Competition Law and Behavioural Influence, Intelligence and Competition Law

1. Introduction

Modern competition law increasingly examines not only prices, output and market shares, but also how firms use behavioural influence, data, algorithms and artificial intelligence to shape the decisions of consumers and competitors.

“Behavioural influence intelligence” may be understood as the use of consumer-behaviour data, predictive analytics, AI, machine learning, personalised recommendations, ranking systems and choice architecture to predict, influence or automate market behaviour.

This creates a competition-law problem because a firm with substantial market power may be able to influence what consumers see, what they choose, how easily they switch, which competitors receive visibility, and even how rival firms set prices.

Recent academic literature specifically identifies behavioural economics as increasingly relevant to EU competition enforcement involving Microsoft, Google, Amazon and Apple.

2. Meaning of Behavioural Influence Intelligence

Behavioural influence intelligence combines three elements:

A. Behavioural data

Platforms collect information concerning:

• searches;
• clicks;
• purchases;
• browsing history;
• location;
• transaction patterns;
• time spent on products;
• switching behaviour;
• responses to advertisements;
• consumer preferences.

B. Predictive intelligence

AI and algorithms analyse that information to predict:

• what a consumer is likely to purchase;
• whether a consumer will switch;
• what price a consumer may accept;
• which recommendation is most likely to generate a transaction;
• how competitors are likely to react.

C. Behavioural influence

The firm then designs the interface or market environment to influence the predicted behaviour through:

• defaults;
• rankings;
• recommendations;
• personalised prices;
• search-result placement;
• nudges;
• automatic renewals;
• switching barriers;
• tying;
• preferential visibility.

The competition-law significance arises when this intelligence is used by a dominant firm to exclude competitors, exploit customers, facilitate coordination or reinforce market power.

3. Why Behavioural Intelligence Matters to Competition Law

Traditional economic models often assume that consumers:

1. have complete information;
2. compare alternatives;
3. make rational choices;
4. switch when another product is better or cheaper.

Actual consumer behaviour may differ.

Consumers can exhibit:

• status-quo bias — preference for the existing/default option;
• salience bias — disproportionate attention to prominent information;
• inertia — reluctance to switch;
• limited attention;
• information overload;
• loss aversion;
• hyperbolic discounting;
• choice overload.

Consequently, a seemingly small design decision can have substantial competitive effects.

For example, placing a dominant platform's own product at the top of a search page may matter because users do not necessarily examine every alternative. The UK's CMA has specifically identified algorithmic ranking and self-preferencing as potential mechanisms through which algorithms can reduce competition.

4. Behavioural Influence and Market Power

Behavioural intelligence can reinforce market power through a feedback loop:

Large user base → more behavioural data → better predictions → greater personalisation → higher engagement → more users/data → stronger market position

This can create a data-driven competitive advantage.

The concern is particularly important in digital markets because network effects and switching costs can make behavioural advantages self-reinforcing.

The competition authority therefore may need to examine not simply:

“Does the consumer have another option?”

but:

“Will the consumer realistically notice, evaluate and use that alternative?”

That distinction is central to behaviourally informed competition analysis.

5. Major Competition-Law Concerns

A. Self-preferencing

A dominant platform may use behavioural intelligence to determine which products users are most likely to click and then preferentially display its own products.

Potential concerns include:

• foreclosure of rivals;
• reduced visibility;
• diversion of traffic;
• leveraging dominance from one market into another.

Example

A search engine could place its own comparison-shopping service prominently while competitors appear lower in organic results.

B. Personalised pricing

Algorithms can use individual or group-level behavioural information to determine prices.

This raises competition questions concerning:

• price discrimination;
• exploitation;
• transparency;
• exclusionary pricing;
• discriminatory treatment of consumers;
• coordination between competitors.

Personalisation itself is not necessarily anticompetitive. The legal issue depends on market power, purpose/effects and the applicable competition and consumer-protection rules.

The CMA has identified personalised prices and online choice architecture among the potential consumer harms associated with algorithms.

C. Algorithmic collusion

AI can monitor competitors' prices and rapidly adjust prices.

There are several possible models:

1. Explicit cartel + algorithm

Competitors agree on prices and use software to implement the agreement.

2. Hub-and-spoke coordination

Competitors use the same pricing intermediary or algorithm, which may facilitate coordination.

3. Tacit algorithmic coordination

Independent algorithms may learn that maintaining higher prices produces greater returns.

Economic experiments have demonstrated that reinforcement-learning algorithms can learn supracompetitive pricing strategies without direct communication.

Importantly, economic possibility is not the same as legal liability. Competition authorities still need to establish the legally relevant agreement, concerted practice or other prohibited conduct under the applicable jurisdiction.

6. Behavioural Intelligence and Choice Architecture

“Choice architecture” refers to the manner in which options are presented to consumers.

Examples include:

• pre-selected options;
• default subscriptions;
• prominent “Buy Now” buttons;
• difficult cancellation mechanisms;
• personalised recommendations;
• ranking algorithms;
• limited comparison interfaces;
• warnings and prompts;
• recommended products.

Competition law becomes particularly relevant where a dominant firm deliberately structures the environment in a manner that makes competing products difficult to discover or switching unnecessarily difficult.

Behavioural economics has therefore become particularly significant in digital competition law.

7. At Least 6 Important Case Laws

Case 1 — Google Search (Shopping) — European Commission, 2017

Issue: Self-preferencing and manipulation of search-result visibility.

The European Commission found that Google had abused its dominant position by favouring its own comparison-shopping service in general search results.

Google Shopping results were displayed prominently, whereas competing comparison-shopping services were subject to Google's general ranking mechanisms.

The behavioural significance is substantial: visibility affects consumer attention and traffic.

The General Court subsequently upheld the Commission's core findings in 2021.

Competition-law principle

A dominant digital platform's control over an important gateway can potentially be used to disadvantage competing services through ranking and presentation.

The case is frequently regarded as an important example of behavioural considerations entering digital dominance analysis.

Case 2 — Google Android — European Commission, 2018

Issue: Defaults, preinstallation, tying and consumer inertia.

The Commission examined Google's contractual arrangements concerning Android devices, including preinstallation of Google applications and services.

The behavioural relevance concerns status-quo bias.

Consumers may continue using applications that are already installed or configured rather than searching for alternatives.

The General Court's judgment also considered the practical reality of user behaviour rather than relying exclusively on an assumption that consumers will independently search for competing products.

Principle

Where a dominant undertaking controls defaults or preinstallation, competition analysis may need to consider whether consumers realistically switch away from the default.

The case therefore illustrates the relationship between:

defaults → inertia → reduced switching → competitor foreclosure.

Case 3 — Amazon Buy Box — European Commission, 2022

Issue: Algorithmic selection and preferential treatment.

The Commission investigated Amazon's Buy Box mechanism and concerns that Amazon's own offers, together with offers from sellers using Amazon's logistics services, could receive preferential treatment.

The behavioural concern was that the Buy Box is highly prominent and consumers frequently focus on the featured offer rather than investigating numerous competing sellers.

Consequently, an algorithmically selected position can have a substantial commercial effect.

Principle

An apparently neutral ranking or recommendation algorithm can become competitively significant where:

• the platform controls consumer access;
• the selected position is highly salient;
• consumers rarely investigate alternatives.

The Amazon case is consequently important for understanding algorithmic visibility as a source of market power.

Case 4 — Apple Music Streaming — European Commission, 2024

Issue: Anti-steering restrictions and consumer decision-making.

The Commission examined Apple's restrictions concerning the ability of music-streaming applications to inform users about alternative purchasing options outside Apple's ecosystem.

The behavioural dimension concerns the difference between theoretical consumer choice and actual consumer behaviour.

Even if an alternative technically exists, consumers may not actively search for it.

The analysis therefore considers:

• consumer attention;
• switching behaviour;
• transaction friction;
• immediate versus future costs;
• the practical visibility of alternatives.

Principle

Competition analysis can examine whether consumers actually make use of alternatives rather than merely whether those alternatives formally exist.

Case 5 — Microsoft — European Commission, 2004 / Microsoft II, 2007

Issue: Tying and leveraging dominance.

Microsoft's conduct involving Windows and other products became a landmark example of leveraging a dominant position into neighbouring markets.

The behavioural dimension arises because integration or default availability can influence consumer adoption.

Consumers may use the integrated product because it is:

• already installed;
• familiar;
• convenient;
• immediately available.

Thus, competitors may face disadvantages even where consumers could technically download alternative products.

Principle

Competition analysis of tying can require attention to consumer inertia, convenience and the practical consequences of integration, rather than simply theoretical availability of substitutes.

The broader EU literature identifies Microsoft as an early example of behavioural reasoning entering Article 102 analysis.

Case 6 — Trod Ltd / GB Eye Ltd — UK CMA, 2016

Issue: Algorithm-assisted price coordination.

Two online sellers of posters agreed not to undercut one another on Amazon Marketplace and used automated repricing software to implement the arrangement.

The important point is that the algorithm did not eliminate the underlying competition-law violation.

Instead, software became the mechanism for implementing the coordination.

Principle

Businesses cannot avoid cartel liability merely because a prohibited pricing arrangement is implemented automatically.

This is particularly important for AI-based pricing systems.

The CMA identifies the case as an example of competitors agreeing not to undercut one another and using pricing software to implement the arrangement.

Case 7 — Eturas — Court of Justice of the European Union, 2016

Issue: Algorithmic platform and concerted practice.

The case concerned a common online booking platform and a system message that effectively restricted the maximum level of discounts that participating travel agencies could offer.

The CJEU considered whether knowledge of the platform's restriction could support a finding of participation in a concerted practice.

Principle

Digital platforms can become mechanisms through which commercially sensitive conduct is coordinated.

The case demonstrates that competition law can apply where coordination is transmitted through a technological system rather than conventional face-to-face communication.

Case 8 — Intel — European Commission / Court of Justice

Issue: Loyalty incentives and actual competitive effects.

Intel concerned conditional rebates and the manner in which a dominant firm's incentive structure could influence customers' purchasing behaviour.

The later EU judicial analysis placed significant emphasis on the assessment of actual or potential exclusionary effects.

Behavioural relevance

Rebates can affect:

• customer loyalty;
• switching;
• purchasing decisions;
• competitors' ability to obtain sufficient scale.

Principle

Competition analysis should distinguish between an apparently attractive commercial incentive and its actual ability to influence customer behaviour and foreclose competitors.

8. Behavioural Influence and Artificial Intelligence

AI makes these problems more complex because algorithms can process behavioural information at enormous scale.

An AI system may predict:

• consumer willingness to pay;
• probability of switching;
• likelihood of accepting a recommendation;
• competitor reactions;
• demand elasticity;
• optimal timing for price changes.

This creates a potentially powerful behavioural intelligence infrastructure.

The competition-law risks can be divided into four categories:

The CMA has recently highlighted that increasingly sophisticated AI and pricing algorithms can create new forms of algorithmic collusion and other competition risks.

9. Behavioural Intelligence and Dominance

A useful analytical model is:

Stage 1 — Data accumulation

The firm collects large amounts of behavioural data.

↓

Stage 2 — Behavioural prediction

AI predicts individual or collective behaviour.

↓

Stage 3 — Choice architecture

The firm modifies:

• rankings;
• defaults;
• recommendations;
• prices;
• interface design.

↓

Stage 4 — Behavioural response

Consumers:

• click;
• purchase;
• remain;
• renew;
• fail to switch.

↓

Stage 5 — Competitive consequence

Rivals experience:

• reduced traffic;
• higher acquisition costs;
• reduced scale;
• reduced data access;
• lower visibility.

↓

Stage 6 — Reinforcement

Reduced rivalry generates additional market power and additional behavioural data.

This creates a behavioural-market-power feedback loop.

10. Relevant Competition-Law Doctrines

A. Abuse of Dominance

Behavioural intelligence may become relevant under abuse-of-dominance rules where a dominant undertaking uses its position to:

• foreclose competitors;
• impose discriminatory conditions;
• tie products;
• self-preference;
• restrict interoperability;
• create artificial switching barriers.

B. Restrictive Agreements

Where behavioural intelligence is shared or used collectively by competitors, issues may arise under rules governing:

• price fixing;
• information exchange;
• concerted practices;
• hub-and-spoke arrangements;
• algorithmic coordination.

C. Merger Control

A behavioural-intelligence merger may produce competitive concerns where the transaction combines:

large user base + large behavioural dataset + AI capability + distribution infrastructure.

Authorities may therefore consider whether the transaction:

• eliminates an emerging competitor;
• combines complementary datasets;
• strengthens network effects;
• increases entry barriers;
• creates an ecosystem advantage.

D. Consumer Protection and Competition Law

Some behavioural practices sit at the boundary between competition law and consumer law.

Examples:

• dark patterns;
• misleading rankings;
• hidden cancellation;
• forced continuity;
• personalised manipulation;
• deceptive defaults.

A practice may be problematic even where it does not independently satisfy the legal test for an antitrust violation.

11. Evidence in Behaviourally Informed Competition Cases

Authorities increasingly need evidence concerning actual behaviour, rather than merely theoretical possibilities.

Important evidence can include:

Quantitative evidence

• click-through rates;
• conversion rates;
• switching rates;
• traffic diversion;
• retention rates;
• price responses;
• search-result engagement;
• algorithmic experiments.

Qualitative evidence

• consumer surveys;
• internal business documents;
• product-design documents;
• algorithm specifications;
• employee communications;
• strategy presentations.

Experimental evidence

Authorities may use:

• A/B testing;
• eye-tracking;
• controlled experiments;
• field trials;
• behavioural surveys.

The literature concerning EU enforcement identifies statistics, surveys, online reviews, eye-tracking and large-scale experiments as possible evidence for demonstrating the competitive effects of behavioural biases.

12. Challenges for Competition Authorities

1. Causation

It is difficult to prove that consumer behaviour was caused by the firm's intervention rather than by ordinary product quality.

2. Intent versus effect

An algorithm may unintentionally produce exclusionary consequences.

3. Algorithmic opacity

Authorities may not understand precisely how a machine-learning system reached a particular output.

4. Dynamic markets

AI systems continuously learn and change.

5. Counterfactual difficulty

Authorities must ask:

What would consumer behaviour have looked like without the algorithmic intervention?

6. False positives

Not every successful recommendation system or personalised price is anticompetitive.

7. Innovation

Over-regulation may interfere with legitimate technological improvements.

Therefore, competition law must distinguish efficient behavioural personalisation from conduct that materially restricts competitive conditions.

13. Relationship Between Behavioural Economics and Traditional Competition Economics

Traditional competition analysis often focuses on:

• price;
• output;
• market shares;
• costs;
• barriers to entry;
• elasticity;
• efficiencies.

Behavioural analysis adds:

• actual consumer decision-making;
• attention;
• defaults;
• switching costs;
• cognitive biases;
• salience;
• inertia;
• information limitations.

The two approaches are complementary.

Traditional question:

Can consumers switch?

Behavioural question:

Do consumers actually notice and undertake the steps necessary to switch?

That difference can materially affect an assessment of competitive effects.

14. Competition-Law Compliance for Businesses Using Behavioural AI

Businesses should consider:

1. Documenting algorithm objectives
2. Testing for discriminatory outcomes
3. Monitoring competitor-information inputs
4. Avoiding automatic implementation of unlawful agreements
5. Auditing ranking and recommendation systems
6. Separating competitively sensitive information
7. Testing whether defaults unfairly disadvantage rivals
8. Maintaining human oversight
9. Recording algorithm changes
10. Conducting competition-law risk assessments before deploying common pricing systems

The use of AI itself is not prohibited. The legal concern arises from how the technology is used and its competitive consequences.

15. Key Legal Principles Emerging

The case law and regulatory developments support several important propositions:

Principle 1

Algorithmic conduct is subject to ordinary competition law.

Principle 2

Automation does not immunise an otherwise unlawful agreement.

Principle 3

Consumer choice should sometimes be assessed realistically rather than purely theoretically.

Principle 4

Defaults and ranking can have competitive significance because of consumer inertia and limited attention.

Principle 5

Data-driven behavioural advantages can reinforce existing market power.

Principle 6

Self-preferencing can be particularly significant where a dominant platform controls a critical gateway to consumers.

Principle 7

Common algorithms can create risks of coordination between competitors.

Principle 8

Behavioural evidence can supplement traditional economic evidence in establishing competitive effects.

16. Conclusion

Behavioural influence intelligence represents an important intersection between competition law, behavioural economics, data analytics and artificial intelligence.

The central issue is not whether businesses may understand consumer behaviour. Businesses routinely and legitimately use behavioural information to improve products, advertising and pricing.

The competition-law concern arises when behavioural intelligence becomes a mechanism for leveraging dominance, suppressing rivals, manipulating competitive visibility, creating switching barriers, facilitating coordination or reinforcing market power.

The Google Shopping, Google Android, Amazon Buy Box, Apple Music, Microsoft, Trod/GB Eye, Eturas and Intel matters demonstrate different aspects of this relationship.

The emerging approach can therefore be expressed as:

Market power + behavioural intelligence + strategic intervention + material competitive effect = heightened competition-law scrutiny.

At the same time, behavioural influence should not automatically be equated with illegality. Competition authorities must distinguish legitimate personalisation, efficient algorithms and innovation from conduct that produces legally relevant exclusionary or coordinative effects. Recent EU scholarship similarly notes that behavioural economics is increasingly informing Article 102 enforcement and the Digital Markets Act, while raising questions about evidence, context and legal certainty.