Competition Law and Governance of Optimization-Driven Markets

1. Introduction

Optimization-driven markets are markets in which prices, rankings, recommendations, allocation of resources, advertising, logistics, credit decisions, search results, inventory, or other commercial outcomes are increasingly determined by algorithms designed to optimize particular objectives.

Examples include:

• dynamic pricing;
• algorithmic price matching;
• online marketplace ranking;
• search-result optimization;
• advertising auctions;
• ride-hailing surge pricing;
• hotel and airline pricing;
• recommendation systems;
• automated inventory allocation;
• credit and insurance scoring;
• cloud-resource allocation;
• energy-grid optimization;
• AI-assisted procurement.

Optimization itself is not inherently anticompetitive. Indeed, algorithms can reduce transaction costs, improve forecasting, increase output and enhance consumer choice. The competition-law problem arises where the objective function, data inputs, market feedback, or control architecture of an optimization system produces exclusion, coordination, discrimination, self-preferencing or excessive market power.

Modern competition law therefore increasingly asks not merely:

“What price did the undertaking charge?”

but also:

“What data, algorithm, optimization objective and competitive constraints produced that outcome?”

2. Meaning of Optimization-Driven Markets

An optimization-driven market can be represented as:

Market Data → Algorithm → Optimization Objective → Commercial Decision → Market Response → New Data → Re-optimization

For example:

Competitor prices + consumer demand + inventory + historical transactions → pricing algorithm → revenue maximization → new price → consumer response → new data → revised price

This creates a continuous competitive feedback loop.

Unlike traditional markets, where a manager may periodically determine prices, an optimization system can make thousands or millions of decisions automatically.

Core characteristics

1. Data dependency
2. Continuous automated decision-making
3. Real-time adaptation
4. Network effects
5. Feedback loops
6. High switching costs
7. Algorithmic personalization
8. Platform intermediation
9. Rapid competitive responses
10. Potential opacity

3. Why Optimization Creates Competition-Law Problems

A. Algorithmic coordination

Competitors may independently employ similar algorithms, yet their systems can react to one another's prices extremely quickly.

The difficult legal question is whether:

independent optimization → parallel behaviour

has become:

algorithmic coordination → unlawful concerted conduct.

The distinction is particularly important when algorithms use competitors' commercially sensitive information.

B. Algorithmic price fixing

An undertaking may use an optimization system that incorporates:

• competitors' prices;
• confidential pricing data;
• future pricing intentions;
• inventory information;
• discounts;
• demand forecasts.

Where competitors effectively delegate pricing decisions to a common algorithm, the algorithm may become a mechanism for coordination rather than genuine independent competition.

The U.S. Department of Justice's RealPage litigation illustrates this concern: the government alleged that competing landlords supplied competitively sensitive information to a common pricing system whose recommendations were used to align rental pricing.

4. Self-Preferencing Through Optimization

Optimization systems operated by dominant platforms can prioritize the platform's own services.

For example, a search engine could optimize rankings according to criteria that systematically place its own products above competing services.

The competition-law concern is not simply that an algorithm produces a ranking. It is whether the optimization criteria are structured or manipulated to disadvantage competitors.

Google Shopping

In Google Search (Shopping), the European Commission found that Google favoured its own comparison-shopping service in general search results while competing comparison-shopping services were treated less favourably.

The General Court examined the conduct as an abuse of dominance involving Google's preferential treatment of its own specialised search service. The Court of Justice subsequently upheld the essential infringement finding in 2024.

Competition-law principle

A dominant undertaking cannot necessarily hide an exclusionary strategy behind the statement:

“The algorithm selected the result.”

The legal inquiry can extend to how the algorithm was designed, what criteria it optimized and how those criteria affected competing services.

5. Six Major Case Laws

Case 1 — T-Mobile Netherlands BV v Raad van Bestuur van de Nederlandse Mededingingsautoriteit

CJEU, Case C-8/08

Facts

Mobile-network operators participated in discussions concerning dealer remuneration.

The issue was whether information exchange between competitors could constitute a restriction of competition even without an explicit agreement fixing prices.

Principle

The CJEU emphasized that exchanges of strategically sensitive information can reduce uncertainty concerning competitors' future market behaviour.

Relevance to optimization-driven markets

The principle becomes particularly significant when algorithms continuously process competitively sensitive information.

If competitors systematically feed sensitive information into a common optimization mechanism, the information-exchange problem can be considerably intensified.

Lesson

Algorithmic coordination does not require a traditional face-to-face cartel meeting.

Case 2 — Eturas UAB and Others v Lietuvos Respublikos konkurencijos taryba

CJEU, Case C-74/14

Facts

Eturas operated a common online booking platform used by travel agencies. A technical message and subsequent system modification imposed restrictions concerning online discounts.

Principle

The CJEU considered when participants in a common electronic platform could be attributed knowledge of an anticompetitive mechanism communicated through the platform.

The case demonstrates that an electronic system can constitute an important mechanism through which competitors receive and implement commercially significant restrictions.

Relevance

In an optimization-driven marketplace, competition authorities may examine:

• platform instructions;
• automated restrictions;
• default settings;
• system notifications;
• algorithmic parameters;
• users' knowledge of the mechanism.

Lesson

Digital architecture can be part of the mechanism through which concerted conduct occurs.

Case 3 — Google and Alphabet v European Commission (Google Shopping)

CJEU, Case C-48/22 P / T-612/17

Facts

Google's general search service displayed its own comparison-shopping results prominently while competing comparison-shopping services were subjected to different treatment.

Principle

The case demonstrates that a dominant platform's ranking mechanism may be examined under Article 102 TFEU where its operation advantages the platform's own service and disadvantages rivals.

The CJEU dismissed Google's appeal in 2024.

Relevance to optimization-driven markets

Search and recommendation systems are optimization engines.

Consequently, competition authorities may examine:

• ranking criteria;
• relevance parameters;
• weighting mechanisms;
• visibility;
• default placement;
• traffic allocation;
• treatment of first-party and third-party services.

Lesson

Algorithmic optimization cannot automatically immunize self-preferencing from competition law.

Case 4 — United States v. Apple Inc.

U.S. Department of Justice, 2024

Facts

The U.S. Department of Justice challenged Apple's conduct concerning the structure and operation of the iPhone ecosystem.

The case concerns, among other things, how Apple's ecosystem restrictions allegedly limit competition and reinforce its position.

Relevance to optimization-driven markets

Modern digital ecosystems optimize multiple dimensions simultaneously:

users + developers + applications + payments + devices + data + distribution.

Competition therefore cannot always be analysed through a single transaction or price.

Lesson

Competition analysis may need to examine the architecture of an ecosystem and the cumulative effects of interconnected optimization decisions.

Case 5 — FTC v Amazon.com, Inc.

U.S. District Court for the Western District of Washington

The FTC and several states alleged that Amazon used interconnected practices to maintain monopoly power, including conduct affecting sellers, pricing and competition on its marketplace.

Relevance to optimization

Amazon's marketplace involves continuous optimization of:

• product ranking;
• search visibility;
• advertising placement;
• seller participation;
• fulfilment;
• pricing;
• consumer recommendations.

The legal significance is that optimization systems can become instruments of platform governance.

A platform may determine which sellers become visible, which products receive prominence and how commercial opportunities are allocated.

Lesson

Where a dominant platform controls an optimization layer essential to market access, competition analysis may need to consider algorithmic control over visibility and commercial opportunity.

Case 6 — United States v RealPage, Inc.

U.S. Department of Justice, 2024 onward

This is one of the clearest contemporary examples of algorithmic optimization becoming an antitrust issue.

Facts alleged by DOJ

The DOJ alleged that competing landlords supplied non-public, competitively sensitive information to RealPage's pricing system. The algorithm then generated rental-price recommendations using information obtained from competing landlords.

The DOJ alleged that the system could facilitate alignment of rental prices and reduce independent price competition.

The litigation subsequently produced settlements and proposed remedies concerning algorithmic coordination and sensitive-information sharing.

Legal significance

The case illustrates a crucial distinction:

Legitimate optimization

“Use our historical data to estimate demand.”

versus

Potentially anticompetitive optimization

“Use competitors' confidential current information to determine prices for competing firms.”

Lesson

An algorithm is not a legally neutral intermediary when its design facilitates coordination among competitors.

6. Algorithmic Pricing and Competition

Optimization-driven pricing may take several forms:

1. Independent dynamic pricing

Each firm independently determines prices using its own information.

Generally, dynamic pricing itself is not unlawful.

2. Algorithmic price matching

A firm automatically matches competitors' prices.

This can potentially reduce incentives to compete aggressively.

3. Common algorithm

Several competitors use the same third-party algorithm.

Risk increases where the system uses confidential information from multiple competitors.

4. Autonomous pricing

AI systems continuously modify prices without human intervention.

This creates difficult questions concerning responsibility and attribution.

5. Predictive coordination

Algorithms predict competitors' responses and adjust prices accordingly.

The competition-law inquiry may focus on whether the system merely responds to observable market conditions or facilitates coordinated conduct.

7. Optimization and Information Exchange

Information is the fuel of optimization.

Competition concerns increase when algorithms process:

• current prices;
• future prices;
• production quantities;
• inventory;
• margins;
• customer-specific information;
• strategic plans;
• discounts;
• capacity;
• demand forecasts.

The fundamental distinction is:

8. Feedback Loops and Market Power

Optimization systems can create self-reinforcing feedback loops.

Example

Large platform

↓

More users

↓

More data

↓

Better algorithm

↓

Better recommendations

↓

More users

↓

More data

↓

Stronger algorithm

This can generate a data–scale–optimization feedback loop.

The competitive concern is not merely possession of data. It is the possibility that optimization produces increasingly powerful advantages that rivals cannot realistically replicate.

9. Algorithmic Self-Preferencing

A platform may optimize its marketplace according to criteria such as:

• conversion rate;
• advertising revenue;
• engagement;
• retention;
• profitability;
• fulfilment;
• platform fees.

If the platform simultaneously competes with businesses using the platform, it faces a structural conflict.

For example:

Platform → designs ranking algorithm → ranks products → owns competing product → controls visibility

This creates a potential vertical conflict between platform governance and downstream competition.

10. Optimization and Predatory Conduct

Optimization can also affect predatory-pricing analysis.

An algorithm can calculate:

• competitor weakness;
• customer elasticity;
• switching probability;
• expected losses;
• optimal price reductions;
• duration of promotional campaigns.

A dominant firm could theoretically optimize short-term losses against particular competitors while maintaining profitability elsewhere.

Competition authorities therefore may examine not simply the observed price, but:

1. objective of the pricing strategy;
2. cost structure;
3. duration;
4. geographic targeting;
5. customer targeting;
6. competitor targeting;
7. recoupment possibilities;
8. internal algorithmic parameters.

11. Optimization and Discriminatory Pricing

AI systems can produce highly individualized prices.

Possible inputs include:

• location;
• purchasing history;
• browsing behaviour;
• device;
• loyalty status;
• demand elasticity;
• transaction history.

This raises questions about personalized pricing.

Competition law may become relevant where personalization:

• exploits market power;
• excludes particular customer groups;
• forecloses rivals;
• reinforces dominance;
• facilitates discriminatory exclusion.

However, personalization is not automatically an antitrust violation.

12. Optimization in Digital Advertising

Advertising markets are heavily optimization-driven.

Algorithms determine:

• advertising placement;
• bid allocation;
• relevance;
• audience targeting;
• auction outcomes;
• pricing;
• advertiser visibility.

The competitive structure can become complicated where the same undertaking controls:

advertiser → ad exchange → auction → publisher → measurement → optimization data

This creates potential vertical and horizontal conflicts.

Recent U.S. enforcement concerning Amazon's advertising-auction practices illustrates how algorithmic auction design itself can become a competition-related enforcement issue. The FTC alleged in 2026 that Amazon used undisclosed pricing mechanisms in its advertising auctions.

13. Optimization and Essential Facilities

In some digital markets, the optimization layer may become economically indispensable.

Examples include:

• app-store ranking;
• search ranking;
• payment routing;
• cloud allocation;
• digital advertising;
• logistics allocation;
• online marketplace visibility.

A dominant undertaking controlling the optimization layer may therefore possess the ability to influence who receives market access and on what terms.

The competition-law question becomes:

Can the dominant operator legitimately optimize its own system, or is the optimization mechanism being used to exclude competitors?

14. Governance Framework

A competition-law governance framework for optimization-driven markets should contain at least eight components.

1. Algorithmic accountability

Undertakings should know:

• what their algorithms optimize;
• what data they use;
• who can modify them;
• what constraints exist.

2. Data governance

Particular scrutiny should apply to:

• competitor data;
• non-public information;
• real-time information;
• commercially sensitive information.

3. Independent decision-making

Competing firms should retain genuine independence in:

• pricing;
• output;
• discounts;
• inventory;
• commercial strategy.

4. Auditability

Important algorithms should be capable of retrospective examination.

5. Human oversight

High-risk commercial decisions should not necessarily operate without meaningful human supervision.

6. Non-discrimination

Dominant platforms should establish transparent criteria where algorithmic ranking determines access to markets.

7. Explainability

Competition authorities increasingly need to understand:

input → model → objective → output → competitive effect.

8. Remedy design

Possible remedies include:

• prohibiting particular data inputs;
• separating datasets;
• changing ranking criteria;
• prohibiting information exchange;
• independent algorithm audits;
• access obligations;
• monitoring;
• interoperability;
• structural remedies in exceptional circumstances.

15. Competition-Law Test for Optimization-Driven Conduct

A useful analytical framework is:

Step 1 — Identify the optimization system

What algorithm or automated system controls the commercial decision?

Step 2 — Identify the market

Determine:

• relevant product market;
• geographic market;
• competitors;
• customers;
• platform relationships.

Step 3 — Identify the optimization objective

Is the system optimizing:

• profit?
• price?
• market share?
• engagement?
• advertising revenue?
• ranking?
• exclusion?
• customer retention?

Step 4 — Examine data inputs

Ask:

Whose data is being used?

Step 5 — Examine competitors' access

Does the algorithm give the operator an advantage unavailable to rivals?

Step 6 — Examine feedback effects

Does greater scale produce better data and therefore better optimization?

Step 7 — Examine competitive effects

Possible effects include:

• higher prices;
• reduced output;
• reduced innovation;
• exclusion;
• reduced choice;
• increased switching costs;
• foreclosure;
• coordinated conduct.

Step 8 — Consider efficiencies

Optimization may generate legitimate efficiencies such as:

• lower costs;
• better matching;
• reduced waste;
• faster delivery;
• improved forecasting;
• increased output.

These must be distinguished from merely increasing a dominant firm's ability to exploit or exclude.

16. Important Legal Distinction: Algorithm ≠ Liability

A central principle is:

The existence of an algorithm does not itself establish an infringement.

Competition law should examine the conduct surrounding the algorithm.

An algorithm may be:

Pro-competitive

• improving logistics;
• reducing costs;
• forecasting demand;
• matching buyers and sellers;
• reducing search costs.

Potentially anticompetitive

• coordinating prices;
• excluding rivals;
• manipulating rankings;
• exploiting confidential competitor data;
• imposing discriminatory access conditions;
• facilitating monopolization.

Therefore:

Algorithm + legitimate objective + independent data → potentially pro-competitive

whereas:

Algorithm + sensitive competitor information + coordinated decisions → significant antitrust risk.

17. Optimization-Driven Markets and Dominance

Dominance can become particularly durable where optimization generates cumulative advantages.

Traditional market power

Market share → market power

Digital optimization market

Users → data → optimization → better service → more users → more data

This can create algorithmic entrenchment.

The competition authority may therefore need to examine dynamic factors rather than relying exclusively upon static market shares.

18. Remedies for Optimization-Related Competition Problems

Possible remedies include:

Behavioural remedies

• prohibit certain algorithmic inputs;
• prohibit competitor data sharing;
• require independent pricing;
• prohibit self-preferencing;
• require transparency.

Technical remedies

• algorithmic firewalls;
• data separation;
• access APIs;
• audit logs;
• independent monitoring.

Structural remedies

In exceptional cases:

• separation of business units;
• divestiture;
• separation between platform and downstream operations.

Procedural remedies

• algorithmic compliance programmes;
• periodic audits;
• competition-law training;
• board-level supervision;
• incident reporting.

19. Six-Case Comparative Summary

20. Emerging Competition-Law Issues

Optimization-driven markets are likely to generate disputes concerning:

1. AI pricing agents
2. Autonomous purchasing agents
3. AI-powered procurement
4. Algorithmic collusion
5. Generative-AI recommendation systems
6. Dynamic electricity pricing
7. Autonomous vehicle routing
8. Cloud-resource optimization
9. Digital advertising auctions
10. AI-powered financial trading
11. Personalized pricing
12. Algorithmic credit allocation
13. Marketplace ranking
14. Search-result optimization
15. AI-driven mergers and acquisitions
16. Automated inventory allocation
17. Algorithmic labour-market matching

The central regulatory challenge is that competitive decisions increasingly occur inside technical systems rather than through visibly human decisions.

21. Conclusion

Competition law in optimization-driven markets must move from a purely outcome-based analysis toward a more sophisticated examination of the architecture that produces competitive outcomes.

The critical chain is:

DATA → ALGORITHM → OBJECTIVE → OPTIMIZATION → COMMERCIAL DECISION → MARKET EFFECT

The major cases demonstrate that existing competition principles remain relevant even when market conduct is technologically sophisticated. T-Mobile Netherlands highlights the importance of strategic information exchange; Eturas demonstrates the significance of electronic coordination mechanisms; Google Shopping shows how ranking and preferential optimization can implicate dominance law; the Amazon litigation illustrates competition concerns surrounding platform architecture; and RealPage demonstrates how common algorithmic pricing systems can raise direct antitrust concerns.

Thus, the appropriate governance model is not to prohibit optimization. It is to ensure that optimization remains compatible with independent decision-making, contestable markets, fair access, transparent competitive parameters and effective competition.