Algorithmic Unilateral Effects in Merger Control for Digital Markets

1. Introduction

Algorithmic unilateral effects arise where a merger enables the combined digital firm to exercise market power without requiring coordination with competitors, because algorithms, data, artificial intelligence, recommendation systems, pricing engines, ranking systems, or automated decision-making tools amplify the merged firm's ability to raise prices, reduce quality, restrict access, degrade privacy, suppress innovation, or disadvantage rivals.

Traditional merger analysis often asks whether a transaction will eliminate an important competitive constraint and thereby allow the merged firm to act independently. In digital markets, that constraint may be embedded in data, algorithms, user attention, interoperability, network effects, recommendation systems, or platform architecture, rather than simply in a conventional price.

The principal concern can therefore be expressed as:

Merger + algorithmic capability + data/network effects + loss of competitive constraint = potential unilateral digital-market effects.

The analysis is particularly important in platform markets because the acquired firm may appear to have modest current revenues while possessing strategically important technology, data, users, engineers, algorithms, or future competitive potential.

2. Meaning of Unilateral Effects

A unilateral effect occurs when a merger reduces competitive pressure sufficiently for the merged undertaking to behave less competitively on its own.

This differs from coordinated effects, where the concern is that the merger makes coordination between remaining competitors easier.

Traditional unilateral-effects mechanism

Suppose:

• Firm A competes closely with Firm B;
• A acquires B;
• consumers who would previously have switched from A to B have fewer alternatives;
• A can consequently increase price or reduce quality.

In digital markets, the same mechanism may involve:

• algorithmic pricing;
• search rankings;
• recommendation algorithms;
• personalised advertising;
• app-store ranking;
• data-driven targeting;
• interoperability;
• API access;
• privacy;
• innovation;
• AI training data; or
• platform access conditions.

Thus, the competitive harm may occur even where consumer prices remain zero.

3. Why Algorithms Change Merger Analysis

Digital algorithms can transform the competitive significance of an acquisition in several ways.

A. Automated pricing

The merged firm may possess superior pricing algorithms capable of:

• personalised pricing;
• demand forecasting;
• real-time price optimisation;
• customer segmentation;
• dynamic discounts; and
• automated experimentation.

B. Recommendation systems

A platform controlling a recommendation algorithm may determine:

• which products users see;
• which applications receive visibility;
• which sellers are promoted;
• which content is recommended; and
• which competing services become less discoverable.

C. Data advantages

A merger may combine datasets that cannot easily be replicated by rivals.

For example:

Platform A's behavioural data + Platform B's transaction data → superior prediction algorithm → stronger competitive position.

D. Network effects

An algorithm may become more effective as the platform obtains more:

• users;
• transactions;
• searches;
• reviews;
• behavioural information; and
• advertising interactions.

The merger may therefore produce a feedback loop:

More users → more data → better algorithm → better service → more users.

E. Reduced innovation

A digital acquisition may remove a potential competitor whose algorithm or technology could otherwise develop into a significant competitive constraint.

4. Algorithmic Unilateral Effects vs Traditional Unilateral Effects

Accordingly, merger authorities increasingly examine non-price competitive parameters.

5. Principal Forms of Algorithmic Unilateral Effects

5.1 Algorithmic Pricing Power

A merger may allow the combined undertaking to deploy better algorithms for pricing.

For example, Firm A and Firm B independently use pricing systems. After the merger, the merged firm can combine:

• historical transaction data;
• consumer profiles;
• competitor information;
• demand elasticity estimates; and
• behavioural data.

The resulting algorithm may identify precisely how much each consumer segment is willing to pay.

The competitive concern is not necessarily that the algorithm colludes with competitors. Rather:

The merged firm may independently optimise prices because the merger has enhanced its information and computational capabilities.

6. Algorithmic Self-Preferencing

A platform may use an algorithm to give preferential treatment to its own services after acquiring another company.

Potential mechanisms include:

• higher search ranking;
• preferred recommendation;
• preferential placement;
• lower latency;
• better access to APIs;
• greater advertising visibility;
• default installation; or
• preferential interoperability.

The unilateral effect becomes particularly significant where the acquired business provides technology that can be integrated into the platform's ranking or recommendation architecture.

7. Data-Driven Unilateral Effects

Data can constitute an important competitive input.

A merger can combine datasets relating to:

• consumer behaviour;
• purchasing patterns;
• location;
• browsing;
• financial transactions;
• search queries;
• advertising interactions;
• health or fitness activity;
• device usage.

The combined dataset may allow the merged company to produce algorithms that competitors cannot readily reproduce.

Competitive mechanism

Merger

↓

Data combination

↓

Improved prediction

↓

Improved personalisation

↓

Higher user engagement

↓

More data

↓

Further algorithmic improvement

This is sometimes described as a data-feedback loop.

8. Algorithmic Quality Reduction

Digital competition does not depend exclusively on monetary price.

A merged platform may reduce:

• search quality;
• recommendation neutrality;
• privacy protection;
• customer service;
• interoperability;
• content diversity;
• security; or
• innovation.

Consumers may nevertheless remain because switching costs are high.

Thus, the relevant unilateral effect can be:

Reduced quality without an equivalent monetary price increase.

9. Algorithmic Innovation Effects

This is particularly important in acquisitions of start-ups.

A start-up may currently have:

• low revenue;
• limited users;
• negative profits; or
• a small market share.

Yet its algorithm may represent a significant future competitive constraint.

The merger can eliminate:

1. an emerging competitor;
2. a source of innovation;
3. a technological alternative;
4. a potential substitute; or
5. a platform that could challenge an incumbent.

Consequently, current market share may understate competitive significance.

10. Algorithmic Network Effects

Digital markets frequently exhibit direct and indirect network effects.

For example:

More users → more transactions → more data → better algorithm → better matching → more users.

If a dominant platform acquires a rapidly growing platform, the acquisition can strengthen the network effect.

This may result in:

• increased entry barriers;
• reduced multi-homing;
• greater user lock-in;
• increased data concentration; and
• reduced contestability.

11. Algorithmic Switching Costs

Algorithms can also create personalised ecosystems.

A consumer's:

• search history;
• recommendations;
• playlists;
• purchase history;
• social graph;
• preferences; and
• stored data

may be embedded within a platform.

The merger can therefore make switching more difficult because consumers lose the accumulated benefits of personalisation.

12. Role of Market Definition

Algorithmic unilateral-effects analysis still requires careful market definition, but conventional definitions can be inadequate.

Relevant dimensions may include:

Product dimension

• search;
• social networking;
• online advertising;
• app distribution;
• cloud computing;
• digital payments;
• e-commerce;
• digital mapping;
• online travel;
• wearable technology; or
• AI services.

Geographic dimension

Digital markets can be:

• national;
• regional;
• global; or
• platform-specific.

Non-price dimensions

Authorities may also examine:

• quality;
• privacy;
• innovation;
• data access;
• interoperability;
• latency;
• algorithmic accuracy.

13. Evidence Used by Competition Authorities

Authorities may examine:

Internal documents

• strategy presentations;
• product roadmaps;
• acquisition documents;
• emails;
• board papers;
• engineering documents.

Algorithmic evidence

• ranking methodologies;
• pricing models;
• recommendation systems;
• machine-learning models;
• training datasets.

Economic evidence

• diversion ratios;
• switching data;
• customer surveys;
• bidding data;
• elasticity;
• concentration measures.

Technical evidence

• API architecture;
• interoperability;
• data portability;
• technical compatibility;
• system integration.

14. Six Major Case Laws and Merger Decisions

Case 1: Facebook/WhatsApp — European Commission

Case: Facebook/WhatsApp, European Commission, Case COMP/M.7217 (2014).

The transaction involved Facebook's acquisition of WhatsApp.

The Commission examined, among other things, the competitive significance of:

• consumer communications;
• social networking;
• online advertising; and
• data.

The case is important for algorithmic unilateral-effects analysis because it illustrates the difficulty of assessing data advantages in digital mergers.

The Commission considered whether Facebook's access to WhatsApp user data could strengthen Facebook's position in online advertising.

Although the Commission ultimately cleared the transaction subject to its assessment at the time, the case became an important reference point for later digital merger analysis.

Principle

A digital merger may create competitive concerns through data accumulation even where the immediate product markets have zero monetary prices.

15. Case 2: Google/Fitbit — European Commission

Case: Google/Fitbit, European Commission, Case M.9660 (2020).

Google proposed acquiring Fitbit, a provider of wearable devices and associated health and fitness data.

The Commission examined whether the transaction could strengthen Google's position in:

• online advertising;
• digital healthcare;
• wearable devices; and
• data-related markets.

The Commission's analysis is particularly relevant because the transaction potentially combined:

Google's advertising/data ecosystem + Fitbit's health and fitness data.

The Commission imposed commitments relating to the use of certain Fitbit data for Google's advertising activities.

Algorithmic significance

The case demonstrates how merger control can address potential unilateral effects resulting from combining datasets that improve algorithmic targeting and personalisation.

Principle

Data accumulation may constitute a competitive parameter even where the acquired firm's immediate market share does not fully capture its strategic importance.

16. Case 3: Microsoft/Activision Blizzard

Case: Microsoft/Activision Blizzard, European Commission, Case M.10646 (2023).

The transaction involved Microsoft's acquisition of Activision Blizzard, including important gaming assets such as the Call of Duty franchise.

The Commission examined competition involving:

• gaming consoles;
• cloud game streaming;
• PC operating systems; and
• distribution of gaming content.

The Commission accepted commitments concerning cloud gaming.

Algorithmic significance

Cloud gaming is strongly dependent on:

• technical infrastructure;
• recommendation and distribution systems;
• user data;
• platform ecosystems;
• access conditions; and
• automated allocation and streaming technology.

The case illustrates that digital merger analysis can focus on ecosystem effects and access to technological infrastructure, rather than simply traditional price effects.

Principle

A merger involving valuable digital content can affect competition in adjacent technology markets through ecosystem integration and control over access.

17. Case 4: Meta/Kustomer

Case: Meta Platforms/Kustomer, European Commission, Case M.10262 (2022).

Meta proposed acquiring Kustomer, a customer relationship management software provider.

The Commission examined competition in relation to customer relationship management and related digital services.

The transaction raised concerns about Meta's position in digital ecosystems and the potential importance of customer-data resources.

Algorithmic significance

CRM systems can generate extensive information concerning:

• customer interactions;
• purchasing behaviour;
• communication;
• preferences;
• engagement;
• business relationships.

Combining such data with a large digital platform may increase the platform's ability to develop prediction, targeting, recommendation and automation algorithms.

Principle

The strategic value of a digital target may extend beyond its current revenues because its technology and data can reinforce an existing ecosystem.

18. Case 5: Amazon/iRobot

Case: Amazon/iRobot, European Commission, Case M.10920 (2024).

Amazon's proposed acquisition of iRobot was examined in the context of the European Commission's digital-platform and connected-device concerns.

The transaction involved a combination of:

• Amazon's ecosystem;
• iRobot's robotic products;
• consumer data;
• online retail;
• connected devices; and
• platform distribution.

The Commission ultimately prohibited the transaction.

Algorithmic significance

Smart-home and connected-device markets can generate extensive data concerning:

• consumer behaviour;
• household environments;
• product usage;
• purchasing patterns; and
• interaction with digital ecosystems.

The transaction therefore illustrates how competition analysis can extend beyond immediate product overlaps and consider ecosystem and data-driven competitive dynamics.

Principle

Digital and connected-device acquisitions can raise concerns where control over technology, distribution and data may reinforce an ecosystem's competitive position.

19. Case 6: Booking Holdings/eTraveli

Case: Booking Holdings/eTraveli Group, European Commission, Case M.10615 (2023).

Booking proposed acquiring eTraveli, a major flight-booking service.

The Commission's investigation focused on the relationship between:

• flight distribution;
• hotel booking;
• online travel platforms;
• consumer traffic; and
• ecosystem effects.

The Commission prohibited the transaction.

Algorithmic significance

Online travel platforms rely heavily upon algorithms for:

• ranking;
• recommendation;
• personalisation;
• pricing;
• search results;
• cross-selling;
• customer acquisition.

A transaction that strengthens an ecosystem can therefore affect competitive conditions even where the acquired service operates in a related rather than identical market.

Principle

Digital merger analysis may consider ecosystem leverage and the ability to use one service to reinforce another service.

20. Additional Important Authorities

20.1 Illumina/Grail

Case: Illumina/Grail, European Commission.

The transaction concerned genomic cancer-detection technology.

The matter is important to digital/technology merger analysis because it demonstrates the importance of:

• innovation competition;
• nascent technologies;
• pipeline products; and
• future competitive constraints.

It illustrates why authorities cannot always rely upon current market shares when evaluating innovative technology markets.

20.2 Microsoft/LinkedIn

Case: Microsoft/LinkedIn, European Commission, Case M.8124 (2016).

The Commission examined the combination of Microsoft's software ecosystem with LinkedIn's professional social-networking platform.

Relevant concerns included:

• data;
• professional networking;
• software integration;
• interoperability;
• platform ecosystems.

The case remains useful for understanding data-driven ecosystem effects.

21. The Counterfactual Problem

Algorithmic merger analysis requires identifying what would happen without the merger.

This is particularly difficult where the target is a rapidly developing technology company.

The authority may have to determine whether:

• the target would remain independent;
• the target would innovate;
• its algorithm would improve;
• it would expand into an adjacent market;
• it would become a competitive constraint; or
• another firm would acquire it.

The counterfactual is therefore frequently more uncertain than in mature industries.

22. The "Killer Acquisition" Problem

A dominant platform may acquire a small company precisely because its algorithm represents a potential future competitive threat.

A transaction can therefore produce unilateral effects even when:

Target's current market share ≈ small

but:

Target's innovation potential = significant.

This is particularly relevant to:

• AI;
• fintech;
• cybersecurity;
• health technology;
• search;
• social media;
• cloud computing;
• digital advertising;
• recommendation systems.

23. Algorithmic Effects and Non-Price Competition

Competition authorities should consider multiple competitive parameters.

24. Efficiencies Defence

The merging parties may argue that combining algorithms and datasets produces legitimate efficiencies.

Potential efficiencies include:

• better fraud detection;
• improved search;
• reduced delivery times;
• better cybersecurity;
• improved recommendations;
• lower transaction costs;
• improved AI models;
• better product matching.

However, the relevant legal question is whether such efficiencies are:

1. verifiable;
2. merger-specific;
3. sufficiently likely;
4. timely; and
5. capable of benefiting consumers.

An asserted algorithmic efficiency cannot automatically eliminate a unilateral-effects concern.

25. Remedies

Competition authorities may employ structural or behavioural remedies.

Structural remedies

• divestiture;
• sale of technology;
• separation of business units.

Behavioural remedies

• data-access commitments;
• interoperability;
• API access;
• non-discrimination;
• licensing;
• restrictions on data combination;
• firewall obligations;
• portability;
• restrictions on self-preferencing.

Algorithm-specific remedies

Authorities may potentially require:

• independent auditing;
• transparency obligations;
• monitoring of ranking algorithms;
• restrictions on data integration;
• technical interoperability;
• access to essential interfaces.

The effectiveness of behavioural remedies is especially important because algorithms can evolve after the merger.

26. Difficulties in Algorithmic Merger Control

A. Black-box algorithms

Authorities may not understand exactly how a machine-learning system reaches decisions.

B. Rapid technological change

Market conditions may change before the investigation is completed.

C. Data valuation

It is difficult to quantify the precise competitive value of a dataset.

D. Future innovation

The competitive significance of an emerging technology may be uncertain.

E. Zero-price services

Traditional price-based merger tests become less informative.

F. Multi-sided markets

A platform may simultaneously serve:

• consumers;
• advertisers;
• sellers;
• developers;
• service providers.

G. Ecosystem effects

Harm may arise through several connected markets rather than one narrowly defined market.

27. Analytical Framework

A useful framework for analysing algorithmic unilateral effects is:

Step 1 — Identify the digital ecosystem

Determine the relevant:

• platform;
• service;
• technology;
• users;
• suppliers.

Step 2 — Identify the algorithmic capability

Determine whether the target possesses:

• pricing technology;
• recommendation technology;
• ranking algorithms;
• AI models;
• matching algorithms;
• predictive analytics.

Step 3 — Identify data assets

Examine:

• quantity;
• quality;
• uniqueness;
• timeliness;
• replicability.

Step 4 — Determine competitive closeness

Ask whether the target constrains the acquirer through:

• price;
• quality;
• innovation;
• technology;
• privacy;
• user experience.

Step 5 — Analyse network effects

Determine whether the transaction increases:

• user numbers;
• data accumulation;
• interoperability;
• switching costs;
• ecosystem dependence.

Step 6 — Analyse foreclosure

Consider whether the merged firm can use algorithms to:

• rank rivals lower;
• deny API access;
• degrade interoperability;
• bundle services;
• restrict data;
• favour its own products.

Step 7 — Examine efficiencies

Assess whether claimed algorithmic efficiencies are merger-specific and verifiable.

Step 8 — Evaluate remedies

Determine whether behavioural remedies can realistically constrain an adaptive algorithm.

28. Important Case-Law Principles — Consolidated

29. Relationship with Modern Digital Merger Regulation

Algorithmic unilateral effects have become increasingly relevant to modern merger regimes because digital markets exhibit characteristics that conventional merger analysis may understate:

Data concentration

Network effects

Algorithmic optimisation

Low switching

Zero-price services

Innovation competition

=

Potential unilateral digital-market power

This does not mean that every acquisition involving algorithms is anticompetitive. The legal analysis remains transaction-specific and requires evidence that the merger is likely to eliminate or substantially weaken a competitive constraint.

30. Conclusion

Algorithmic unilateral effects represent an important extension of traditional merger-control theory into digital markets. The essential concern is that a merger may allow the combined undertaking to exploit algorithms, data, network effects and ecosystem integration to exercise greater market power independently.

The most significant issues include:

1. algorithmic pricing;
2. data accumulation;
3. personalised targeting;
4. self-preferencing;
5. ranking manipulation;
6. reduced interoperability;
7. innovation suppression;
8. increased switching costs;
9. ecosystem leverage; and
10. elimination of nascent algorithmic competitors.

The decisions in Facebook/WhatsApp, Google/Fitbit, Microsoft/LinkedIn, Microsoft/Activision Blizzard, Meta/Kustomer, Booking/eTraveli, Amazon/iRobot and Illumina/Grail demonstrate different aspects of the modern approach to digital merger control.