Did Google's 2017 'fix' Actually Fix Anything? - Swedish Court Awards €1.3 Billion in Record Competition Damages

On 1 July 2026, the Stockholm Patent and Market Court handed down judgment in PMT 1860-22, ordering Google to pay Klarna Technologies AB (formerly PriceRunner International AB) damages of roughly €1.3 billion before interest, and approximately €1.74 billion with accrued interest.

It is the largest competition damages award in Swedish legal history, yet, on the court's own characterisation, only about a quarter of the roughly €5.8 billion claimed by PriceRunner.

A Field of Frozen Ground

To appreciate the judgment's significance, the landscape in which it was delivered must be understood. Sweden has permitted private competition damages since the Competition Act of 1993. Three decades on, the results have been modest: Swedish tort law takes a restrictive approach to pure economic loss, and competition cases have been no exception.

The Euroclear case (Svea Court of Appeal, January 2011) remains the only stand-alone abuse of dominance case in which Swedish courts have ultimately awarded damages. Three claimants recovered just under SEK 1.9 million in total for a refusal to supply data, on a 'before and after' method and a 'reasonable amount' standard. The sums were negligible and the case produced no framework for larger claims.

The TeliaSonera litigation cast a longer shadow. After the competition authority and the Market Court established that Telia had abused its dominance through a margin squeeze in broadband infrastructure, two competitors brought substantial claims. In Tele2 v Telia (Svea Court of Appeal, December 2017), the Court of Appeal reversed a first-instance award of SEK 240 million and dismissed the SEK 708 million claim in full: notwithstanding the established infringement, Tele2 had failed to prove any damage. The parallel Yarps claim of SEK 369 million was dismissed on the ground that no abuse was established. Neither claimant obtained leave to appeal; other stand-alone actions have settled or failed on proof of infringement.

The message to prospective claimants has thus been clear: an established infringement, even one confirmed by a competition authority and upheld through appellate proceedings, does not translate into recoverable damages without precise proof of harm and causation.

The Competition Damages Act, in force since 27 December 2016 and implementing the EU Competition Damages Directive, was welcomed as lowering that threshold. But its most meaningful benefit, the rebuttable presumption of harm for cartels, could not apply in a case resting on abuse of dominance. PriceRunner had to prove injury, causation and quantum from scratch, across three jurisdictions and up to 16 years of alleged harm. In the end, the Act played a limited role, governing the substantive claim only for the 2017–2023 tail.

The Commission Decision and What Came After

Before turning to the substance, it is worth setting out the structure of PriceRunner’s claim.

The Commission's 2017 Google Shopping decision (AT.39740, 'the Decision') found that Google abused its dominance by positioning its own comparison shopping service (CSS) more prominently in general search results than competing CSSs, while ranking algorithms systematically demoted rivals but were never applied to Google's own service. The General Court upheld the Decision in all material respects (T-612/17), annulling it only as regards the national markets for general search, and the Court of Justice dismissed Google's appeal in full (C-48/22 P). The infringement on the comparison-shopping markets was definitively established.

PriceRunner's claim had three layers: a follow-on claim for the infringement period covered by the Decision; a stand-alone claim for the period from Google's modifications of 28 September 2017 until 31 December 2023, the date at which PriceRunner procedurally capped the alleged infringement; and a run-off claim for the period thereafter.

No Shortcut: Where Article 16 Ends, the Stand-Alone Assessment Begins

The follow-on layer was straightforward: under Article 16 of Regulation 1/2003, the Decision bound the court from January 2008 in the UK and November 2013 in Denmark and Sweden. Although the operative part records that the infringement "is continuing" as at adoption, the post-Decision period concerned a market situation never assessed by the Commission or the EU courts; the court treated itself as bound only up to 27 September 2017.

For 28 September 2017 to 31 December 2023, the burden of proof lay entirely on PriceRunner, and the Court worked through every element of a full Article 102 assessment from market definition to objective justification. The Decision's findings were treated as the analytical baseline; the operative question at each step was whether the evidence showed that the underlying circumstances had materially changed.

Within that framework, the decisive question on conduct was what happened from 28 September 2017, when Google introduced its so-called modifications.

Did the Modifications End the Infringement?

The Decision required Google to bring the infringement to an end within 90 days of notification. Google's answer, in place from 28 September 2017, was to convert the prominent product-results display into an auction in which competing CSSs could purchase placement. Google argued this ended the abuse; the court disagreed.

The analytical cornerstone was the court's characterisation of the auctioned display as still being, in substance, Google's own CSS: it met the Decision's definition, and users perceived and used it as one. Rivals that bought their way in acted as paid advertising intermediaries for merchants rather than as comparison services competing in their own right, and links identifying them attracted negligible engagement.

Three structural conditions of the abuse persisted unchanged. First, Google's own product results continued to appear in a prominent, visually distinct box at the top of the page. Second, the adjustment algorithms were not changed in any material respect in response to the Decision, as Google itself confirmed. Third, traffic data corroborated the finding: organic traffic to competing CSSs remained depressed after September 2017 while traffic through Google's display kept growing, supported by independent studies, Google having declined to produce its own traffic figures.

Nor did the Commission's silence help Google. Asked by the court under Article 15(1) of Regulation 1/2003, the Commission confirmed that it had never taken a position on compliance, had continued informing Google of desirable further changes, and had factored the forthcoming Digital Markets Act into its monitoring. Non-intervention reflected regulatory sequencing, not approval.

The compliance lesson: a dominant firm that modifies the conduct most visible to the regulator, while leaving intact the mechanisms generating the harm, cannot assume the infringement has ended.

Breaking the Evidentiary Barrier

In quantifying its claim, PriceRunner took as its starting point a counterfactual scenario consistent with the Commission's guidance; Google did not dispute that approach. The parties differed entirely, however, on the hypothetical course of events, a question that came to dominate the case. Ultimately, the Court declined to adopt either counterfactual, citing the duration of the infringement, the pace of technological change, and the complexity of platform competition. No reliable non-infringement scenario could be constructed, the more so since the Commission had acknowledged more than one lawful way to end the infringement.

Rather than dismissing the main claim (as the Svea Court of Appeal had done in TeliaSonera), the court decided to exercise its estimation power.

But before turning to estimation, two rulings independently reduced the award. First, part of the UK claim was time-barred: damage arising between 1 January 2008 and 10 January 2009 was statute-barred under English law. Second, the run-off claim failed entirely. The court accepted in principle that loss can continue after an abuse ends, but PriceRunner's case rested on a procedural assumption that the abuse ended on 31 December 2023; the court refused to rescue that claim through estimation precisely because the difficulty was self-inflicted by the choice of model, not a genuine impossibility of proof.

In regard of the estimation, the court found that PriceRunner had presented all the evidence that could reasonably be required and adopted PriceRunner's expert's methodology as a framework while replacing his assumptions. The resulting figures are far below the claim. But the methodology matters as much as the numbers: where full proof is genuinely impossible, the right response is estimation grounded in observable economic data, not dismissal. That is exactly the signal Swedish private enforcement has needed.

What Now?

Google has stated it disagrees with the judgment and is reviewing its legal options; an appeal to the Patent and Market Court of Appeal requires leave and is anticipated.

The cost order is itself a striking data point for would-be claimants: because PriceRunner succeeded on roughly a fifth of what it claimed, the Court treated it as partly losing and ordered it to pay a substantial portion of Google's costs, roughly €7.5 million, plus interest. One could say that estimation saved the claim but not the costs consequence of overclaiming.

But the analytical framework, and the signal it sends on estimation, will be difficult to confine to this case alone. After three decades of largely frozen ground, Swedish private competition enforcement has produced something that may finally defrost it.