Competition Law and Recommendation Hierarchy Competition Issues

1. Introduction

Recommendation hierarchy refers to the order in which a digital platform presents products, services, sellers, advertisements, search results, videos, apps, restaurants, hotels, news stories, or other content to users. The hierarchy may be generated by algorithms using factors such as relevance, price, quality, engagement, conversion probability, user history, commissions, advertising payments, inventory, or platform-specific commercial objectives.

Recommendation ranking is not inherently anti-competitive. A platform may legitimately design an algorithm to improve relevance, quality, safety, or consumer experience. Competition concerns arise when a platform with substantial market power uses control over ranking or recommendation visibility to disadvantage competing businesses, favour its own products, discriminate among business users, manipulate access to customers, or make market entry and expansion more difficult.

The issue has become particularly important because algorithms can determine commercial visibility without an obvious contractual exclusion. The European Court of Justice's Google Shopping judgment is a leading authority: Google was found to have favoured its own comparison-shopping service through prominent placement while competing services were demoted by its ranking mechanisms. The Court of Justice upheld the €2.4 billion fine in September 2024.

2. Meaning of Recommendation Hierarchy

A recommendation hierarchy can operate at several levels:

A. Search-result ranking

Examples:

• first-page search results;
• featured results;
• shopping results;
• local-business results;
• hotel or travel recommendations.

B. Marketplace ranking

A marketplace may determine:

• which seller appears first;
• which seller receives the "Buy Now" or "Buy Box";
• which products are recommended;
• which seller receives default selection.

C. Content recommendation

Platforms may rank:

• videos;
• news;
• social-media posts;
• music;
• games;
• apps;
• advertisements.

D. Personalised recommendations

The algorithm may use individual data to determine what each user sees.

Thus, recommendation hierarchy can effectively become a commercial allocation mechanism: ranking first can translate into greater traffic, sales, data accumulation and advertising revenue.

3. Why Recommendation Hierarchy Creates Competition Concerns

3.1 Self-preferencing

The principal concern is self-preferencing.

A vertically integrated platform may simultaneously operate:

Platform + competing downstream business

For example:

Marketplace → independent sellers + platform's own products

If the platform systematically places its own products above independent sellers despite equivalent or inferior objective ranking criteria, competition concerns may arise.

The ACCC has specifically identified preferential treatment of a platform's own products and services as a potential digital-platform competition problem.

4. Algorithmic Discrimination

A recommendation system may treat otherwise comparable businesses differently.

For example:

The difference does not automatically establish an infringement. Competition authorities would need to examine:

1. the relevant market;
2. platform dominance;
3. ranking criteria;
4. actual treatment of comparable firms;
5. economic effects;
6. legitimate efficiency explanations;
7. whether the conduct forecloses competitors.

The critical question is therefore not simply "Was the ranking different?", but:

Was the difference connected to legitimate competition on the merits, or was ranking power used to distort competitive conditions?

5. Foreclosure of Competitors

Recommendation hierarchy can produce foreclosure effects because users generally concentrate disproportionately on highly visible results.

A competitor pushed from:

Position 1 → Position 20

may lose substantial traffic even though the competitor remains technically available.

The competition concern is therefore particularly significant where:

• users rarely examine lower-ranked results;
• switching costs are high;
• the platform controls access to a critical customer base;
• competitors depend heavily on platform traffic;
• ranking data is unavailable to competitors;
• the platform's own downstream service receives preferential visibility.

The UK's government discussion of algorithmic competition specifically identifies manipulation of ranking algorithms to favour a platform's own products as a potential mechanism of competitive harm.

6. Recommendation Hierarchy and Article 102 / Abuse of Dominance

Under Article 102 TFEU and comparable national provisions, recommendation hierarchy can become problematic when undertaken by a dominant undertaking and capable of producing exclusionary effects.

Potential theories include:

6.1 Discriminatory treatment

A dominant platform may apply different ranking conditions to competing businesses without objective justification.

6.2 Leveraging

Market power in an upstream platform or search market may be leveraged into:

• shopping;
• travel;
• maps;
• advertising;
• financial services;
• local search;
• app distribution.

6.3 Self-preferencing

The platform may systematically give its own downstream service preferential treatment.

6.4 Refusal or degradation of access

Competitors may technically remain on the platform but receive substantially reduced visibility.

6.5 Exploitative ranking conditions

Ranking may also affect business users through:

• increased commissions;
• paid visibility;
• advertising requirements;
• mandatory platform services;
• opaque ranking conditions.

7. Recommendation Hierarchy and Market Definition

Market definition is particularly difficult in digital markets.

A platform may operate several interconnected markets:

General search

↓

Specialised search

↓

Shopping / travel / maps / local services

↓

Advertising

Recommendation hierarchy may therefore allow a dominant undertaking in one market to influence competitive conditions in another.

The Google Shopping litigation illustrates this structure: the EU courts considered Google's position in general search and its conduct concerning specialised comparison-shopping services. The Court of Justice characterised the case as involving an abuse of dominance and favouring Google's own specialised search service.

8. Six Important Case Laws / Authorities

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

European Commission decision of 27 June 2017; General Court T-612/17; CJEU C-48/22 P, Google and Alphabet v Commission, judgment of 10 September 2024

This is the most important authority for recommendation and ranking hierarchy.

Google displayed its own comparison-shopping results prominently on general search-result pages while competing comparison-shopping services were subject to algorithms that could demote them.

The CJEU upheld the infringement and the approximately €2.4 billion fine.

The Court's 2024 judgment confirmed that the conduct could constitute abuse where the dominant search platform gives preferential display to its own specialised service and thereby affects competition.

Principle

A dominant platform's ranking and presentation mechanisms can constitute an abuse of dominance when they favour its own downstream service and disadvantage competing services.

Relevance

This authority establishes the importance of:

• visibility;
• ranking;
• algorithmic demotion;
• self-preferencing;
• foreclosure;
• causal effects.

Case 2 — Streetmap.EU Ltd v Google Inc

[2016] EWHC 253 (Ch)

Streetmap challenged Google's prominent presentation of Google Maps within search results.

The claimant argued that Google's dominant general-search position was being used to favour Google Maps over competing online mapping services.

The High Court considered the conduct in terms of dominance, discrimination, foreclosure, effects and objective justification. It ultimately rejected Streetmap's claim, finding insufficient causal evidence that the Maps OneBox caused the alleged competitive harm.

Principle

Preferential visibility alone does not automatically establish an infringement.

A claimant must establish the necessary competitive effects and causal relationship.

Importance

This case provides an important counterpoint to Google Shopping:

Self-preferencing allegations require evidence of competitive harm, not merely proof that the platform's own service received prominent placement.

Case 3 — Yelp Inc. v Google LLC

U.S. District Court for the Northern District of California, 2024–2025 litigation

Yelp alleged that Google used its general-search position to favour Google's own local-search services and place them prominently in search results while competing services such as Yelp were pushed downward.

The litigation is significant because it illustrates the different treatment of self-preferencing under U.S. antitrust law compared with European competition law.

In 2025, the court allowed aspects of Yelp's refined tying theory to proceed, including allegations concerning Google's presentation of local-search results through its OneBox.

Principle

U.S. antitrust analysis may require additional proof concerning:

• tying;
• coercion;
• monopoly power;
• foreclosure;
• competitive effects.

Importance

The case demonstrates that the same ranking conduct may generate different legal theories under different competition regimes.

Case 4 — FTC Investigation into Google Search Practices

The FTC's investigation into Google's search practices is an important regulatory authority concerning search ranking and self-preferencing.

The FTC staff considered allegations that Google had:

• scraped content from vertical competitors;
• used competitors' information to improve its own services;
• preferentially positioned its own content;
• demoted competing websites.

The FTC staff ultimately did not recommend proceeding on the self-preferencing theory at that time, citing Google's efficiency justifications, although it considered other conduct more problematic. The record is discussed in later U.S. litigation concerning Google's ranking practices.

Principle

Competition analysis must distinguish between:

legitimate product improvement

and

anticompetitive exclusion.

Importance

This authority demonstrates why algorithmic ranking cases require detailed economic and technical evidence.

Case 5 — United States v Microsoft Corp.

D.C. Circuit, 2001

Although not a recommendation-engine case in the modern sense, Microsoft is highly relevant to platform-controlled distribution and visibility.

Microsoft used contractual arrangements and platform control to make Internet Explorer more difficult for competing browsers to distribute.

The case established important principles concerning:

• exclusionary conduct;
• platform power;
• distribution restrictions;
• foreclosure;
• effects on technological competition.

Its relevance to recommendation hierarchy is conceptual:

A platform can influence competition not merely by preventing access, but by controlling the conditions through which users encounter competing products.

The modern equivalent can be an algorithmic ranking system rather than an explicit contractual restriction.

Case 6 — FTC / State Attorneys General v Amazon

FTC et al. v Amazon.com Inc., U.S. District Court for the Western District of Washington, filed 2023

The FTC and 18 states alleged that Amazon maintained monopoly power through interconnected practices affecting sellers and consumers. The case includes allegations concerning Amazon's marketplace architecture and treatment of sellers, making it relevant to the broader question of how platform rules can affect seller visibility and competition.

The litigation remains significant because Amazon's marketplace simultaneously functions as:

platform + retailer + logistics provider + advertising intermediary.

That combination creates potential conflicts where platform rules determine which sellers or offers receive commercially valuable visibility.

Principle

Where a platform controls access to customers while simultaneously competing with businesses using the platform, ranking and marketplace design can become relevant to exclusionary-conduct analysis.

Important: allegations in the FTC case should not be treated as judicial findings of infringement.

9. Comparative Significance of the Six Authorities

10. Ranking Neutrality

A major regulatory concept is ranking neutrality.

It does not necessarily mean that every result must receive identical treatment.

Rather, competition law may ask whether:

1. the same ranking criteria are applied consistently;
2. platform-owned products receive hidden advantages;
3. competitors can satisfy the same criteria;
4. ranking criteria change selectively;
5. paid placement is clearly distinguished;
6. commercial relationships influence organic ranking;
7. the algorithm systematically disadvantages rivals.

The UK's current digital-markets regime provides an especially important modern development: in June 2026 the CMA imposed a fair-ranking conduct requirement on Google concerning general search services following Google's designation as having strategic market status.

This shows the movement from traditional ex-post antitrust enforcement toward ex-ante obligations governing platform ranking behaviour.

11. Transparency Problems

Recommendation algorithms may be proprietary.

Consequently, competitors may not know:

• why they were demoted;
• which ranking factor changed;
• whether competitors receive preferential treatment;
• whether advertising affects organic ranking;
• whether commissions influence recommendations;
• whether platform-owned products receive exceptions.

This creates an evidentiary asymmetry.

The platform possesses:

algorithm + data + logs + experiments + ranking history

while the affected competitor may possess only:

before-and-after traffic statistics.

Therefore, competition authorities may need access to:

• ranking logs;
• A/B testing records;
• algorithmic documentation;
• internal communications;
• source-code documentation where appropriate;
• seller-level data;
• traffic data;
• click-through rates;
• conversion rates;
• historical rankings.

12. Paid Ranking and Advertising

A particularly important distinction is between:

Legitimate sponsored placement

A platform clearly identifies:

Sponsored / Advertisement

and allows businesses to purchase visibility under transparent conditions.

Potential competition concern

The problem becomes more serious where:

• payment is required to obtain ordinary visibility;
• platform-owned products receive equivalent benefits without paying;
• independent sellers are systematically disadvantaged;
• paid ranking is disguised as organic recommendation;
• the platform uses advertising dominance to disadvantage non-advertising competitors.

Thus, competition law must distinguish advertising competition from manipulation of organic recommendation systems.

13. Data Advantage and Recommendation Hierarchy

Recommendation systems become more powerful as platforms accumulate data.

A dominant platform may possess:

• search histories;
• purchases;
• clicks;
• conversion rates;
• seller performance;
• consumer preferences;
• inventory data;
• pricing data;
• advertising performance.

This produces a feedback loop:

More users

↓

More behavioural data

↓

Better recommendations

↓

More transactions

↓

More data

↓

Better recommendations

This can create data-driven entry barriers.

A new competitor may therefore face difficulty reproducing the incumbent's recommendation quality even where the underlying algorithmic technology is technically replicable.

14. Network Effects

Recommendation hierarchy can also reinforce network effects.

For example:

More sellers

→ more products

→ more consumers

→ more transactions

→ more data

→ better recommendations

→ more sellers

This can create substantial advantages for established platforms.

The competition concern becomes stronger when the platform uses this position to systematically disadvantage competing platforms or downstream businesses.

15. Algorithmic Feedback Loops

One of the most important emerging issues is that algorithms may create their own competitive advantages.

Suppose an algorithm ranks products according partly to historical sales.

Then:

Product A receives #1 ranking

→ more users see A

→ A receives more sales

→ algorithm observes higher sales

→ algorithm ranks A even higher.

This is a feedback loop.

The original ranking advantage may therefore become self-reinforcing.

Competition authorities should consequently distinguish:

• ranking based on genuine consumer demand;
• ranking based on platform-controlled visibility;
• ranking based on paid placement;
• ranking based on self-preferencing.

16. Recommendation Hierarchy and Consumer Harm

Competition concerns can eventually affect consumers through:

Higher prices

Reduced competitive pressure may permit higher prices.

Lower quality

Competitors receiving insufficient traffic may reduce investment.

Reduced innovation

New entrants may find it difficult to achieve sufficient scale.

Reduced choice

Users may repeatedly encounter the platform's preferred products.

Reduced discovery

Consumers may never see relevant competing products located lower in the ranking.

The consumer-harm analysis should therefore examine both:

short-term convenience

and

long-term competitive effects.

17. Objective Justification

A platform can defend a ranking algorithm by showing legitimate reasons such as:

• relevance;
• product quality;
• safety;
• fraud prevention;
• delivery performance;
• consumer satisfaction;
• availability;
• compatibility;
• privacy;
• cybersecurity.

For example, placing a product first because it has a substantially better delivery record may be legitimate competition on the merits.

The legal problem arises where the stated criterion is merely a pretext for:

favouring the platform's own product or excluding competitors.

18. Remedies

Possible competition-law remedies include:

1. Algorithmic non-discrimination

Require comparable products to be evaluated using comparable criteria.

2. Self-preferencing prohibition

Prevent preferential treatment of the platform's own downstream services.

3. Ranking transparency

Require disclosure of significant ranking parameters.

4. Audit obligations

Allow regulators or independent auditors to test recommendation systems.

5. Data-access remedies

Provide competitors with appropriate access to certain data.

6. Interoperability

Allow competitors to connect with platform systems.

7. Choice screens

Provide users with competing alternatives rather than automatically selecting the platform's own service.

8. Structural remedies

In extreme circumstances, competition authorities may consider separation of platform and downstream commercial activities, depending on the applicable legal regime.

19. Key Legal Tests

A useful analytical framework is:

Step 1 — Identify the platform

Who controls the recommendation system?

Step 2 — Define the market

What market is affected?

Step 3 — Establish market power

Does the platform possess dominance or strategic market power?

Step 4 — Identify the ranking mechanism

What determines recommendation position?

Step 5 — Compare treatment

Are the platform's products treated differently from rivals?

Step 6 — Establish competitive effects

Does the conduct reduce:

• traffic;
• sales;
• entry;
• innovation;
• multi-homing;
• competitor scale?

Step 7 — Examine causation

Can the loss of competitive opportunity actually be connected to the ranking practice?

Step 8 — Examine efficiencies

Does the ranking improve:

• relevance;
• quality;
• safety;
• consumer welfare?

Step 9 — Assess proportionality

Is the alleged restriction reasonably necessary to achieve the claimed efficiency?

Step 10 — Select remedy

Possible remedies include transparency, non-discrimination, algorithmic auditing, access, interoperability, or behavioural restrictions.

20. Emerging Competition Issues

Recommendation hierarchy is likely to generate further competition-law questions involving:

1. Generative-AI recommendation engines
2. AI shopping assistants
3. voice assistants
4. autonomous purchasing agents
5. travel recommendation systems
6. financial-product recommendations
7. healthcare recommendations
8. app-store ranking
9. social-media recommendation algorithms
10. news-feed ranking
11. retail-media algorithms
12. hotel and restaurant ranking
13. marketplace Buy Box systems
14. algorithmic personalised pricing
15. AI-generated product comparison

The problem becomes particularly complex where an AI system simultaneously determines what the consumer sees, which products are recommended, and which product is ultimately purchased.

21. Conclusion

Recommendation hierarchy is becoming an important competition-law control point because ranking can determine commercial visibility without formally excluding a competitor from the market.

The central legal distinction is between:

competition-enhancing recommendation

and

strategic manipulation of recommendation power to disadvantage rivals.

The leading Google Shopping litigation demonstrates that algorithmic placement and demotion can form the basis of an abuse-of-dominance finding. Conversely, Streetmap v Google demonstrates that preferential display alone does not necessarily establish liability; competitive effects, causation and objective justification remain important.

Accordingly, future competition-law analysis of recommendation engines will increasingly focus on self-preferencing, ranking discrimination, data advantages, algorithmic feedback loops, foreclosure, transparency, platform conflicts of interest and measurable effects on competitors and consumers. Recent regulatory developments, including the UK's 2026 fair-ranking requirement for Google, indicate that ranking governance is also moving beyond traditional case-by-case antitrust enforcement toward more specific ex-ante digital-platform obligations.