Competition Law and Algorithmic Consumption Coordination Risks

Introduction

Algorithmic consumption coordination refers to situations where digital platforms, retailers, service providers, or other businesses use algorithms and large amounts of consumer or competitor data to coordinate commercially important decisions. The algorithms may determine prices, discounts, product recommendations, advertising, supply levels, or other terms offered to consumers.

Using an algorithm is not itself unlawful. Businesses routinely use algorithms to forecast demand, manage inventories, personalize services, and improve efficiency. Competition-law concerns arise when algorithms facilitate an agreement or concerted practice between competitors, exchange competitively sensitive information, stabilize coordinated conduct, or help a dominant firm restrict competition.

The central competition-law principle remains independent commercial decision-making. A business should normally determine its own competitive strategy rather than substitute coordination with competitors for the risks of competition.

1. What Is Algorithmic Consumption Coordination?

An algorithm can process information about consumer demand, previous purchases, browsing behaviour, competitors' prices, inventories, and market conditions. It can then recommend or automatically implement commercial decisions.

Competition problems can arise where several competing businesses:

• use the same algorithm or intermediary;
• contribute confidential pricing or demand information to a shared system;
• receive recommendations derived from competitors' non-public information;
• automatically follow coordinated recommendations;
• use software to monitor whether competitors are following a common strategy; or
• design algorithms specifically to implement an existing anticompetitive agreement.

The legal issue is therefore generally not whether a computer made the decision. The question is whether the underlying arrangement amounts to an unlawful agreement, concerted practice, information exchange, or exclusionary strategy.

2. Algorithmic Price Coordination

Pricing is one of the clearest areas of risk.

Suppose competing online sellers independently use ordinary pricing software. Seller A lowers its price, Seller B's software notices the public price and reacts, and Seller C responds as well. Parallel algorithmic reactions do not automatically establish an unlawful agreement.

The situation is different where competitors agree that their algorithms will maintain a particular relationship between their prices or otherwise implement a common pricing strategy.

The U.S. prosecution involving online poster seller David Topkins illustrates this distinction. According to the Department of Justice, Topkins and other sellers agreed to coordinate prices for posters sold through an online marketplace and used pricing algorithms and computer code to implement the agreement. The case resulted in a guilty plea and is an important early example demonstrating that traditional price-fixing rules can apply when software implements the coordination.

3. Hub-and-Spoke Algorithmic Coordination

A more difficult problem occurs when competitors do not communicate directly but use the same technological intermediary.

Imagine:

Retailer A → Algorithm Provider ← Retailer B

and

Retailer C → Algorithm Provider ← Retailer D

Each retailer gives commercially sensitive information to the central provider. The provider combines that information and generates recommendations for all participating firms.

This can resemble a hub-and-spoke arrangement. The algorithm provider operates as the hub while competing businesses constitute the spokes.

A major contemporary example is the U.S. government's RealPage litigation. The Justice Department alleged that competing landlords supplied RealPage with non-public information concerning rents and other leasing conditions and that RealPage's algorithm used this information to generate pricing recommendations. The government alleged violations of Sections 1 and 2 of the Sherman Act. These are allegations in ongoing litigation rather than a general judicial rule that shared pricing software is automatically unlawful.

4. Exchange of Competitively Sensitive Information

Algorithms require data. Consequently, competition authorities may examine not merely the algorithm but also the information entering it.

Particularly sensitive information can include:

future prices, planned discounts, expected production, capacity, inventories, projected vacancies, customer-specific information, future commercial strategy, and non-public demand forecasts.

Sharing sufficiently detailed and current information can reduce uncertainty about competitors' future behaviour.

The RealPage complaint, for example, alleges that competing landlords provided non-public commercially sensitive information that was incorporated into algorithmic recommendations. The Justice Department argues that the arrangement reduces independent competitive decision-making.

The competition concern therefore exists even though businesses may never communicate directly with each other.

5. Algorithms as Monitoring Devices

Successful coordination can be difficult because participants may have incentives to deviate—for example, by secretly lowering prices to attract additional customers.

Algorithms can potentially make deviation easier to detect.

Modern software can continuously observe:

• competitors' publicly observable prices;
• discounts;
• product availability;
• inventory movements; and
• changes in market conditions.

Rapid monitoring and response can make coordinated market behaviour more stable.

However, monitoring public prices is not automatically unlawful. Competition law generally requires examination of the surrounding arrangement and evidence of an agreement or other legally relevant anticompetitive conduct.

6. Consumer Demand and Consumption Data

Consumption algorithms may also analyse consumer behaviour.

Businesses increasingly know when consumers purchase products, how frequently they purchase them, what alternatives they examine, and how sensitive they are to price changes.

This information can produce legitimate efficiencies. For example, demand forecasting can reduce shortages and unnecessary inventory.

Competition concerns become stronger when competitors pool non-public consumption data in ways that reveal their commercial strategies or reduce uncertainty about how competitors intend to respond to demand.

Thus, the important distinction is between better understanding consumers and using shared information to coordinate competitors.

7. Personalized Pricing

Algorithms can potentially calculate different prices or offers for different consumers.

Personalized pricing itself is not necessarily a competition-law violation. Businesses have long offered different discounts to different customers.

Competition concerns can nevertheless arise where personalization forms part of a broader exclusionary or coordinated arrangement—for example, where competitors collectively use a common mechanism that reduces independent pricing.

Separate consumer-protection, privacy, data-protection, or discrimination rules may also apply even when competition law does not prohibit the practice.

Important Case Laws and Enforcement Proceedings

1. United States v. David Topkins — United States, 2015

This is one of the most important early algorithmic price-fixing cases.

Topkins and competing online sellers were accused of agreeing to fix prices for posters sold online. According to the DOJ, pricing algorithms were programmed to implement the agreed pricing relationship.

Topkins agreed to plead guilty.

Principle: An illegal agreement does not become lawful simply because computer software implements it. Algorithms can constitute the technological mechanism through which traditional price fixing is carried out.

2. Eturas UAB and Others v Lithuanian Competition Council — CJEU, Case C-74/14

Travel agencies used a common online booking platform. A system message concerned a restriction on the discounts available through the booking system, and the technical system implemented that restriction.

The Court of Justice examined when travel agencies using the common electronic system could be regarded as participating in a concerted practice.

The Court emphasized that participation could not simply be presumed merely because a business used the platform. Knowledge and the relevant evidentiary circumstances remained important.

Principle: Coordination implemented through a shared digital platform can potentially fall within traditional rules governing concerted practices, while proof of participation remains necessary.

3. United States and Plaintiff States v. RealPage, Inc. — United States

The Justice Department and several states brought an antitrust action against RealPage concerning revenue-management software used by landlords.

The government alleges that competing landlords supplied commercially sensitive non-public information and received algorithm-generated pricing recommendations derived from the collected data. It alleges violations of Sections 1 and 2 of the Sherman Act.

Principle: Shared algorithmic intermediaries can create competition concerns where competitors contribute sensitive information and use recommendations generated from pooled competitor data.

Because this matter involves allegations and continuing proceedings, it should not be described as a final judicial determination that every shared revenue-management system violates competition law.

4. Gibson v. Cendyn Group LLC — United States

This litigation concerns allegations involving hotel revenue-management software.

The dispute is significant because it addresses when competitors' use of common algorithmic pricing technology can satisfy the Sherman Act's requirement of concerted action.

The U.S. Department of Justice filed an amicus brief explaining that pricing algorithms are simply a newer technological means through which competition issues can arise and arguing that common algorithms can raise antitrust concerns depending upon the underlying arrangement.

Principle: Courts must distinguish genuinely independent adoption of technology from circumstances supporting an agreement among competitors.

5. FTC v. Amazon — U.S. E-Commerce Antitrust Litigation

In its broader monopolization litigation against Amazon, the FTC alleged that Amazon used an internal algorithm known as Project Nessie.

According to the FTC's complaint, the algorithm identified products for which Amazon predicted other online stores would follow Amazon's price increases. When competitors followed, Amazon allegedly maintained the higher price.

These remain allegations made by the FTC and should not be treated as established judicial findings.

Principle: Algorithmic pricing can be examined not only under collusion theories but also within broader monopolization and exclusionary-conduct analysis where a firm allegedly possesses substantial market power.

6. RealPage Landlord Proceedings and Related Enforcement

The RealPage litigation has developed beyond the original action against the software provider. The Justice Department's case materials record proceedings involving participating property-management companies and proposed or final judgments concerning particular defendants.

These developments are relevant because algorithmic coordination analysis does not necessarily stop with the technology provider. Authorities can examine the conduct of businesses supplying information to, adopting recommendations from, or otherwise participating in the system.

Principle: Liability analysis can encompass both the central algorithm provider and participating competitors, depending on evidence concerning agreement, knowledge, information exchange, and implementation.

7. Traditional Hub-and-Spoke Cases as an Analogy

Older hub-and-spoke cases remain important even though they did not involve artificial intelligence.

A traditional arrangement might involve one distributor communicating separately with multiple competing manufacturers and facilitating a common strategy between them.

Algorithmic systems can potentially reproduce the same economic structure digitally:

Competitors → common platform → recommendations → competitors.

The technology therefore changes the mechanism, but not necessarily the underlying competition-law principles.

8. Tacit Algorithmic Coordination

One of the hardest legal questions concerns algorithms that independently learn that maintaining higher prices is more profitable than aggressive competition.

Suppose two firms deploy separate AI systems without communicating. Through repeated market interaction, both algorithms eventually learn not to start price wars.

Economically, this could resemble coordinated behaviour.

Legally, however, similarity of behaviour does not automatically establish an agreement. Traditional competition rules generally distinguish prohibited concerted conduct from mere conscious parallelism or independent adaptation to market conditions.

This creates an important enforcement challenge: increasingly autonomous algorithms may produce coordinated-looking outcomes even where proving communication or agreement is difficult.

9. Factors Competition Authorities May Examine

Investigators can consider several factors when determining whether algorithmic coordination presents competition concerns:

Common provider: Do several competitors use the same algorithmic service?

Nature of data: Does the system receive public information or confidential competitor information?

Data freshness: Historical aggregated information usually raises different issues from detailed real-time or forward-looking data.

Recommendations: Does the system merely provide neutral analytics, or does it recommend commercially significant decisions?

Automation: Are recommendations automatically implemented?

Compliance mechanisms: Does the provider monitor whether customers follow its recommendations?

Communication: Were competitors informed that their rivals were participating?

Market structure: Is the market concentrated and transparent?

Purpose and design: Was the system designed to improve independent decision-making or to reduce competitive uncertainty?

No single factor necessarily determines legality. Authorities generally examine the arrangement as a whole.

10. Potential Consumer Harm

Algorithmic coordination can potentially reduce normal competitive pressure.

Possible effects include higher prices, fewer discounts, reduced output, reduced product variety, weaker incentives to innovate, standardized commercial terms, and diminished consumer choice.

Algorithms can amplify these effects because they operate quickly and can process enormous quantities of information.

At the same time, algorithms can produce substantial benefits. They can improve inventory management, reduce operating costs, predict demand, identify shortages, optimize distribution, and enable faster price reductions when market conditions change.

Competition analysis therefore requires distinguishing efficiency-enhancing automation from automation that facilitates anticompetitive coordination.

11. Compliance Measures

Businesses using algorithmic systems should preserve independent commercial decision-making.

Important safeguards include controlling access to competitors' confidential information, examining the source of data used by third-party algorithms, maintaining meaningful human oversight, reviewing automated pricing recommendations, documenting legitimate business reasons for important pricing decisions, and conducting competition-law assessments before competitors participate in shared data systems.

Companies should also be particularly cautious where a vendor requests current or future pricing, output, capacity, discount, inventory, or customer information from multiple competitors.

Conclusion

Competition law generally does not require entirely new legal principles merely because businesses use artificial intelligence or algorithms. Traditional concepts such as price fixing, concerted practices, information exchange, hub-and-spoke arrangements, monopolization, and independent commercial decision-making remain central.

Cases and proceedings such as Topkins, Eturas, RealPage, Gibson v. Cendyn, and the FTC's Amazon litigation demonstrate different ways in which established competition principles can interact with algorithmic markets.

The decisive issue is usually not whether an algorithm exists. It is what information the algorithm receives, how competitors participate, whether commercially sensitive decisions remain genuinely independent, and what competitive effects the arrangement produces.