Reflections on US v Bazaarvoice
The Antitrust Division signals its readiness to litigate closed mergers
In June 2012, an attorney fresh out of law school walked into my office and handed me a press report that Bazaarvoice had acquired PowerReviews for $170 million. The story did not have much in the way of detail other than to note that both companies offered ratings and review, or R&R, platforms. R&R platforms allow online retailers and manufacturers to collect and display consumer-generated product ratings and reviews that many online shoppers consider before making a purchasing decision.
The attorney first explained that he had consulted with the Antitrust Division’s experts, and they did not believe the transaction would be reportable under the Hart-Scott-Rodino (HSR) Act of 1976. The attorney also said he had reviewed an HSR filing about six months before that involved the e-commerce industry. Although the filing did not involve the market for R&R platforms, it contained documents that described the market for R&R platforms as a “duopoly”. That filing had since been lost to the shredder, but he was certain that the two duopolists were Bazaarvoice and PowerReviews.
As a result of this attorney’s diligence, the division was able to open an investigation of this non-reportable transaction two days after it was consummated. When the attorney left my office, I did not expect that the division was about to embark on 30-month journey that would result in a fully litigated merger case – one of the few during the Obama administration – a litigated remedy proceeding and, ultimately, a settlement that remedied the potential harm caused by the deal and restored the market to its pre-merger condition.
The division conducted a six-month investigation of the deal after Bazaarvoice agreed to maintain PowerReviews’ assets until the DOJ had decided the matter one way or the other. During the investigation, division staff came to believe that the deal was likely to lead to less innovation and higher prices for online retailers to use Bazaarvoice’s R&R platform, which would essentially be the only such platform on the market were the merger to stand. On 10 January 2013, the division filed a lawsuit in the San Francisco federal court to unwind the transaction and restore the competition that it extinguished. The trial lasted from 23 September through 15 October that year, with the trial team working through the 17-day government shutdown that began at the end of September.
Although furloughed as “non-essential” workers, the trial team appeared in court on 1 October and presented a motion to postpone the trial. The division also advised the court that both parties and the witnesses scheduled to testify that day were prepared to proceed if the court denied the motion. The court without hesitation denied the motion and asked Bazaarvoice to call its next witness. The court heard testimony from 40 witnesses, read more than 100 depositions and reviewed approximately 1,000 exhibits.
The court on 8 January 2014 issued a “necessarily lengthy” 141-page opinion concluding that Bazaarvoice’s acquisition of its primary competitor would lead to anticompetitive, unilateral effects – the ability for Bazaarvoice to wield unchecked power over its customers – and therefore violated Section 7 of the Clayton Act. The court’s opinion contains several key takeaways for the enforcement of the antitrust laws in the high-tech industry.
First, the Bazaarvoice decision supported the proposition that antitrust laws apply with full force to dynamic, high-tech markets. The court concluded that “while Bazaarvoice indisputably operates in a dynamic and evolving field, it did not present evidence that the evolving nature of the market itself precludes the merger’s likely anticompetitive effects.” Secondly, the court effectively neutralised the holding in US v Oracle1 that a post-merger monopoly is necessary for unilateral, anticompetitive effects. Citing US v H&R Block,2 the court declined Bazaarvoice’s invitation to impose a market share threshold for proving unilateral effects.
Instead, the court firmly endorsed the likely anticompetitive effects approach in the revised 2010 Horizontal Merger Guidelines. Thirdly, the court rejected the argument, commonly made by merging parties in the high-tech industry, that large technology companies such as Amazon, Google and Facebook could quickly enter the market and dissipate any anticompetitive effects, particularly where Bazaarvoice offered no evidence that these firms had even considered entering. On this point, the court stated: “[t]o conclude otherwise would give e-commerce companies carte blanche to violate the antitrust laws with impunity with the excuse that Google, Amazon, Facebook or any other successful technology company stands ready to restore competition to any highly concentrated market.” Finally, the court rejected Bazaarvoice’s argument that US v Syufy3 established an alternative methodology for analysing consummated mergers. The court confirmed that consummated mergers must be evaluated using the same standards that apply to mergers reported under the HSR Act.
Having established liability, the court ruled that the DOJ was entitled to an injunction that required Bazaarvoice to divest PowerReviews. On 12 February 2014, the division filed a proposed final judgment designed to eliminate the anticompetitive effects of the merger. Bazaarvoice initially contested the division’s remedy, but ultimately the parties reached a settlement that required the divestiture of PowerReviews to a third party, Viewpoints Inc, and other remedial measures designed to restore the level of competition that would have developed if the merger had not occurred. The final twist in the case occurred in July 2014 when the division learned that Bazaarvoice may have improperly used certain PowerReviews technology that should have been divested to Viewpoints. Following an investigation, the parties resolved the dispute with an amended settlement that required Bazaarvoice to appoint an internal compliance officer to oversee the company’s efforts to comply with the final judgment and the antitrust laws. With the entry of the amended final judgment on 2 December, the court ended the Bazaarvoice journey.
The role of customer testimony
Bazaarvoice relied heavily on customer testimony to support its argument that the merger had not substantially lessened competition. Bazaarvoice deposed more than 100 customers – averaging more than one customer deposition a day – during the three months of fact discovery. Bazaarvoice did not depose any PowerReviews customers. Instead, Bazaarvoice selected customers that for a variety of idiosyncratic reasons selected a fringe competitor, developed an in-house solution or determined they did not need an R&R platform. Most depositions were short, averaging about an hour.
Most of these customers testified at trial or appeared by deposition. In response to leading questions, most testified that the merger wouldn’t harm and didn’t particularly concern them. Often there was little or no foundation for these opinions. Bazaarvoice argued that because this case involved a consummated merger, evidence that the merger hadn’t hurt its customers should be given substantial weight. The division also called four customer witnesses. These witnesses testified that they viewed the merged firms as close substitutes; viewed the fringe competitors as weak alternatives; and, in the face of an R&R price increase, would not switch to alternative e-commerce products, such as Facebook or blogs. These witnesses did not offer opinions on whether the merger should be unwound.
Acknowledging the importance of customer testimony, Bazaarvoice called a customer as its first witness. The defence got off to a rocky start. On cross-examination, the customer made a series of admissions that appeared to undermine his own credibility and to support the division’s theory. It could have been a record for the fastest attack on a witness’s credibility in any case, in any court, anywhere: division lawyers challenged the witness immediately after taking the oath. After being sworn, the witness was asked if he took the oath at his deposition seriously. He said that he did. The division then played the video of his swearing-in at his deposition and noted that the witness winked at the camera. “Why did you wink?” the division lawyer asked. The witness responded that he winked at the female attorney taking the deposition, who was now sitting at the defence table. This insensitive remark may have been the high-water mark of his testimony. The witness next admitted that the sworn declaration he submitted to the division during its investigation was not accurate. He admitted that his statement claiming he had many options for R&R platforms was not accurate. When asked why he signed an inaccurate declaration, the witness explained he was concerned Bazaarvoice would raise his price if it knew he had no other options. He further undermined any favourable testimony when he next admitted he had a financial interest in the outcome of the case.
The last customer witness called by Bazaarvoice came as a surprise. He was outside the jurisdiction of the court and at first was not willing to come to San Francisco to testify. But the trial ran longer than expected and it turned out the witness was going to be in San Francisco anyway, so he agreed to testify. He was a well-informed and sophisticated customer, who had recently conducted a detailed evaluation of his R&R options and selected another competitor. His testimony prompted the division to adopt the rare cross-examination strategy: “We have no questions for this witness.” Better to get the witness off the stand before he has the chance to hurt the division’s case any more than he could under direct questioning from the company’s lawyers.
The division’s economic expert was the last witness called at trial. Bazaarvoice focused its cross-examination on the views of customers that supported the deal. Bazaarvoice asked questions that suggested those customers’ views should be given significant weight. Bazaarvoice asked the witness who was in the best position to understand the effects of the merger: 100 customers or the expert. The expert answered that he was. With a dramatic flourish, defence counsel sat down and Bazaarvoice rested. The defence presumably intended that this abrupt ending to the trial would give the impression that the division and its expert were arrogant to suggest that they were in a better position to determine what was best for customers than the customers themselves.
The court agreed with the division and its expert. The court found that the customer testimony on the likely effect of the merger “was speculative at best and is entitled to virtually no weight”, given the customers’ narrow perspective on the legal and economic questions before the court. The court found that these customers were not positioned to predict the competitive effect of the merger because they were not privy to most of the evidence in the case, including the expert testimony; had paid little or no attention to the merger; had never given any thought to the merger; were either unaware of alternatives or had conducted a limited review of their alternatives; and each had an idiosyncratic understanding of R&R platforms based on the priorities of their company and different levels of knowledge. The court also recognised that evidence regarding pricing and the effects of the merger could have been subject to manipulation by Bazaarvoice. The division informed Bazaarvoice of its investigation two days after the transaction closed. Given this scrutiny, the court, relying on US v General Dynamics,4 found that it was unlikely Bazaarvoice would have imposed an anticompetitive price increase while under federal investigation.
The Bazaarvoice case demonstrates that customer views on a merger can be very informative regarding their preferences, but of limited value when trying to judge the effects of a merger, whether positive or negative. The court gave significant weight to customer testimony on important factual issues relevant to market definition and market power. The court described customers as “the most credible source of information on their need for, use of and substitutability of social commerce products, as well as regarding their companies’ past responses to price increases.” Less helpful to the court are uniform customer opinions on the ultimate question of whether a merger should be blocked, or cleared, because customers do not have access to the information or expertise required to predict how a deal might ultimately affect competition.
Despite the impressive number of customer testimonials Bazaarvoice presented, it was not true that most customers believed they would not be harmed by the merger. During the investigation, the division interviewed well over 100 customers and all were asked their views on the merger. As one might expect, some favoured the merger, some opposed it and some, perhaps the majority, held no views. Soliciting the views of customers is useful for internal decision making, but great care must be used in how their testimony is presented in court. Customer testimony is likely to be given the most weight when supported by evidence that the customer has recently analysed the market, and when the customer testimony is consistent with documents, expert testimony and other objective evidence. Here, the opinions offered by Bazaarvoice’s customer witnesses were not supported by any analysis they had performed, and their views were inconsistent with most of the economic and other objective evidence presented in the case.
Generally, it is the division that has overly relied on customer testimony to block a merger. Oracle and US v SunGard5 are just two of the many examples where courts expressed scepticism about the value of customer testimony opposing a merger. Bazaarvoice did not learn the lessons from the DOJ’s losses in these cases.
The role of hot (bad) documents
The parties’ internal business documents played a critical role in the outcome of the case. Virtually every argument Bazaarvoice made to rebut the division’s case was undercut by dozens of pre-merger documents drafted by high-level Bazaarvoice and PowerReviews executives. The documents established that the purpose of the merger was to eliminate a significant competitor, erect entry barriers and raise prices. Although intent is not an element of a Section 7 violation, the court noted that “the admissions made by Bazaarvoice and PowerReviews officers and employees prior to the merger and during their intense competition with each other undergirds the government’s case and conflicts with much of Bazaarvoice’s presentation at trial.”
The reason Bazaarvoice generated such a large volume of “hot” documents is that Bazaarvoice and PowerReviews engaged in three separate negotiations leading up to the deal. In addition, Bazaarvoice conducted its initial public offering in the middle of these negotiations where it generated additional documents analysing the market. When outlining the benefits of the transaction, Bazaarvoice executives noted that acquiring PowerReviews would eliminate Bazaarvoice’s “primary competitor”; create a “monopoly in the market”; “avoid margin erosion” caused by “tactical ‘knife-fighting’ over competitive deals”; and provide “relief from the price erosion that sales experiences in 30-40% of deals”. In its documents PowerReviews’ executives mirrored those beliefs, and said that Bazaarvoice would be willing to pay a premium because the acquisition would eliminate the “knife-fighting” between the two companies. Documents from the Bazaarvoice and PowerReviews sales teams also suggested intense, war-like, head-to-head competition. A member of the Bazaarvoice trial team described these documents as “antitrust porn”.
At trial Bazaarvoice emphasised an alternative rationale for the deal. Bazaarvoice executives, with some support from contemporaneous documents, testified that the rationale for the deal was to better compete in a market for Big Data. They argued that the combination of the Bazaarvoice and PowerReviews content, advertising and data would benefit consumers. With more data, Bazaarvoice would have the scale to allow online retailers to better personalise and target advertising. The court was unconvinced by this attempt to walk away from the anticompetitive rationale evidenced throughout the pre-merger documents. The court found that “[t]he evidence that Bazaarvoice and PowerReviews expected the transaction to have anticompetitive effects is overwhelming.”
While on the witness stand, Bazaarvoice’s executives struggled to establish their credibility. Bazaarvoice clearly understood it would be difficult to refute their own documents and, in their opening argument, lawyers for Bazaarvoice told the court that its executives would “own them”, rather than distance themselves from the seemingly incriminatory emails and memos. Bazaarvoice executives did not walk away from their documents; they ran. A Bazaarvoice executive was the first witness called to testify and he initially refused to admit that an email he sent to the company’s board outlining the anticompetitive motives and recommending the company go forward with the deal reflected his personal views. He was then impeached with other documents and forced to admit that he agreed with many of the points in his email.
Another executive used a demonstrative exhibit to present data on business lost to competition. But, when presented with the underlying document on which the exhibit was based, the witness had to admit that the data he presented grossly overstated the amount of revenue that was lost to competition. These patterns were repeated by other Bazaarvoice executives. In addition, the court found that Bazaarvoice’s economic expert submitted an exhibit that materially overstated the impact of in-house R&R solutions in the market. The court found the expert presented this evidence even though he was aware of an error in his analysis that was revealed on cross-examination. The court noted that his attempt to explain away his mistake “fell flat”.
While the court emphasised the parties’ internal documents, the documents were just one factor the court considered. The documents were several years old by the time of trial and the issue in any merger case is not the market structure at some time in the past, but the structure of the market on the day of trial and looking forward at least two years. The best strategy to deal with bad documents is to present evidence that the views expressed are inconsistent with current market realities, or to present evidence that the market has since changed so much that the views are no longer operative. The court gave the documents considerable weight because the evidence produced at trial was fully consistent with its executives’ anticompetitive predictions. The market on the day of trial was still dominated by Bazaarvoice: no fringe competitor had expanded in the 18 months since the deal was consummated; and high-entry barriers would prevent fringe competitors from expanding in the foreseeable future.
While much has been written about the Bazaarvoice opinion on liability, the remedy proceedings have attracted little attention or analysis. Fashioning a remedy 18 months after a merger has been consummated, and some of the assets have already been commingled, is not a simple proposition. The division didn’t know much about the current state of the PowerReviews assets. For example, the division did not know how many customers using PowerReviews at the time of the acquisition were still using the platform. It was also unknown whether Bazaarvoice had invested to improve the PowerReviews R&R platform and whether it was still commercially viable. The division was unable to get any information from Bazaarvoice about the status of the PowerReviews R&R platform, and also had to account for the possibility that Bazaarvoice could appeal and thereby delay implementation of any final judgment entered by the court.
Facing a deadline to submit a proposed final judgment, the division had to craft a remedy that accounted for the possibility that the PowerReviews assets may not be commercially viable at the time the final judgment’s remedy would be implemented. To address this concern, the division’s original, proposed final judgment required the full divestiture of the PowerReviews assets to a buyer acceptable to the division. In addition, the proposed final judgment required Bazaarvoice to offer the acquirer a perpetual, irrevocable licence for its own current R&R platform in the event that less than 80% of legacy PowerReviews customers remained on the PowerReviews R&R platform. While an extreme remedy, the potential licence of Bazaarvoice’s own R&R platform would only have been triggered if the PowerReviews customer base had diminished substantially at the time of a divestiture sale. This provision ensured a potential buyer would have access to a viable R&R platform and also provide Bazaarvoice with the incentive to maintain the PowerReviews R&R platform until a divestiture could be implemented.
After the division filed court papers in support of its proposed final judgment, Bazaarvoice agreed to waive its right of appeal, engage in settlement negotiations and provide evidence of the current status of the PowerReviews’ assets. The evidence convinced us that PowerReviews had retained most of its customers and remained a viable, stand-alone product. Bazaarvoice got to keep its R&R platform, but the settlement contained a number of novel provisions designed to ensure that a critical mass of customers would remain on the PowerReviews platform after divestiture, and to restore the level of competition that would have developed had the merger never occurred.
The settlement prohibited Bazaarvoice from poaching legacy PowerReviews customers; required Bazaarvoice to help keep PowerReviews customers on the platform; required Bazaarvoice to waive breach-of-contract claims against any of its customers who switched to the PowerReviews acquirer during a limited time period; and prohibited Bazaarvoice from soliciting PowerReviews’ customers for a six-month period. These requirements were designed to give the new competitor the opportunity to grow its market share to where PowerReviews would otherwise have been, absent the merger. In addition, the settlement required Bazaarvoice to waive trade secret restrictions related to its R&R technology for any of its employees hired by the new company, and to license its patents to the PowerReviews buyer. These provisions were designed to ensure that Bazaarvoice would not engage in strategic behaviour to raise its rival’s costs through litigation related to intellectual property that had been commingled through the transaction.
Finally, the court found that Bazaarvoice’s syndication network was a barrier to entry in the market for R&R platforms. R&R platforms allow manufacturers to share or “syndicate” ratings and reviews with their retail partners. The court also found that PowerReviews had the only other competitive syndication network at the time of the merger. To better enable the new company to replace the competition that PowerReviews would have provided, Bazaarvoice was required to provide the acquirer access to its syndication network for four years, ensuring customers of the PowerReviews buyer had access to the syndication relationships on the Bazaarvoice network.
The court approved the proposed settlement on 24 April 2014, and on 2 July 2014 Bazaarvoice divested the PowerReviews assets to Viewpoints. Customers appear to have benefited from lower prices and more choices among R&R platforms. In its first annual report following the divestiture, Bazaarvoice warned investors that “[t]he average sales price of our solutions may decline for a variety of reasons, including competitive pricing pressures.” Or, as Bazaarvoice might say in its internal documents, “tactical knife-fighting” has resumed. It remains to be seen whether PowerReviews will be able to offer an independent syndicate network when it loses access to Bazaarvoice’s network.
The Bazaarvoice victory was the division’s second litigated merger win since its 2004 defeat in Oracle. The division’s first litigated merger win since Oracle was the division’s successful challenge of H&R Block’s acquisition of TaxAct. Both cases were tried by the same team of attorneys, and these victories demonstrated that the division staff was “litigation-ready” and set the stage for the division’s high-profile courtroom victories that followed.
* Jim Tierney is an antitrust partner at Orrick, Herrington & Sutcliffe in Washington, DC. Before joining Orrick, he was the chief of the Antitrust Division’s Network & Technology Enforcement Section and had the privilege of supervising the team that investigated and litigated the Bazaarvoice case.
- US v Oracle Corp, 331 F. Supp. 2d 1098 (N.D. Cal. 2004).
- US v H&R Block Inc, 833 F. Supp.2d 36 (D.D.C. 2011).
- US v Syufy Enterprises, 903 F.2d 659 (9th Cir. 1990).
- US v Gen Dynamics Corp, 415 US486 (1974).
- US v SunGard Data Systems Inc, 172 F. Supp. 2d 172 (D.D.C. 2001).