Algorithmic Pricing UK EU US Case Law Review

Maskot via Getty Images

---

---

I. Introduction

In recent years, algorithmic pricing has been the subject of substantial antitrust scrutiny in the United States. Private litigation has generated a small but growing body of American case law, with the first appellate decision coming in August 2025. Algorithmic pricing is also on the radar of various competition authorities in the UK and the EU. While authorities across Europe are examining algorithmic pricing, private damages actions can also be expected to be brought in various European jurisdictions. Algorithmic pricing presents novel questions under UK and EU competition law, and cases in the US provide important lessons for both authorities and private litigants. This article will review recent US developments and their relevance to enforcement in Europe, identifying key takeaways for practitioners.

---

---

---

---

II. What Is “Algorithmic Pricing”?

Before diving into the case law, the term “algorithmic pricing” requires definition. An algorithm is merely a formal process or set of rules by which a problem is solved. A gas station owner who looks across the street and, as a rule, sets his prices $.01 lower than those of his competitor is, as a technical matter, engaged in algorithmic pricing. In the context of antitrust litigation, “algorithmic pricing” is typically shorthand for a situation in which a single company provides a pricing algorithm to multiple participants in an industry, and that algorithm’s recommendations are based at least in part on data provided by the industry participants. References to algorithmic pricing in this article refer to that situation.

Companies in a range of industries, including real estate, hospitality, and travel, use algorithms in their pricing. As a general matter, it is accepted that the use of an algorithm to price products in an industry can benefit consumers by allowing competitors to compete on price more efficiently and effectively. However, private plaintiffs and authorities challenging algorithmic pricing argue that, in some cases, the algorithm provider could facilitate price collusion by enabling (1) the exchange of proprietary, competitively sensitive information between market participants used in pricing; and/or (2) outright agreement on prices. As to the former, the theory is that multiple competitors contribute competitively sensitive information to the algorithm provider, which then uses all of that information to propose prices to each competitor. As to the latter, the theory is that multiple competitors agree (expressly or tacitly) to accept the algorithm’s pricing recommendations, which can then trend above the competitive level without being undercut in the normal course of price competition.

III. Lessons from United States Case Law

Algorithmic pricing litigation in the United States has turned on two key questions: (1) whether there was any agreement between industry participants, as opposed to between each participant and the algorithm provider; and (2) assuming there was an agreement, whether it was a per se restraint or subject to the rule of reason. Each court to face these questions has approached them differently. Almost all United States decisions have occurred at the motion to dismiss stage, meaning that courts have taken as true the allegations in the complaint.

A. Existence of a Horizontal Agreement

Section 1 of the Sherman Act proscribes agreements in restraint of trade, so the first step in any Section 1 case is to determine whether there is an agreement. In the context of algorithmic pricing, industry participants generally sign express agreements to license the algorithm. But the alleged harms from algorithmic pricing depend on horizontal competitors either exchanging competitively sensitive information or colluding on price through the algorithm provider; looking at each vertical agreement in and of itself would not capture that alleged harm. As such, the alleged agreement in algorithmic pricing claims is typically characterized as a “hub-and-spoke” agreement in which a horizontal agreement between competitors (the “spokes”) depends on each competitor executing a vertical agreement with a common “hub.” The industry participants are the spokes, and the algorithm provider is the hub. Proving the vertical agreements alone is not enough. United States law consistently holds that, to move forward on a hub-and-spoke theory of liability, a plaintiff must also prove the existence of a horizontal agreement among the spokes.

In algorithmic pricing cases that survived motions to dismiss, courts focused on allegations that purported to provide circumstantial evidence of agreement in the form of actions by competitors acting against their independent interests. In RealPage, the court reviewed a range of alleged circumstantial evidence, but focused on the allegation that the defendants gave RealPage (the algorithm provider) their proprietary commercial data knowing both that: (1) their competitors would do the same; and (2) each user would receive price recommendations based on all users’ data. The Court called this the “most persuasive” alleged evidence of a horizontal conspiracy because “the contribution of sensitive pricing and supply data for use by RealPage to recommend prices for competitor units is in Defendants’ economic self-interest if and only if Defendants know they are receiving in return the benefit of their competitors’ data in pricing their own units.”

In Duffy v. Yardi, another district court found that an agreement had been sufficiently alleged on similar logic, citing RealPage. That court rejected the defendants’ argument that a conspiracy was implausible absent allegations of an express agreement to accept the prices Yardi’s algorithm suggested, finding it reasonable to infer from the allegations that, “having turned over their commercially-sensitive data and paid for the services Yardi offered,” the defendants intended to use the pricing recommendations Yardi provided. However, a California state court dismissed similar claims against Yardi at summary judgment, holding that evidence had shown that Yardi’s pricing algorithm did not provide recommendations to users based on the confidential information of competitors.

Where algorithmic pricing complaints were dismissed, courts found the inverse: that the most plausible inference from the allegations was that competitors acted in their independent interests. In Gibson v. Cendyn Group, LLC, the Court found that Plaintiffs had not plausibly alleged a horizontal agreement, resting this finding on two key points: (1) the lack of allegations that industry participants exchanged confidential information through the algorithm provider; and (2) the lack of allegations that the prices proposed by the algorithm would be binding. The court reasoned that either allegation would make a horizontal agreement among competitors more plausible by showing actions that would not make sense absent agreement. Absent either, the more plausible inference was that the industry participants lawfully and independently decided to purchase helpful software without agreeing to any coordinated pricing strategy. On appeal, the plaintiffs dropped their hub-and-spoke claim and challenged only the individual license agreements, which the Ninth Circuit found did not restrain competition individually; the court refused to consider their aggregate effect absent allegations of horizontal agreement. In Cornish-Adebiyi v. Caesars and Dai v. SAS Institute, courts in California and New Jersey faced a similar alleged fact pattern and declined to find an agreement on similar logic, again focusing on the lack of allegations that the algorithm providers used proprietary information from any one defendant in calculating pricing for any other defendant.

The key to finding that circumstantial evidence of a horizontal agreement has been sufficiently alleged, then, is that defendants allegedly took actions against their interests, namely the knowing provision of competitively sensitive information to a third party that would use the information to recommend prices to competitors, and/or substantial (even if not total) deference to pricing decisions made by an algorithm. To be clear, more straightforward direct or circumstantial evidence of a conspiracy (e.g., communications between competitors, meeting records) could also serve as evidence of agreement, but was largely absent from all of the above cases.

B. Per Se or Rule of Reason

Agreements are unlawful per se if they are so obviously anticompetitive that further analysis of their effects is not required to render them unlawful (e.g., a price fixing conspiracy). The rule of reason is applied to restraints whose anticompetitive effects are not immediately obvious, and often requires an examination of market power, the alleged restraints’ anticompetitive effects, procompetitive benefits, and the availability of alternative means of obtaining any procompetitive benefits, though the actual level of analysis required will depend on the case.

As the only cases to find sufficient allegations of horizontal agreement, RealPage and Duffy v. Yardi are the only two cases to determine the appropriate standard of review, because all vertical restraints are analyzed under the rule of reason. While the RealPage court noted that hub-and-spoke price fixing conspiracies are subject to the per se rule, the court identified several “imperfections” in the instant allegations that meant the restraint at issue was not a “straightforward form of horizontal price fixing” and thus not subject to the per se standard. Specifically, the Court looked to the lack of alleged direct communications between defendants, that defendants purportedly diverged from pricing recommendations 10%-20% of the time, and that there was no enforcement mechanism against those who did not follow the algorithm’s pricing recommendation. Duffy v. Yardi, decided after RealPage on similar allegations, took a different approach. The court “respectfully disagree[d]” with the RealPage court, finding that the “imperfections” went to the strength of the conspiracy allegations, not to the nature of the restraint. Its analysis was simple: the alleged restraint was a horizontal agreement with respect to price, and thus subject to per se treatment regardless of whether the means for effecting that agreement were novel.

To some extent, these courts simply diverged. But a close review reveals an important difference in the way each court framed the admittedly similar allegations. The RealPage court framed the alleged conspiracy as an agreement between competitors “to each contribute their commercially sensitive pricing and supply data for use by RealPage to calculate their horizontal competitors’ pricing recommendations,” while the Duffy v. Yardi court framed it as an agreement “to entrust Yardi with their sensitive commercial information in order to obtain and implement the supracompetitive rental rates generated by Yardi’s algorithm.” The key analytical difference, then, is that the Duffy v. Yardi court found allegations of a horizontal agreement to implement supracompetitive prices by exchanging confidential information, while the RealPage court found only an alleged agreement to exchange confidential information.

IV. Algorithmic Collusion under UK and EU Competition Law

Algorithmic pricing enforcement and litigation will present a range of novel issues for UK and EU competition law. This section will focus on two critical questions where existing US case law may be instructive for the future assessment of algorithmic pricing under UK and EU competition law: (1) whether competition authorities and private claimants in the UK and EU are likely to seize the fact that UK and EU competition law applies to both agreements and concerted practices; and (2) whether algorithmic pricing could be considered a restriction by object or by effect.

A. Algorithmic Pricing as a Concerted Practice

Unlike the Sherman Act, the relevant provisions of UK and EU competition law, namely Section 2(1) of the Competition Act 1998 and Article 101(1) of the Treaty on the Functioning of the European Union, which are substantively similar, explicitly apply to both agreements and “concerted practices.” As the Court of Justice of the EU (the “CJEU”) reaffirmed in 2016 in Eturas, a seminal judgment providing guidance to national courts across the EU in relation to collusion using electronic means, a concerted practice has three elements: concertation, subsequent conduct on the market, and a relationship of cause and effect between concertation and conduct. The UK Court of Appeal reiterated these elements in its recent judgment in Phones 4U. The third element is subject to the Anic presumption, a rebuttable presumption that there is a relationship of cause and effect.

The Court of Appeal explained in Phones 4U that one-way disclosure of competitively sensitive information can amount to concertation if the recipient at least “accepts” it, even without requesting it, and noted that an “obvious example would be attendance at a meeting knowing that confidential information will be disclosed or exchanged.” Competition authorities and claimants may argue that knowing use of an algorithmic pricing system that pools competitors’ competitively sensitive information is analogous, and thus satisfies the concertation element. Further, authorities and claimants may argue that subsequent conduct and a cause-and-effect relationship are satisfied where there is proof (in line with the allegations in US cases) that the participating firms use the algorithm to make pricing decisions. It is worth noting that the two US courts that found an alleged exchange of proprietary, competitively sensitive information also found sufficient allegations of a horizontal agreement. Allegations of actions against interest were enough to infer an agreement at the motion to dismiss stage, but in the US, plaintiffs will have the burden of showing a dispute of material fact as to its existence at summary judgment and proving an agreement at trial. In the UK and EU, for the reasons described above, there is a risk that mere use of an algorithmic pricing system that pools competitively sensitive information could be found sufficient to establish a concerted practice, which is a lower legal standard than agreement.

B. By Object or By Effect

The “by object” and “by effect” classification in UK and EU competition law roughly mirrors the distinction between per se and rule of reason in the US. Restrictions “by object” are restrictions that by nature are so likely to have negative anticompetitive effects that no further analysis of effects is needed. In Budapest Bank, the CJEU explained that there must be reliable and robust experience with a restraint to determine that it is harmful to the function of competition. While the interchange fee agreement at issue there fixed certain fees, unique aspects of the card payment market made the court unwilling to declare the restriction “by object” based on the information it had available. However, courts are willing to classify novel restraints as “by object” where an examination of market conditions suggests that is appropriate.

Whether a given algorithmic pricing arrangement is assessed under UK or EU competition law as a “by object” or “by effect” restraint will depend on its particulars, but the distinction between the allegations and court reasoning in RealPage and Duffy v. Yardi may be instructive. If industry participants send data to an algorithm provider “in order to obtain and implement . . . supracompetitive prices,” then that restriction could potentially be treated as “by object.” By contrast, if industry participants agreed to contribute data for use by the algorithm provider “to calculate their horizontal competitors’ pricing recommendations,” the picture is less clear, especially if there were compelling reasons to think that the algorithm had other procompetitive effects, in which case the restriction might be treated as “by effect.” However, this classification will only become clearer over time once competition authorities and courts in Europe have had an opportunity to consider algorithmic pricing restrictions.

V. Takeaways and What to Watch in Europe

A. Algorithmic Conduct Giving Rise to Antitrust Concerns

While much remains to be decided, US case law provides at least three clear takeaways that may inform antitrust enforcement and litigation in relation to algorithmic collusion in Europe. First, the threshold to prove anticompetitive conduct under Section 2 of the Competition Act 1998 and Article 101 of the TFEU is lower than it is under Section 1 of the Sherman Act: authorities and claimants in Europe should in principle have an easier time proving a concerted practice relative to US plaintiffs’ burden to prove a horizontal agreement. Second, the assessment of liability will likely come down to the facts at issue in each specific case, just as it does in the US. Third, and most importantly, the success of some US plaintiffs in pressing algorithmic pricing claims portends significant antitrust enforcement and litigation risk for companies in the UK and EU using pricing algorithms. Companies and practitioners in these jurisdictions would be wise to familiarize themselves with the key factors US courts have looked to in assessing these claims, specifically:

• Whether proprietary, competitively sensitive information sent to the algorithm provider is used in calculating suggested prices for competitors;
• Whether there has been any agreement or communication, including public signaling, between industry participants as to which algorithms to use; and
• Whether the algorithm sets or enforces limits on the user’s ability to set pricing or otherwise compete in the market, or whether there is an agreement among competitors to use the prices set by the algorithm. Where these features exist, or potentially exist, companies in the UK and EU should review their practices, and assess the risk of facing an investigation and private damages litigation.

B. Risk of Enforcement Action by European Competition Authorities

In the UK, the Competition and Markets Authority (the “CMA”) is no longer only “watching and learning from developments in the US” as the authority announced an investigation into potential algorithmic collusion in early March 2026. The CMA’s investigation comes shortly after the authority identified deterrence of algorithmic collusion amongst its priorities for the current year. Interestingly, like the Gibson v. Cendyn litigation in the US, the CMA’s investigation also concerns the hotel sector with the CMA “investigating the suspected sharing of competitively sensitive information among competing hotel providers, via a hotel data services provider.” Consumer-facing businesses were also at the heart of the CMA’s decision in 2016 fining an online seller of posters for an agreement with a competitor not to undercut each other’s prices for products sold on Amazon Marketplace using an automated re-pricing software to execute that agreement.

At the EU level, the European Commission (the “EC”) revealed the existence of multiple ongoing investigations into algorithmic pricing in July 2025. Additionally, certain member states’ competition authorities have also been investigating potential algorithmic collusion at the national level, and other national authorities are also likely to be monitoring this space.

These developments show that companies should guard against antitrust risk in connection with algorithmic conduct alongside other current enforcement priorities (such as restrictive labor practices) in order to mitigate financial and reputational costs of a potential investigation. Anticompetitive horizontal restrictions can potentially result in competition authorities imposing fines of up to 10% of the worldwide annual turnover of the corporate group, with pricing restrictions typically attracting particularly severe financial penalties.

C. Risk of Damages Claims in the UK and the EU

Trends indicate that private litigation in the US and enforcement action by authorities in Europe tend to encourage antitrust damages claims in European jurisdictions. Accordingly, private antitrust claims in the UK and across EU member states arising from algorithmic conduct pose a related risk which can in certain cases surpass financial penalties imposed following a competition investigation.

In addition to claims traditionally brought by corporates, consumer collective litigation has also grown in significance in recent years in a number of European jurisdictions. In the UK in particular, the collective proceedings regime before the Competition Appeal Tribunal under Section 47B of the Competition Act 1998 – the UK equivalent of a US-style class action – has evolved rapidly since 2020, followed by various regimes across the EU, including the Mass Damages in Collective Action Act (WAMCA) in the Netherlands. The risk of collective proceedings might be higher in particular where algorithmic conduct relates to customer-facing products and services.

The coming months and years will show whether any damages claims arising from algorithmic conduct will follow on from any decisions of competition authorities, or whether any claimants will proceed to file standalone claims without waiting for the outcome of any related investigation. However, given that all of the private claims in the US outlined above have been pursued on a standalone basis, and given the general proliferation of standalone claims (at least in the UK), companies in relevant industries which use pricing algorithms should be alive to the potential risks of private litigation.

---

Endnotes

---

Authors

Samuel Sullivan

Sam Sullivan is an Antitrust Associate in the New York office of Arnold & Porter Kaye Scholer LLP. Sam represents clients across a range of antitrust and competition matters, including complex civil litigation, government...

View Bio →

Samuel Milucky

Arnold & Porter Kaye Scholer LLP

...

View Bio →

---

Authors

Samuel Sullivan

Samuel Milucky

Arnold & Porter Kaye Scholer LLP

---

Committees

This content was produced by:

• Global Private Litigation Committee

Related Content