month, but it is not possible to tell.
(c) In relation to the Overall Score Graph, as Professor Malthouse noted in his second report, when there are hundreds of reviews, adding a few less positive reviews would have a negligible effect on the average star rating. Another difficulty raised in Professor Malthouse's second report is that "we cannot know which reviews were written by people who were affected by the service outages", as customers do not necessarily leave reviews contemporaneously with their stay at a particular property and may do so well after their date of departure.
(d) Apart from the Overall Score Graph and the New Reviews Graph, it is not apparent from Mr Emmins's reports how he relied on the balance of the dashboard report for each of the three properties. No submissions were made indicating how any other part of the dashboard reports supported Mr Emmins's conclusions. It is not clear how the balance of the dashboard reports supports Mr Emmins's opinion.
171 In relation to Mr Emmins's third approach, I note the following:
(a) In sections 12-13 of his first report, Mr Emmins stated that he had carried out research on the Booking.com site "in order to further evaluate the question of whether the service disruptions created negative perceptions by guests that may have led to the posting of negative reviews". However, very little detail was provided as to the work carried out or its outcome.
(b) In his supplementary report, Mr Emmins stated that, from the Booking.com data (which would seem to refer to data he derived from his research on the Booking.com website) he drew the conclusion that there was no discernible difference in the (relative) volume or rating given in reviews, in the relevant periods when the masking practice was in operation, between the Booking.com website and the TripAdvisor website. Mr Emmins also stated that his personal examination of the Booking.com website corroborated his opinion, formed on the basis of the TrustYou data, that the service disruption periods cited by the ACCC did not cause any increase in negative reviews.
(c) In cross-examination, Mr Emmins said that, when he prepared his first report, he had looked at the Booking.com data himself. He said that this material was not identified (in section 18 of his first report) as a document relied upon because he had made the mistake of not actually printing out the documents that he had looked at. He explained that: he had used a computer to look at the data on the Booking.com website; and he had then manually counted up the number of reviews. When asked how he did this when the Booking.com website would not enable a person to go back beyond two years, he responded "actually, I cannot recall". Mr Emmins accepted that annexure CE-6 to his second affidavit, which was said to comprise the Booking.com reviews relating to Meriton's Bondi Junction property that he had looked at in preparing his first report, did not in fact comprise those reviews. Apart from one 2015 review, the rest of the 62 pages comprised reviews from 2017.
172 In section 14 of his first report, Mr Emmins stated that he had carried out some further research, comparing reviews of Meriton properties on the TripAdvisor website before and after the introduction of Review Express. A summary of reviews for Meriton's Bondi Junction property posted in July 2013 (before Meriton started using Review Express) and in September 2013 (after it started using Review Express) was set out. Mr Emmins expressed the view that this data helped to substantiate that the use of Review Express: increases the overall number of reviews; increases the number of positive reviews; and appears not to increase the number of negative reviews. He stated that the data appeared to confirm that "reducing the use of the Review Express system would not be of any benefit to Meriton".
173 In section 15 of his first report, Mr Emmins set out his responses to certain questions he had been asked by Meriton. He expressed the opinions (based on the analysis set out earlier in the report) that: neither the masking practice nor the failure to send emails referred to in the concise statement would have had the effect, or likely effect, of reducing the number of recent negative reviews of Meriton properties submitted to TripAdvisor; he does not believe that the actions of the Meriton employees referred to in the concise statement would have been likely to have improved the relative number of favourable reviews compared with unfavourable reviews of Meriton properties that would otherwise have been posted on TripAdvisor; and he does not believe that the actions of Meriton's employees would have (or would have been likely to have) improved, maintained or prevented a reduction in the TripAdvisor ratings of the Meriton properties.
174 In response to Professor Malthouse's first report, Mr Emmins stated in his first report that: he did not find any evidence of the number of negative reviews (being reviews submitted by guests who were not using the Review Express system) increasing during the periods of service disruption cited by the ACCC in the concise statement; and he disagreed with Professor Malthouse's assumption that a service disruption issue would necessarily increase the number of negative reviews.
175 I note the following further matters arising from the cross-examination of Mr Emmins:
(a) Mr Emmins was taken to the part of his résumé where he referred to work he had done for the UK Competition & Markets Authority with regard to online reviews and endorsements. The résumé stated that he had been asked to provide, among other things, "[e]xamples of gaming review sites such as TripAdvisor". In cross-examination, Mr Emmins accepted that the masking of emails so as to prevent emails going out to persons whom the relevant accommodation provider feared might write negative reviews was an example of the "gaming" he referred to in his résumé.
(b) Mr Emmins was taken to a document published by the International Consumer Protection and Enforcement Network (ICPEN) titled "Online Reviews & Endorsements - ICPEN Guidelines for Traders & Marketing Professionals" (June 2016). Mr Emmins said he was very familiar with the document. In the context of that document, Mr Emmins was asked some questions about the principle of not preventing consumers from seeing the whole picture of genuine, relevant and lawful reviews. Mr Emmins referred to a situation where a review is published that the business wishes to have taken down on the basis that it is biased or unfair. Mr Emmins said that the situation is a matter between the business and the review site to determine "what is fair and what is not". The following exchange then occurred:
… It follows from your answer a few moments ago that you consider the appropriate response, if a business regards a review as unfair, as you put it, that the business will deal with that review not by attempting to silence the critic in any respect, but, rather, by dealing with the review site; is that correct?---Yes.
And you would regard it, having regard to these principles, as improper for a business to endeavour to silence its critics in any respect, wouldn't you?---I would.
Do you agree that's what Meriton was doing in this [case]?---I do.
And you agree, therefore, it was wrong to do it?---I do.
(c) Mr Emmins was taken to a bundle of documents comprising correspondence between Meriton and himself that had been produced during the trial in response to a notice to produce (the notice to produce documents). In an email dated 2 April 2017, Mr Emmins expressed the view that Professor Malthouse's first report "whilst competent in terms of a general academic interpretation … lacks sufficient focus on the more precise ways that consumers actually search and the processes that are involved which may or may not lead to a purchasing decision". He expressed the view that, in this respect, it was "fundamentally flawed". In cross-examination, Mr Emmins accepted that he did not elaborate on these matters in his own report and, in this respect, his own report was flawed.
(d) Mr Emmins said that, in the course of preparing drafts of his first report, Meriton's in-house counsel cautioned him to "stay on track in terms of what the case was about, and the case was about Meriton's actions, not TripAdvisor's flaws" and he (Mr Emmins) "believed that was good advice".
(e) Mr Emmins was taken to a draft of his first report, which contained material that was subsequently removed and not included in the final version of the report. Although he initially sought to explain the removal of this material on the basis that it "looks like an attack on TripAdvisor", he subsequently accepted that this did not provide an explanation (T290). Although Mr Emmins rejected the proposition that this material was an "inconvenient truth" for Meriton in this case, he was unable to provide a reason why it was taken out. The fact of the removal of this material, and the absence of any real explanation, raises a concern that it may have been removed to assist Meriton's case. (See also T295, 297-298.)
(f) Mr Emmins was taken to correspondence in the notice to produce documents in which he had written that, unless certain additional information was provided to him, "I do not believe the evidence will be as much in the favour of your defence as possible". It was put to Mr Emmins in cross-examination that this was not the language of an independent expert. He responded: "I would take your point on that."
176 In the discussion that follows, I adopt the definition of "negative review" used by Meriton, namely a one, two or three-star review.
177 In my view, the opinion of Professor Malthouse that the impugned practices had the effect of reducing the number of negative reviews of Meriton properties appearing on TripAdvisor is to be preferred to the contrary view of Mr Emmins. First, Professor Malthouse's written material was presented in a clear and logical manner. In contrast, Mr Emmins's written evidence did not clearly articulate the research he had carried out, or provide detailed reasons for the conclusions he had reached based on that research. Further, there was a lack of documentary support for some of this work, making it difficult to test his conclusions. Secondly, in his oral evidence, Professor Malthouse impressed me as having a better command of the subject matter. He was able to explain why he reached the conclusions that he did with greater clarity and cogency, in my view. Thirdly, a difficulty with Mr Emmins's evidence generally was that he adopted a different definition of "negative review" to that adopted by Meriton. There is force in the proposition that Meriton had a better understanding of what constituted a negative review in the context of its own business. In any event, the implications of the difference were not addressed by Mr Emmins. Fourthly, a fundamental difficulty with the second approach taken by Mr Emmins is that the data in the New Reviews Graphs is presented on a monthly basis. It is therefore not possible to determine from this data whether there were fluctuations, at or about the times of the service disruptions referred to in [16] of the concise statement, in the volume of new reviews on Booking.com or TripAdvisor. Further, in relation to the Overall Score Graph, as discussed further below, when there are many hundreds of reviews for a given property, adding a few less positive reviews is likely to have a negligible effect on the average star rating. Fifthly, Professor Malthouse's opinion, to the effect that the impugned practices reduced the number of negative reviews appearing on the TripAdvisor website, is inherently plausible and accords with the lay evidence referred to earlier in these reasons. As set out in the reasons above, Meriton's intention in engaging in the impugned practices was to reduce the likelihood of negative reviews being posted on TripAdvisor. Moreover, Meriton's staff responsible for monitoring the TripAdvisor website (in particular, Ms Nazha) clearly thought that the impugned practices had the effect of reducing the number of negative reviews. For example, as set out at [93] above, Ms Nazha sent an email commending the spreadsheet (with an "Email MSA'd" column) adopted by Kent Street. She commented that they had recently implemented this approach, and were showing "great results with their reviews". Ms Nazha accepted in cross-examination that, by this comment, she meant far fewer negative reviews.
178 It should be noted that Professor Malthouse expressed the opinion that, in cases where a complaint or a service disruption issue was resolved or dealt with by Meriton to the customer's satisfaction, the impugned practices would not have had the effect of reducing the number of negative reviews. I accept this evidence. However, as noted above, the evidence generally suggests that positive service recovery occurred in only a small proportion of cases in which the impugned practices were implemented. I therefore do not consider this scenario to be of such significance as to detract substantially from the overall position.
179 It is true that the impugned practices did not prevent a self-motivated reviewer from posting a negative review. However, the evidence establishes that there are many guests who will leave a review (including a negative review) if prompted by an email invitation, but who would not otherwise post a review on TripAdvisor given the additional effort that is involved. Accordingly, while it may be accepted that negative reviews may still have occurred, this does not detract from the proposition that the impugned practices had the effect of reducing the number of negative reviews.
180 Further, in my view, that the impugned practices had the effect of improving the relative number of favourable reviews compared to unfavourable reviews of Meriton properties on the TripAdvisor website. The impugned practices reduced the number of negative reviews (compared with the position if the impugned practices had not been implemented). And while the impugned practices may also have reduced the number of new positive reviews, I consider that these would have been far fewer than usual on account of the occurrence of a service disruption issue or customer complaint that led to the masking.
181 Importantly, the TripAdvisor website generally displays the most recent reviews first. This means that, if (as I have found) the impugned practices had the effect of reducing the number of negative reviews, there are likely to have been more extended periods of time when, but for the impugned practices, the first review (or the first few reviews) that a consumer would have seen on accessing the page for the relevant property on the TripAdvisor website would have been a negative review (or negative reviews). While the circumstances of consumers looking at the TripAdvisor website to choose accommodation vary from case to case, it may be inferred that a recent negative review (or a few recent negative reviews) complaining, for example, about lift delays at high rise accommodation, is likely to have conveyed a more negative impression and influenced the consumer. Based on the evidence generally, and the findings I have made above, I consider it likely that there were periods of time when, but for the impugned practices, a negative review or a few negative reviews would have appeared first on the TripAdvisor page for a Meriton property. Although it is true that a self-motivated reviewer may have in any event posted a negative review about the particular issue at the property, it is likely that in some cases that did not occur or did not occur until some time later.
182 Whether the impugned practices improved or maintained the TripAdvisor ratings and rankings of the Meriton properties, is a more difficult question. I will deal with the question of rating and ranking separately.
(a) In relation to the rating of a property, the evidence of Professor Malthouse establishes that, where there are a large number of reviews (as was the case with the Meriton properties), a single review will have very little effect on the overall rating. As Professor Malthouse said during cross-examination, in relation to TripAdvisor: "We do not know the exact weight function, but … unless it were a very extreme case where they gave zero weight to the majority of reviews, the statement would hold that when you have a lot of reviews, and you add one more, the effect is very small." He also said, in relation to the recency of reviews: "There would be slightly more weight given to that more recent one, but we do not know exactly how much more weight." In light of this evidence, it is not established that the impugned practices affected the rating of the properties on TripAdvisor.
(b) The evidence generally suggests that a single review may have a relatively greater influence on ranking than on rating. While there are effectively only 10 different positions with respect to rating (including half stars), in some geographical areas the number of different ranking positions was much greater. For example, several of the Meriton properties are in the Sydney area. There were, at the relevant time, 183 properties in this area (see [73] above). As there are more increments in ranking, this suggests that the relative influence of a single review may be greater with respect to ranking compared with rating. Further, it is very possible that many properties in a given area may attract similar ratings, and be ranked closely together, such that a few negative reviews might affect the ranking. The lay evidence supports the view that a recent negative review had the capacity to affect a property's ranking. On a number of occasions, Ms Nazha expressed the view in contemporaneous documents that a property's ranking had been affected by recent negative reviews. I refer, in particular, to Ms Nazha's email of 18 January 2016: see [158] above. I refer also to Ms Nazha's email of 10 August 2015: see [124] above. Given how closely Ms Nazha monitored reviews of Meriton's properties on the TripAdvisor website, I place considerable weight on her contemporaneous observations in this regard. Further, as set out in [159] above, Mr Chan accepted that the receipt of negative reviews was a factor in the decline in the ranking of the hotels referred to in the 18 January 2016 email. On the basis of this evidence, I infer that a single review, or at least a small number of reviews, may affect the ranking of a property on the TripAdvisor website. I would also find, on the basis of this evidence, and taking into account my earlier findings, that in some cases the impugned practices affected the ranking of a Meriton property on the TripAdvisor website. For example, but for the impugned practices, a property may have been ranked "3 out of 183" rather than "2 out of 183". While the circumstances of consumers looking at the TripAdvisor website to choose accommodation vary from case to case, it may be inferred that a difference in ranking is likely to have influenced the consumer in some cases.
183 For these reasons, I find that the MSA-masking practice and the bulk withholding practice created a more positive or favourable impression of the quality or amenity of the Meriton properties on the TripAdvisor website. Moreover, when the scale of the MSA-masking practice and the frequency of the bulk withholding practice (as discussed earlier in these reasons) are taken into account, this effect was substantial.