Last week was the official conclusion to the Google Shopping Saga. The Court of Justice ruled that self-preferencing is indeed a stand-alone abuse by validating the theory of harm put forward by the Commission. With the ruling against Ireland in their long-running Apple state aid tax case, the week was a shining conclusion of Margrethe Vestager’s blemishing enforcement record against Big Tech, who was crowned the winner in a never-ending battle of wits between firm and regulator. Now that it’s over, we can finally look back at what happened in Google Shopping. A close look raises the question: who is the real winner in Google Shopping, and who is (or are) the loser(s)?
By Jasper van den Boom
What made Google Shopping special?
Most EU competition law enthusiasts will know the answer (or have their own answer) to the question what made the case of Google Shopping so special. Google Shopping is considered a landmark case for a number of reasons, the most important being that (a) it was the first big Art. 102 TFEU case against a big digital firm since Microsoft, (b) it highlighted how much competition in the digital sector had transformed in a relatively short amount of time; and (c) the Commission introduced a new type of abuse, aiming at the core function of the ecosystem business model: self-preferencing. There were other aspects in this as well that thrilled the community, the taking into account of behavioural effects, for instance, or the lack of references to classic IO economics known as “more economic approach”. Oh, and Google was also hit with a record fine of 2.42 bn. EUR (small money, probably, for a company with liquid means of over €100 billion, and with a quarterly revenue of around €80 billion, but quite a sum for someone working in academia). First and foremost, this was a symbol for the battle of Brussels versus Big Tech.
The European case against Microsoft related to the tying of its Windows Media Player service and access to its work-group servers. The Commission Decision in this case happened in 2003, and Microsoft lost its appeal in 2007. After 2007, the year in which Apple released the iPhone and forever changed the relationship between human and computer, competitive strategies of digital firms matured in terms of data collection and processing, the emergence of digital platforms, third-party participation in digital ecosystems, zero price strategies and more. Theorisation on the effect of all these changes began rapidly, also in the area of competition law and economics, yet it took a while to make sense of how exactly these business models fit into competition law and how abuse was to be understood.
The Commission became a frontrunner with their investigation into Google Shopping. Opening this investigation in 2010, the Commission spent 5 years establishing Google’s monopoly in horizontal search and how it leveraged its market power between that market and vertical search. The case did not get off the ground under Commissioner Joaquín Almunia. When Margrethe Vestager took over in 2014, digital became a priority. In 2015, the Commission sent its statement of objections to Google and issued its decision in 2017. The Commission found that by simultaneously promoting its own product – Google Shopping – while demoting competing comparison-shopping services in the organic results, Google had abused its dominance. Giving preferential treatment to one’s own product in a way that involved discriminatory practices did not fit neatly into the box of refusal to deal, nor did it simple constitute tying and bundling or discrimination. This triggered PhD thesis – amongst others by Friso Bostoen, Manuel Bogenreuther or Giulia Ferrari. The practice became known as its own thing: self-preferencing.
Finding the right box in the Article 102-universe was one thing. Finding the right remedies to stop the behaviour another. The Commission had ordered Google to cease their self-preferencing behaviour and anything that would produce an equivalent effect, but told Google to come up with how they would do so.
Shortly after the Commission Decision, Google implemented remedies to resolve the behaviour: it turned Google Shopping into a distinct business unit within Google and introduced an auction model where competing comparison shopping services could bid – in direct competition with the Google Shopping unit – for a preferential placement at the top of the results page. This remedy was subject to heavy criticism from academics and industry alike: it ensured equal treatment for all bidders, yet it was Google that could determine who could bid, there was little certainty that Google Shopping would actually compete on the merits in the auctions, and – perhaps most importantly – everyone now paid Google to be treated fairly instead of being discriminated against. Thomas Höppner, an anti-Google lawyer, produced a 390-pages study to show how ineffective this was.
As the remedies were being imposed, Google appealed against the Commission Decision before the General Court.
Google Shopping before the General Court
The General Court’s judgement came in November 2021. Google raised a number of pleas in its defence. Broadly, these can be summarised as: (i) the changes to Google Shopping were a quality improvement, not an abuse; (ii) the Commission should have proven indispensability if it was going to mandate access, in accordance with the Bronner case; (iii) their practices were not discriminatory; (iv) their practices did not produce anti-competitive effects, or these were not established; and (v) the behaviour was justified.
The General Court dismissed nearly all of Google’s pleas. On competition on the merits, the General Court found that the demotion of other results was an abnormality that showed that Google was not merely trying to compete better, but actually constituted an abuse due to the existence of the two prongs: preferential treatment on one side and demotion on the other. The Commission had also adhered to their evidentiary burdens, it did not need to show indispensability as this was a self-preferencing case and not a refusal to deal, and it did not need to establish a counterfactual as it had shown actual effects.
Perhaps, the most interesting part of the judgement came in the form of the Court’s afterthoughts on the duty of Google to treat their competitors fairly. The General Court had already established how Google’s behaviour had departed from competition on the merits with the two-pronged approach. Then, in a wave of semantic inspiration, found that Google was actually a ‘superdominant’ firm, and that their role as a provider of search was similar to the role of telecommunications providers in the days of yore (i.e., of Generation Y(ore) that still relied on land, while Generation Z only knows their smartphone). The Court decided that on the basis of this new concept of superdominance, extended the neutrality obligation from Directive 2002/22/EC on universal service, imposing on Google the duty of neutrality. This part of the judgement had far-reaching implications: who were these super dominant companies, and would they all face extended regulatory principles from sectoral regulation?
Google and the Commission duked out the rest of the appeals before the Court, and it was found that Google’s diversion of traffic and the resulting foreclosure effects were in fact anti-competitive, and that demoting did constitute discrimination.
The case before the Court of Justice
Google again appealed the judgement of the General Court, arguing that they had erred in applying the law. Advocate General Juliane Kokott sided with the Commission in a strong opinion, which focused mostly on the applicability of the indispensability criterion and was largely followed by the CJEU.
On 10 September 2024 the Court of Justice of the European Union rendered their judgement in appeal, in a much-anticipated ruling that marked the end of an era (the Vestager-era as it is henceforth known). A week before, the Court had set a blow to the Vestager record by annulling the Illumina/Grail approach to merger control – something of importance for the digital sphere since it related to killer acquisitions and Article 14 DMA. The Google ruling came like a warm shower now. While there were not quite celebrations in the streets, Akvavit champagne must have surely been uncorked by Commission officials, and the ruling made waves in popular and social media: the Commission had finally beaten Big Tech.
While the Court ruling was much anticipated, it was also surprisingly predictable and (dare I say) a little boring. It certainly did not send gasps around the antitrust community like its ruling in Illumina/Grail. The CJEU, sitting with the Grand Chamber and with Octavia Spineanu-Matei from Romania as the reporting judge, first affirmed that the Commission did not have to prove indispensability. However, by now, it has already been clarified in case law on constructive refusal to deal (such as Slovak Telekom) that indispensability does not apply in cases with discriminatory treatment. The CJEU also affirmed the Commission’s theory of harm and used the opportunity to restate the meaning of the concept of abuse: competing better is allowed, but using one’s strong position in one market to demote the position of others in another is not allowed. While it is great to have this clarified, it is not exactly surprising.
The CJEU also agreed with the General Court that the Commission could not be expected to demonstrate a counterfactual. First, it could not be expected from the Commission to do so looking at the complexity of the matter. More importantly, the Commission had already demonstrated that Google‘s behaviour produced anti-competitive effects. The case concluded with questions about the as-effective competitors test, but this was also dismissed as there was no question that there would be a competitors as effective as Google (thanks to Google). Finally, and perhaps why I personally read the judgement with a mix of understanding and disappointment, the CJEU brushed away the General Court’s concept of superdominance and the applicability of sectoral regulation as an afterthought, that was not necessary to come to its conclusions. With this, I think we have seen the short-lived existence of superdominance, which was cut short before we could even reach mega- or ultra-dominance. With the CJEU’s ruling the most important elements of the General Court’s ruling remained intact, but ultimately it seems the meal is served without the spice. While disappointing from a personal perspective, and perhaps an unhealthy overattachment to cool concepts like superdominance, it may be that the CJEU has done the General Court a favour in the long term. The concept of superdominance and its choice to expand regulatory principles from telecommunications to search were surprising, and perhaps too unpredictable or unconvincingly motivated. Perhaps the Court was right and just dominance suffices all the same.
The art of misdirection? How Google lost the case but won the market
While the Courts were mulling over whether self-preferencing was real, comparison shopping service providers, practitioners, and academics alike have submitted letters to the Commission in 2018, 2019, 2020, 2021 and 2022 that the remedies were not effective and that competition was not recovering. Ladenzeile went out of business, its platform is offline and it now exists to pursue damages from Google. Kelkoo and Foundem, two of the complainants in the Google case, have been driven to the brink of bankruptcy, as have many other comparison-shopping service providers. Twenga, one of the other complainants, is now a premium partner in Google Shopping’s ecosystem. It seems that the choice is either no business or Google’s business. Case won, market lost, damages as a last resort? In 2019 (!), it was reported that Idealo, a comparison site, was suing Google for over 500 million EUR in damages in the Regional Court of Berlin. The case is ongoing, now with some tailwind from Luxemburg.
There was another winner of this case: The DMA. Without the far-too-long Google Shopping saga and its failure to produce meaningful remedies, the DMA would never have come into existence. The hopefuls now do no longer look at antitrust for helping them in the markets, but at the DMA. Article 6(5) DMA is a direct translation of the Google Shopping theory of harm into regulation.
The compliance workshops have shown that Google – again – has turned faith to its own hand and has implemented its obligations under Art. 6(5) DMA to its own benefit. This behaviour is – again – one of the first types of non-compliance to be investigated, this time under the DMA. It is the litmus test to show if the DMA can deliver what competition law could not, namely effective remedies.
So, who has really won the Google Shopping case? The Commission was able to establish self-preferencing as a stand-alone abuse after a hard-fought battle. Google, after 14 years of litigating, fought off effective remedies and just suffers a a €2.4 billion euro fine (while it has turned a €20 billion profit in just the first quarter of 2024). While either can consider themselves the winner in their own way, there is one group that surely can’t: the complainants who notified the Commission of Google’s practices that started the investigation in 2010. Some lawyers considerably raised their profile (and possibly their salaries). Some Commission officials were able to raise their market value and secured jobs in Big Tech companies or their law firms. Executive Vice-President Vestager is vindicated – she remains the queen of digital antitrust. The courts proved that they are not behind and follow new approaches. Academia had a lot to chew on. For the general public, the case also did some good. And for our heroes, the complainants in the case? The CJEU found that they had to “bear their own costs” for the case. Would this be the nail in the coffin for those that could barely avoid bankruptcy up until now? Probably not, but it will surely not raise spirits after all thist time.
The ineffective antitrust remedies in Google Shopping may also serve as a warning light to the US instances that now debate the remedies in the Google cases. What is the message for them? Do not give Google too much room to decide on the details? Is structural separation the better option? I will be looking into these questions for an upcoming paper with Fiona Scott Morton (who has already voiced her opinion on the matter).
