Competition Law and Competition Implications of Cognitive Marketplaces

1. Introduction

A cognitive marketplace is a digital marketplace in which artificial intelligence (AI), machine learning, predictive analytics, recommendation systems, automated pricing, ranking algorithms, behavioural data, or other computational tools are used to connect buyers and sellers and influence transactions.

Examples include:

AI-powered e-commerce platforms;

online travel and hotel marketplaces;

ride-hailing and delivery platforms;

app stores;

digital advertising exchanges;

financial and insurance comparison platforms;

platforms using personalised recommendations;

marketplaces using algorithmic pricing;

AI systems that match suppliers and consumers automatically.

Cognitive marketplaces can improve competition by reducing search costs, increasing price transparency, improving matching and personalisation, and allowing smaller businesses to reach consumers.

At the same time, they create difficult competition-law questions because the same algorithm can facilitate both competition and anticompetitive conduct.

The principal competition concerns involve:

algorithmic price coordination;

self-preferencing;

discriminatory ranking;

exclusion of competitors;

data advantages;

tying and bundling;

excessive dependence on a dominant platform;

personalised pricing;

network effects;

acquisitions of potential competitors;

interoperability restrictions; and

reduced consumer choice through algorithmic recommendations.

2. Meaning of a Cognitive Marketplace

A traditional marketplace generally provides a physical or digital place where buyers and sellers interact.

A cognitive marketplace goes further. Its technological system may:

predict consumer preferences;

recommend products;

determine search rankings;

dynamically alter prices;

identify potential buyers;

determine which sellers receive visibility;

match consumers with suppliers;

personalise offers;

detect fraud;

allocate advertising space;

forecast demand;

optimise commissions; and

automatically respond to competitors.

Thus, the platform is not merely a passive intermediary.

It may become an important economic decision-maker in the market.

Basic structure

Consumer data → AI/algorithm → prediction → ranking/matching/pricing → transaction → additional data → improved algorithm

This produces a feedback loop.

The more transactions the platform receives, the more data it may collect. More data can improve its algorithm, which can attract additional users, producing still more data.

This can create substantial competitive advantages and entry barriers.

3. Competition-Law Framework

Cognitive marketplaces can potentially be examined under several branches of competition law.

A. Abuse of dominance

A dominant platform may abuse its position through:

discriminatory access;

self-preferencing;

exclusionary ranking;

refusal to provide interoperability;

tying;

predatory pricing;

loyalty mechanisms;

discriminatory commissions.

B. Anti-competitive agreements

AI systems can facilitate:

price fixing;

market allocation;

information exchange;

coordinated output restrictions;

resale-price maintenance.

C. Merger control

Large platforms may acquire:

emerging competitors;

AI startups;

data-rich businesses;

complementary services;

potential future competitors.

Such acquisitions can eliminate competitive threats before they become significant.

D. Consumer protection

Competition law increasingly intersects with:

transparency;

personalised pricing;

misleading rankings;

dark patterns;

manipulation of consumer choice.

4. Market Definition in Cognitive Marketplaces

Market definition becomes particularly difficult.

A platform may simultaneously provide several services.

For example:

consumers → marketplace → sellers → advertisers → payment providers

The platform may operate in several interconnected markets.

Competition authorities may therefore examine:

4.1 Consumer-side market

The relevant market may concern the service provided to consumers.

4.2 Seller-side market

Sellers may depend upon the platform for:

customers;

advertising;

logistics;

payment;

data analytics.

4.3 Advertising market

The platform may simultaneously operate as an advertising intermediary.

4.4 Data-related competitive advantage

Data itself may not always constitute a separate relevant market, but access to unique or commercially valuable data can become an important competitive parameter.

5. Network Effects

Cognitive marketplaces commonly exhibit network effects.

The value of a marketplace increases as more participants join it.

For example:

More buyers → more sellers → better selection → more buyers

and:

More transactions → more data → better AI → better recommendations → more transactions

This can create a self-reinforcing cycle.

A successful incumbent may therefore become difficult for competitors to challenge even when consumers nominally have the ability to switch platforms.

6. Data as a Competitive Advantage

Data is particularly important in cognitive marketplaces.

A large platform can possess information about:

consumer preferences;

search history;

purchasing behaviour;

prices;

conversion rates;

seller performance;

product demand;

geographic patterns;

consumer willingness to pay.

An AI system can use this information to improve predictions.

Potential competition concern

Suppose a marketplace operates as:

marketplace intermediary;

seller;

advertiser; and

data analytics provider.

It may potentially use information obtained from independent sellers to compete against those same sellers.

This creates a vertical and horizontal conflict.

7. Self-Preferencing

Self-preferencing occurs where a platform gives preferential treatment to its own products or services compared with competing products.

For example, an AI marketplace might:

place its own products first;

give its products higher recommendation scores;

provide competitors with inferior visibility;

favour its own logistics service;

favour its own payment service.

The important competition question is not merely whether the platform owns a competing product.

The issue is whether the platform's conduct distorts competitive conditions in a market in which competitors depend upon the platform.

8. Google Shopping – European Commission

Case

European Commission v Google (Google Shopping)

The European Commission found that Google had abused its dominant position by systematically giving prominent placement to its own comparison-shopping service while applying less favourable treatment to competing comparison-shopping services.

The case is highly relevant to cognitive marketplaces because ranking and visibility were central competitive parameters.

Competition principle

A dominant digital intermediary controlling an important gateway to consumers may create competition concerns if its ranking system systematically advantages its own competing service.

Importance

The case demonstrates that competition can be affected even where the platform does not expressly prohibit competitors.

A platform can potentially weaken rivals through:

ranking;

visibility;

traffic allocation;

recommendation;

search placement.

Relevance to cognitive marketplaces

An AI marketplace may make these decisions automatically.

Therefore, the fact that a discriminatory outcome is generated by an algorithm does not necessarily remove competition-law responsibility.

9. Amazon Marketplace – European Commission

Case

The European Commission investigated Amazon's use of non-public seller data obtained through its marketplace.

The Commission's concern was that Amazon could use information generated by independent sellers to compete with those sellers.

The investigation resulted in commitments concerning the use of seller data and the operation of Amazon's Buy Box.

Competition principle

A platform that simultaneously operates a marketplace and competes with marketplace participants creates a potential dual-role problem.

Cognitive-marketplace relevance

AI makes this issue more significant because non-public marketplace data can be used to:

predict demand;

identify successful products;

optimise prices;

determine inventory;

design competing products;

improve recommendations.

The competitive concern therefore extends beyond traditional data access to algorithmic exploitation of marketplace data.

10. Eturas v Lietuvos Respublikos konkurencijos taryba

Case

Eturas UAB and Others v Lietuvos Respublikos konkurencijos taryba, Case C-74/14

The Court of Justice of the European Union considered a situation involving an online travel-booking system in which a technical message concerning restrictions on discounts was communicated through the system.

The case is particularly significant for algorithmic marketplaces.

Principle

An electronic platform can potentially facilitate coordination between competitors.

Importantly, competition law does not become irrelevant merely because communication occurs through software.

Cognitive-marketplace significance

Imagine that competing sellers use the same AI pricing system.

The system could automatically:

monitor competitors;

receive pricing information;

adjust prices;

implement common restrictions.

The resulting conduct may raise questions about whether businesses have participated in coordinated behaviour.

Key lesson

Technology can be the mechanism through which an agreement or concerted practice is implemented.

An algorithm does not automatically transform coordinated conduct into lawful independent conduct.

11. United States v Topkins

Case

United States v Topkins

This matter concerned an online seller who participated in an agreement involving algorithmic pricing of posters.

The conduct involved the use of computer algorithms to implement coordinated pricing.

Competition significance

The case illustrates an important principle:

A computer program does not provide immunity from antitrust law.

If businesses agree to coordinate prices and use software to implement that agreement, the technological method does not eliminate the underlying antitrust concern.

Relevance

Cognitive marketplaces make algorithmic coordination increasingly sophisticated.

An AI system might:

observe competitors;

predict their reactions;

adjust prices;

optimise prices against competitors.

This creates difficult questions concerning the distinction between:

independent algorithmic optimisation

and

algorithmically facilitated coordination.

12. Apple Inc. v Pepper

Case

Apple Inc. v Pepper, 587 U.S. 273 (2019)

The United States Supreme Court considered whether consumers could pursue antitrust claims against Apple concerning Apple's App Store.

The Court allowed the plaintiffs to proceed at the pleading stage because consumers purchasing apps through Apple's platform could be considered direct purchasers from Apple for purposes of the antitrust standing issue.

Importance for cognitive marketplaces

The case illustrates the importance of determining:

who buys from whom;

who controls the transaction;

who bears the competitive harm;

whether the platform is an intermediary or seller;

how platform commissions affect market relationships.

Cognitive-marketplace relevance

Where AI marketplaces simultaneously:

intermediate transactions;

determine rankings;

impose commissions;

control payment;

control access;

the legal relationship between consumers, sellers and the platform becomes particularly important.

13. Ohio v American Express

Case

Ohio v American Express Co., 585 U.S. 529 (2018)

The U.S. Supreme Court examined the economics of a two-sided transaction platform.

Credit-card networks connect:

merchants; and

cardholders.

The Court emphasised the importance of considering both sides of the platform when analysing competitive effects.

Relevance to cognitive marketplaces

Many AI marketplaces are also two-sided or multi-sided platforms.

For example:

buyers ↔ platform ↔ sellers

or:

consumers ↔ platform ↔ advertisers ↔ sellers

Competition analysis may therefore need to consider effects on multiple participant groups.

A practice that benefits one side may potentially affect another side differently.

14. Samir Agarwal v Competition Commission of India

Case

Samir Agarwal v Competition Commission of India, (2021) 3 SCC 136

The case concerned allegations relating to pricing practices in the radio-taxi sector involving platforms such as Uber and Ola.

The Supreme Court considered issues concerning:

competition-law complaints;

standing/informant status;

market dynamics;

algorithmic pricing and platform economics.

Significance

Digital platforms can create new forms of pricing and coordination that are different from traditional cartel arrangements.

Cognitive-marketplace relevance

AI-based pricing systems may use:

demand;

supply;

location;

time;

competitor information;

consumer behaviour

to determine prices.

Competition authorities therefore have to examine whether pricing is genuinely independent or whether the platform's structure facilitates coordinated conduct.

15. Matrimony.com Ltd. v Google LLC

Case

Matrimony.com Ltd. v Google LLC

The Competition Commission of India examined Google's practices involving search preferences and specialised search services.

The case is relevant to digital-platform competition because it concerns the relationship between:

search visibility;

platform power;

competing services; and

preferential treatment.

Relevance

Cognitive marketplaces similarly depend upon ranking algorithms.

A platform may determine which businesses consumers see first.

Consequently, visibility itself becomes an important competitive asset.

16. Algorithmic Pricing and Collusion

One of the most difficult problems is algorithmic collusion.

Traditional cartel

Businesses communicate:

"We will charge ₹100."

Algorithmic coordination

Businesses may not communicate directly.

Instead:

Firm A uses algorithm A.

Firm B uses algorithm B.

Each algorithm observes market prices.

Each predicts the competitor's response.

Both algorithms repeatedly adjust prices.

Prices may converge.

The competition-law question becomes:

At what point does algorithmic interdependence become legally relevant coordination?

17. Tacit Coordination

Tacit coordination differs from an explicit cartel.

Competitors may independently understand that aggressive competition could trigger a price war.

AI can potentially make such coordination easier by:

increasing market transparency;

reducing reaction times;

monitoring competitors continuously;

predicting competitor responses;

rapidly adjusting prices.

However, parallel pricing alone does not automatically establish an unlawful agreement.

Authorities generally need evidence meeting the applicable legal standard for prohibited coordination.

18. Personalised Pricing

Cognitive marketplaces can use AI to estimate a consumer's willingness to pay.

For example:

Consumer A:

predicted willingness to pay = ₹1,000

Consumer B:

predicted willingness to pay = ₹1,500

The platform may potentially show different offers.

Personalised pricing can create competition-law questions concerning:

price discrimination;

exploitation;

transparency;

consumer lock-in;

discriminatory treatment.

Whether personalised pricing violates competition law depends upon the applicable jurisdiction and the market circumstances.

19. Recommendation Algorithms

Recommendation algorithms can influence consumer choice.

For example:

"Recommended for you"

may determine which products receive consumer attention.

A platform may therefore have substantial power even without formally preventing consumers from viewing alternatives.

Competition implications

Potential concerns include:

self-preferencing;

exclusion of smaller sellers;

discriminatory ranking;

manipulation of visibility;

reduced discoverability;

foreclosure of competing products.

20. Ranking as a Competitive Parameter

Traditional competition focused heavily on:

price;

output;

quality.

Digital marketplaces introduce another parameter:

visibility

A product ranked first may receive substantially more consumer attention than one ranked twentieth.

Therefore:

Ranking algorithm → consumer attention → sales → seller dependence

can become an important competitive mechanism.

21. Exclusivity and Lock-In

Cognitive marketplaces can create switching costs through:

stored consumer data;

loyalty programmes;

seller ratings;

transaction history;

subscriptions;

integrated payment;

logistics;

personalised recommendations.

A seller may therefore remain on a platform because leaving means losing:

customer reviews;

historical data;

visibility;

consumer relationships.

This may contribute to platform dependency.

22. Interoperability

Competition may also be affected by whether platforms permit interoperability.

For example:

data portability;

payment interoperability;

API access;

communication between platforms;

compatibility with rival services.

A dominant platform may potentially weaken rivals by restricting access to important technical interfaces.

23. Killer Acquisitions in Cognitive Markets

AI markets develop rapidly.

A dominant platform may acquire a small AI company before the company becomes a substantial competitor.

The acquired business may possess:

innovative technology;

valuable data;

specialist engineers;

a novel recommendation system;

an emerging customer base.

Traditional turnover-based merger thresholds may sometimes fail to capture the competitive significance of such transactions.

Therefore, merger authorities increasingly examine the future competitive potential of acquisitions.

24. Vertical Integration

A cognitive marketplace may operate at several levels simultaneously:

AI infrastructure

↓

Operating system

↓

Marketplace

↓

Payment

↓

Advertising

↓

Logistics

↓

Consumer service

Vertical integration can create efficiencies.

But it can also create opportunities for:

foreclosure;

tying;

discrimination;

self-preferencing;

leveraging dominance from one market into another.

25. Essential Data and Market Power

A difficult issue is whether certain datasets become so valuable that access to them is competitively important.

Relevant questions include:

Is the data unique?

Can competitors obtain comparable data?

Can synthetic or alternative data substitute for it?

Is the dataset continuously updated?

Does the platform have exclusive access?

Is access technically feasible?

Would compulsory access reduce innovation?

Competition authorities must balance access concerns against incentives to invest in data infrastructure.

26. Consumer Choice and Cognitive Manipulation

AI marketplaces do not merely respond to consumer preferences.

They may also shape consumer preferences.

A recommendation engine can determine:

what consumers see;

when they see it;

what alternatives are highlighted;

what products disappear from view.

This creates a distinction between:

consumer choice

and

algorithmically structured choice architecture.

Competition authorities may therefore need to consider whether platform design limits meaningful consumer choice.

27. Innovation Competition

Competition is not limited to current prices.

A cognitive marketplace may reduce competition by weakening:

innovation;

technological development;

alternative business models;

future entrants.

For example, a dominant platform might make it difficult for an innovative competitor to obtain:

data;

users;

developers;

distribution;

advertising;

interoperability.

The result could be reduced innovation competition, even where present prices remain low.

28. Efficiency Benefits

Cognitive marketplaces also produce significant pro-competitive effects.

AI can:

reduce search costs;

match consumers with appropriate products;

reduce transaction costs;

detect fraud;

optimise logistics;

reduce inventory waste;

improve price comparison;

increase market access for SMEs;

improve service quality.

Therefore, the mere use of AI should not itself be regarded as anti-competitive.

The legal analysis should focus on the actual conduct, market power, competitive effects and legitimate efficiencies.

29. Competition Risks by Category

30. Important Case-Law Principles

The major cases can be summarised as follows:

31. Regulatory Approach

Competition authorities should examine cognitive marketplaces through several questions.

Step 1: Identify the relevant market

Determine:

product/service market;

geographic market;

platform sides;

substitute services.

Step 2: Identify market power

Consider:

market share;

network effects;

data advantages;

switching costs;

entry barriers;

ecosystem control.

Step 3: Identify algorithmic conduct

Determine whether AI controls:

price;

ranking;

recommendations;

access;

advertising;

matching.

Step 4: Examine competitive effects

Ask whether the conduct:

excludes rivals;

raises barriers to entry;

reduces choice;

increases prices;

reduces quality;

reduces innovation;

facilitates coordination.

Step 5: Examine efficiencies

Consider:

better matching;

lower costs;

improved quality;

innovation;

reduced fraud;

logistics efficiencies.

32. Corporate Compliance for Cognitive Marketplaces

Businesses using AI marketplaces should implement:

1. Algorithm audits

Regularly examine:

pricing;

ranking;

recommendation;

seller treatment.

2. Competition-law controls

Employees should understand that:

"The algorithm did it"

is not necessarily a defence to unlawful conduct.

3. Data governance

Separate:

confidential seller data;

platform data;

consumer data;

competitively sensitive information.

4. Human oversight

Important competition decisions should have appropriate human review.

5. Documentation

Companies should preserve:

algorithmic design documents;

training objectives;

pricing policies;

changes to ranking algorithms;

compliance reviews.

33. Challenges for Competition Authorities

Cognitive marketplaces create several enforcement difficulties.

Black-box algorithms

Authorities may not understand how an algorithm reaches its decision.

Rapid technological change

AI systems can change rapidly through machine learning.

Attribution

It may be difficult to determine:

Who made the anticompetitive decision—the company, developer, algorithm or user?

Legally, however, responsibility generally remains connected to the relevant undertaking and its conduct rather than treating the algorithm as an independent legal actor.

Cross-border operations

One marketplace can simultaneously operate across dozens of jurisdictions.

Evidence

Competition authorities may need access to:

source code;

training data;

logs;

communications;

internal documents;

model outputs.

34. Future Competition-Law Issues

Future cognitive marketplaces may raise questions concerning:

A. Autonomous pricing agents

AI agents may negotiate prices without continuous human intervention.

B. AI-to-AI negotiation

Consumer and seller agents may independently negotiate transactions.

C. Autonomous procurement

AI may automatically select suppliers.

D. Algorithmic collective behaviour

Multiple AI systems may adapt to one another.

E. AI-generated market structures

Algorithms could potentially determine:

who receives access;

which suppliers survive;

which products receive visibility.

F. Agentic commerce

Consumers may delegate purchasing decisions to AI agents.

This could substantially change the traditional concept of consumer choice.

35. Conclusion

Cognitive marketplaces represent an evolution from digital marketplaces to intelligent marketplaces.

Their competition significance arises from the fact that AI can control or influence:

pricing;

ranking;

recommendations;

matching;

access;

advertising;

data utilisation.

The central competition-law challenge is therefore not simply whether AI is being used. The important question is how AI changes competitive conditions.

The case law concerning Google Shopping, Amazon's marketplace practices, Eturas, Topkins, Apple v Pepper, Ohio v American Express, Samir Agarwal and Matrimony.com demonstrates several important principles: digital platforms can possess significant intermediary power; electronic systems can facilitate coordination; ranking can affect competition; two-sided markets require careful analysis; and platform data can become an important competitive resource.

For examination purposes, the central proposition can be remembered as:

“In cognitive marketplaces, algorithms can simultaneously improve competition through efficiency and create competition risks through pricing coordination, preferential ranking, data advantages, exclusion, and increased platform dependence.”

Short Revision Points

Cognitive marketplace = AI-driven digital marketplace.

AI can control pricing, ranking, matching and recommendations.

Network effects can strengthen platform market power.

Data can create competitive advantages and entry barriers.

Algorithmic pricing may create coordination risks.

Algorithmic ranking can create self-preferencing concerns.

Marketplace data may create conflicts where the platform also competes with sellers.

Two-sided market analysis is particularly important.

AI acquisitions can raise merger-control concerns.

Competition law must balance innovation and efficiency against exclusion and coordination risks.

Eturas is important for electronic facilitation of coordination.

Google Shopping is important for ranking and preferential treatment.

Amazon is important for marketplace data and platform conflicts.

Topkins demonstrates the relevance of algorithms to price coordination.

Ohio v American Express illustrates two-sided-platform analysis.

Apple v Pepper illustrates the legal importance of platform-controlled transactions.