Illusion of Competition Through Controlled Metrics

Introduction

Illusion of competition through controlled metrics describes a situation in which a market appears competitive because firms, platforms, regulators, or industry bodies publish favourable performance indicators—such as prices, market shares, rankings, switching rates, response times, quality scores, or innovation statistics—while the underlying methodology, data inputs, benchmarks, or measurement system is controlled by a dominant undertaking.

The central competition-law concern is that competition can be manipulated not only through prices or output, but also through control over the metrics by which competition is observed and evaluated.

A market may therefore show:

• apparently declining prices while quality deteriorates;
• apparently high switching while switching is practically difficult;
• apparently low market concentration because the relevant market is defined narrowly;
• apparently strong innovation despite exclusionary conduct;
• apparently neutral rankings despite algorithmic manipulation;
• apparently open access despite discriminatory technical standards; or
• apparently competitive bidding despite manipulation of the parameters used to compare bids.

The problem becomes particularly acute in digital markets, where the undertaking controlling the platform may simultaneously control the data, ranking system, measurement methodology, API, analytics, and access to the underlying market.

1. Meaning of Controlled Metrics

A metric is a quantitative or qualitative measure used to assess market performance.

Examples include:

A metric becomes problematic when the undertaking being evaluated can control the construction, collection, weighting, presentation, or interpretation of that metric.

The key distinction is therefore:

Competition measured independently is different from competition measured through a system controlled by one of the competitors.

2. How the Illusion of Competition Arises

The phenomenon can be understood through a six-stage mechanism.

Stage 1 — Creation of the metric

The dominant undertaking establishes a measurement system.

For example:

"Competition is healthy because users can switch providers within five minutes."

But the undertaking may define "switching" as merely opening another account, ignoring data migration, loss of history, interoperability barriers, or contractual penalties.

Stage 2 — Control over data

The undertaking controls the underlying information.

For example, a platform may possess detailed information concerning:

• consumer searches;
• seller impressions;
• ranking;
• transactions;
• cancellations;
• switching;
• advertising expenditure; and
• competitor performance.

Competitors and regulators may receive only aggregated statistics.

Stage 3 — Selective measurement

Only favourable indicators are disclosed.

A platform might highlight:

"Average prices declined by 10%."

while excluding:

• deterioration in product quality;
• increased advertising fees;
• reduced organic visibility;
• increased data extraction;
• higher switching costs.

Stage 4 — Benchmark manipulation

The undertaking establishes the baseline against which performance is measured.

A small improvement can therefore be represented as substantial competitive progress if the baseline is strategically selected.

Stage 5 — Regulatory reliance

Authorities, investors, consumers, or courts may rely on the metric because it appears objective.

The numerical presentation creates an impression of neutrality.

Stage 6 — Competitive reality diverges

The underlying market becomes less contestable even while the published indicators continue to suggest vigorous competition.

This produces the illusion of competition.

3. Competition Law Significance

Controlled metrics may engage several competition-law doctrines.

A. Abuse of dominance

Under abuse-of-dominance principles, metric manipulation may form part of:

• exclusionary conduct;
• discriminatory access;
• self-preferencing;
• refusal to provide essential information;
• exploitative conduct;
• tying or leveraging;
• degradation of interoperability; or
• manipulation of ranking and visibility.

The metric need not itself constitute the abuse. It can be evidence of a broader exclusionary strategy.

4. Market Definition and Controlled Metrics

Market definition is particularly vulnerable.

Suppose a dominant platform claims:

"Our market share is only 25%."

That figure may be misleading if the market is defined around a narrow product category while consumers actually regard several adjacent services as substitutes.

Conversely, an undertaking might artificially broaden the market to dilute its apparent dominance.

Therefore, competition authorities must examine:

• demand substitution;
• supply substitution;
• multi-homing;
• geographic scope;
• switching costs;
• network effects;
• data advantages;
• ecosystem effects; and
• potential competition.

The metric cannot replace substantive market analysis.

5. Quality Metrics and the "Zero-Price" Problem

Digital platforms provide a particularly important example.

Suppose users receive a service for zero monetary consideration.

A platform may therefore claim:

"Consumers pay nothing, so competition is strong."

But the relevant competitive variables may instead include:

• privacy;
• advertising intensity;
• data extraction;
• interoperability;
• algorithmic neutrality;
• service quality;
• security;
• user choice.

Consequently, price can become an incomplete or misleading metric of competition.

A zero-price market may still involve substantial competitive harm.

6. Ranking Metrics as Instruments of Competitive Control

Platforms frequently use rankings to determine who receives visibility.

A ranking algorithm may measure:

• relevance;
• popularity;
• quality;
• engagement;
• conversion;
• seller performance.

But the platform may control the algorithm itself.

This creates a circular problem:

The platform determines the metric → the metric determines visibility → visibility determines performance → performance is then used to validate the metric.

This is a form of metric feedback dominance.

For example, if a platform gives its own product greater visibility, its product may obtain:

1. more impressions;
2. more sales;
3. better conversion statistics;
4. higher ranking;
5. even greater visibility.

The resulting performance data can then falsely appear to demonstrate that the platform's product deserves its superior ranking.

7. Six Major Case Laws

1. United Brands Co v Commission

Case 27/76, United Brands v Commission (1978)

The European Court of Justice examined United Brands' dominant position in the banana market.

The case is important because it demonstrates that market power cannot be determined through a single superficial indicator. The Court examined characteristics of the product, substitution possibilities, consumer preferences and competitive conditions.

Relevance

The case illustrates why controlled or selectively presented metrics should not automatically determine market power.

A dominant undertaking might rely upon:

• market-share statistics;
• price comparisons;
• sales figures;

while ignoring qualitative characteristics that constrain competition.

Principle

Quantitative indicators must be interpreted in their economic and competitive context.

2. Hoffmann-La Roche v Commission

Case 85/76, Hoffmann-La Roche v Commission (1979)

The Court established the classic definition of dominance as a position of economic strength enabling an undertaking to behave to an appreciable extent independently of competitors, customers and consumers.

Relevance to controlled metrics

A firm controlling the metrics by which competitors are evaluated may possess a form of informational or infrastructural power.

The important question is not merely:

"What does the metric say?"

but:

"Who controls the system that produces the metric?"

A dominant undertaking's ability to determine what information is visible can reinforce its ability to act independently.

3. British Airways plc v Commission

Case C-95/04 P, British Airways v Commission (2007)

The case concerned British Airways' incentive scheme for travel agents.

The Court confirmed that an undertaking with a dominant position has a special responsibility not to allow its conduct to impair genuine undistorted competition.

Relevance

Controlled metrics can become problematic where performance indicators are used to create exclusionary incentives.

For example, a platform could design seller-performance metrics that:

• reward dependence on its ecosystem;
• penalise multi-homing;
• favour its own distribution system;
• or make rival platforms appear inferior.

The metric then becomes a competitive instrument rather than a neutral measurement device.

4. Intel Corp v Commission

Case C-413/14 P, Intel v Commission (2017)

Intel is particularly important because the Court required detailed examination of the circumstances surrounding alleged exclusionary rebates.

The case emphasised the importance of assessing actual or potential foreclosure rather than relying on formalistic assumptions alone.

Relevance

The broader methodological lesson is crucial for controlled metrics.

A competition authority should examine:

• the actual economic mechanism;
• the coverage of the conduct;
• duration;
• conditions of competition;
• possible foreclosure;
• and the undertaking's strategy.

Thus, a dominant undertaking should not be able to establish the appearance of competitive neutrality merely by presenting favourable aggregate statistics.

5. Google Shopping

Case T-612/17, Google and Alphabet v Commission (2021)

The General Court examined Google's conduct concerning comparison-shopping services.

Google's general search results could significantly influence the visibility of competing services.

Relevance to controlled metrics

This case is highly relevant to algorithmic metrics because visibility itself can constitute an important competitive variable.

A ranking system may appear to be based upon neutral relevance criteria while its design or application disproportionately benefits the platform's own service.

The critical issue therefore becomes:

Who determines the metric that decides which competitor is visible?

If the dominant platform controls both:

1. the marketplace; and
2. the ranking mechanism,

the metric can become a mechanism for exclusion.

6. Slovak Telekom v Commission

Joined Cases C-165/19 P and C-166/19 P (2021)

The Slovak Telekom litigation concerned exclusionary conduct involving access to telecommunications infrastructure.

Relevance

The case demonstrates the importance of looking beyond nominal access.

A competitor may technically have access to infrastructure while the conditions, technical parameters, costs, or practical usability make meaningful competition difficult.

This provides an important analogy for controlled metrics.

A platform might claim:

"Competitors are allowed to participate."

But if it controls:

• API performance;
• data access;
• ranking;
• authentication;
• technical standards;
• interoperability;
• measurement methodology,

formal participation may coexist with substantive competitive exclusion.

7. Qualcomm v Commission

Case T-235/18, Qualcomm v Commission (2022)

The General Court examined Qualcomm's alleged exclusionary conduct involving payments to Apple.

The case reinforces the importance of analysing actual competitive effects and the economic circumstances surrounding allegedly exclusionary behaviour.

Relevance

Metrics such as:

• market share;
• prices;
• output;
• customer switching;

cannot necessarily establish competitive conditions by themselves.

Where the dominant undertaking controls critical inputs or commercial relationships, apparently favourable metrics may conceal competitive foreclosure.

8. Amazon Marketplace — European Commission

The European Commission's investigation into Amazon's use of marketplace seller data is another important illustration of the problem, although it concerns a regulatory investigation rather than a traditional reported judgment.

Amazon's dual role as:

1. marketplace operator; and
2. competing retailer

created concerns concerning its access to commercially sensitive information generated by independent sellers.

Relevance

The case illustrates informational asymmetry.

A platform can potentially possess more detailed performance information about competitors than the competitors themselves.

That creates a significant danger:

The undertaking controlling the metric may also possess the data necessary to optimise its own competitive strategy.

9. Competition Through Measurement Architecture

Controlled metrics can produce competition distortions through several mechanisms.

1. Metric selection

The dominant undertaking chooses what counts as success.

2. Metric weighting

Different variables receive different importance.

3. Data exclusion

Certain information is deliberately omitted.

4. Data aggregation

Competitive differences disappear inside averages.

5. Time-window manipulation

Short-term improvements are presented while long-term deterioration is ignored.

6. Benchmark manipulation

The comparison standard is strategically selected.

7. Algorithmic opacity

Competitors cannot determine how scores are generated.

8. Feedback effects

The metric itself changes the behaviour that it purports to measure.

10. Controlled Metrics and Artificial Contestability

A particularly important concept is artificial contestability.

A market may appear contestable because several firms technically operate within it.

However:

If the dominant undertaking controls the metrics determining access, ranking, reputation, visibility, or performance, the presence of multiple firms does not necessarily establish effective competition.

For example:

Five sellers

↓

Platform controls ranking

↓

Platform controls seller score

↓

Platform controls consumer visibility

↓

Platform controls transaction data

↓

Platform reports high seller participation

↓

Market appears competitive

But actual competitive constraints may be weak.

11. Metrics and Information Asymmetry

Controlled metrics generate a major information asymmetry.

The dominant platform knows:

• individual consumer behaviour;
• competitor conversion rates;
• search patterns;
• seller performance;
• pricing;
• churn;
• advertising effectiveness.

Competitors may receive only:

"Your ranking score is 72."

They cannot determine:

• the underlying dataset;
• weighting;
• benchmark;
• algorithm;
• error rate;
• counterfactual;
• or reason for the score.

This can create informational dependence.

12. Metric Manipulation and Self-Preferencing

Suppose a platform evaluates products according to:

"conversion probability."

If the platform's own products receive preferential placement, they may generate more transactions.

The platform then obtains superior conversion data.

It can subsequently claim:

"Our products rank higher because they objectively perform better."

This creates an endogenous competitive advantage.

The metric is no longer merely measuring competition.

It is producing the competitive outcome that it subsequently claims to measure.

13. Green Metrics and Competition

The problem also arises in sustainability markets.

A dominant firm might create:

• proprietary carbon scores;
• environmental rankings;
• sustainability certifications;
• ESG benchmarks.

If competitors must participate in the system to reach customers, the metric may become a gatekeeping mechanism.

Competition authorities may therefore need to examine:

• transparency;
• methodology;
• access;
• interoperability;
• discriminatory treatment;
• verification;
• governance.

14. AI-Generated Metrics

Artificial intelligence creates an even more sophisticated problem.

AI systems can generate metrics concerning:

• trust;
• reputation;
• risk;
• quality;
• fraud;
• consumer preference;
• creditworthiness;
• seller reliability;
• algorithmic relevance.

The difficulty is that AI-generated metrics may be:

• difficult to explain;
• continuously changing;
• trained on proprietary datasets;
• affected by feedback loops;
• difficult for competitors to reproduce.

A dominant AI platform can therefore possess metric-generation power in addition to traditional market power.

15. Competition Law Tests for Controlled Metrics

Competition authorities should ask at least eight questions.

1. Who designed the metric?

2. Who owns the underlying data?

3. Who controls the algorithm?

4. Can competitors independently reproduce the metric?

5. Can the metric be audited?

6. Does the metric influence market access?

7. Does the metric reinforce the dominant firm's position?

8. Does the metric accurately measure the competitive variable it purports to measure?

The last question is especially important.

A metric can be accurate mathematically but misleading economically.

16. Evidentiary Problems

Controlled metrics create difficult evidentiary questions.

A regulator may possess:

"Market share = 30%."

But that figure may depend upon:

• the market definition;
• data source;
• reporting methodology;
• treatment of multi-homing;
• treatment of internal transactions;
• geographic assumptions.

Therefore, the evidence should be subjected to metric provenance analysis.

Metric provenance means identifying:

Who → collected the data → how → using what methodology → under whose control → for what purpose?

17. Appropriate Remedies

Competition-law remedies may include:

A. Independent auditing

Require third-party verification of critical metrics.

B. Algorithmic transparency

Require disclosure of material ranking or scoring parameters.

C. Data access

Allow regulators or competitors access to appropriate underlying information.

D. Interoperability

Prevent metric systems from becoming closed ecosystems.

E. Non-discrimination

Require identical measurement standards for the platform and competitors.

F. Separation of functions

Where necessary, separate marketplace operation from competitive participation.

G. Standardised measurement

Use independently governed industry standards.

H. Monitoring trustees

An independent body may monitor compliance.

18. Key Doctrinal Principle

The central competition-law lesson is:

Competition should be evaluated by reference to actual competitive constraints, not merely by metrics selected or controlled by the undertaking whose market power is being assessed.

This does not mean that every proprietary metric is anticompetitive.

The legal concern arises where control over measurement is combined with:

• dominance;
• exclusionary incentives;
• opacity;
• discriminatory application;
• network effects;
• data advantages;
• self-preferencing; or
• barriers to independent verification.

Conclusion

Illusion of competition through controlled metrics represents a shift from traditional market power based on control over prices, output, or physical infrastructure toward control over the measurement architecture of markets.

The most important insight is that a market can be numerically competitive while structurally uncompetitive.

A dominant undertaking may control:

Data → Metric → Ranking → Visibility → Behaviour → Outcome → Reported Metric

creating a self-reinforcing cycle in which the undertaking's own measurement system becomes evidence that the market is competitive.

The cases such as United Brands, Hoffmann-La Roche, British Airways, Intel, Google Shopping and Slovak Telekom demonstrate, from different perspectives, why competition analysis must look beyond superficial indicators and examine the actual conditions, mechanisms and effects of competitive constraint.

In digital and AI markets, this principle becomes particularly significant because control over metrics can itself become a source of market power. The competition-law challenge is therefore not simply to ask "How competitive does the market look?", but rather:

"Who controls the system that tells us how competitive the market looks?"