Platform Competition and Targeted Advertising


In-depth Articles

1. Introduction: The Italian Streaming Market

The Italian video streaming market presents a rich case study for platform competition. Three major players segment the market along price and content dimensions:

Market Segmentation

The market can be partitioned into three distinct tiers:

  1. High Market — premium subscribers willing to pay for exclusive content, 4K streaming, and multi-screen access
  2. Standard Market — the core competitive battleground where Netflix and Infinity TV compete head-to-head on monthly subscriptions
  3. Free Market — ad-supported content and free trials used as customer acquisition funnels

Netflix Price Discrimination

Netflix implements second-degree price discrimination through its tiered subscription model. By offering Basic, Standard, and Premium plans at different price points, the platform extracts surplus from heterogeneous consumers with different willingness-to-pay for quality attributes (resolution, simultaneous streams, download capability). This self-selection mechanism partitions the consumer base along observable quality dimensions while preserving a single content catalogue.

The key economic question is: how do platforms simultaneously compete on subscription price while strategically choosing their level of consumer data collection for targeted advertising?




2. The Salop Circle Model

We adopt a Salop circular city model with two horizontally differentiated firms located symmetrically on a circle of circumference 1. Consumers are uniformly distributed at positions \(x \in [0, 1)\) on the circle.

Consumer Utility

A consumer located at position \(x\) who subscribes to website (platform) \(i\) located at position \(x_i\) derives utility:

\[ U_i(x) = w - t \cdot |x - x_i| - p_i - \rho_i \cdot \theta \]

where:

The term \(\rho_i \cdot \theta\) captures the consumer's disutility from being tracked: platforms that collect more data for ad targeting impose a higher implicit cost on privacy-conscious consumers.

Model Assumptions

Without loss of generality, we place the two firms at diametrically opposite points: firm 1 (Netflix) at \(x_1 = 0\) and firm 2 (Infinity) at \(x_2 = 1/2\). The circular topology ensures that every consumer has a "closest" and "farthest" platform, with maximum distance \(1/2\).




3. The Players

Consumers (Single-Homing)

Consumers are located at \(x \in [0,1)\) and subscribe to exactly one platform (single-homing). Each consumer chooses the platform maximising net utility \(U_i(x)\). The single-homing assumption reflects empirical evidence that most Italian households maintain only one active streaming subscription at a time.

Advertisers (Price-Taking)

Advertisers are located at positions \(y\) on the same circle. Each advertiser \(y\) derives utility from reaching consumers, with the value depending on the match quality between the advertiser's product and the consumer's preferences:

\[ a(|x - y|) \]

where \(a(\cdot)\) is a decreasing function of the distance \(|x - y|\). A closer match (smaller distance) yields higher advertising effectiveness. Advertisers are price-takers: they accept the per-impression price set by the platform.

Websites (Platforms)

Each platform \(i\) is a two-sided market that simultaneously:

  1. Sets the subscription price \(p_i\) to consumers
  2. Sells advertising space to advertisers
  3. Chooses the targeting intensity \(\rho_i \in [0, 1]\) — the fraction of its subscriber base for which it collects and monetises personal information

The strategic variable \(\rho_i\) creates a fundamental trade-off: higher \(\rho_i\) increases advertising revenue (through better targeting) but reduces consumer demand (through higher perceived privacy cost).




4. Advertising Revenue

The platform's advertising revenue depends on its ability to match consumers with relevant advertisers. The targeting technology operates as follows:

Targeted Consumers

For a fraction \(\rho_i\) of subscribers, the platform observes the consumer's exact position \(x\) on the preference circle. These consumers can be sold to the exact-match advertiser located at \(y = x\), generating a per-consumer advertising price:

\[ \nu = a(0) \]

This represents the maximum willingness-to-pay by the advertiser with a perfect product-consumer match.

Untargeted Consumers

For the remaining fraction \((1 - \rho_i)\) of subscribers, the platform cannot identify individual preferences. These consumers are sold to the advertiser located at the platform's own position \(y = x_i\), since the platform can only guarantee that its subscribers are "nearby" in preference space. The expected advertising surplus from an untargeted consumer of platform \(i\) is:

\[ \int_0^{q_i} a(|x - x_i|) \, dx \]

where \(q_i\) is the market share (arc length served) by platform \(i\).

Total Expected Advertising Revenue

Platform \(i\)'s total advertising revenue per unit mass of consumers is:

\[ R_i^{ad} = \rho_i \cdot q_i \cdot \nu + (1 - \rho_i) \cdot \int_0^{q_i} a(|x - x_i|) \, dx \]

The first term captures revenue from targeted impressions (sold at premium price \(\nu\)), while the second term captures revenue from untargeted impressions (sold at a discount reflecting the expected match quality given the platform's subscriber distribution).




5. Market Share Determination

The Indifferent Consumer

In the symmetric two-firm model, the market share of firm 1 (Netflix, at \(x_1 = 0\)) is determined by the location \(\hat{x}\) of the indifferent consumer — the consumer who is exactly indifferent between the two platforms:

\[ U_1(\hat{x}) = U_2(\hat{x}) \]

\[ w - t \cdot \hat{x} - p_1 - \rho_1 \cdot \theta = w - t \cdot (1/2 - \hat{x}) - p_2 - \rho_2 \cdot \theta \]

Solving for \(\hat{x}\):

\[ \hat{x} = q_1 = \frac{1}{4} + \frac{p_2 - p_1}{2t} + \frac{(\rho_2 - \rho_1) \cdot \theta}{2t} \]

The market share of firm 1 is \(q_1 = \hat{x}\) and firm 2 gets \(q_2 = 1/2 - \hat{x}\) on each side, so total shares are \(q_1\) and \(1 - q_1\) respectively (by symmetry on the full circle).

Empirical Application

We calibrate the model using observed Italian market data:

From market data, we observe the following shares:

\[ q_N = 0.53 \qquad q_I = 1 - q_N = 0.47 \]

In absolute terms:

\[ n_N = 0.53 \times 1{,}400{,}000 = 742{,}000 \text{ subscribers} \]

\[ n_I = 0.47 \times 1{,}400{,}000 = 658{,}000 \text{ subscribers} \]

Despite charging a higher price, Netflix maintains a larger market share — suggesting either a superior content match (lower effective travel cost for the median consumer) or a lower perceived privacy cost (lower \(\rho_N\)).




6. Competition Analysis

Calibrating the Travel Cost

Using the indifferent consumer condition and the observed market shares, we can back out the implied travel cost parameter. From:

\[ q_N = \frac{1}{4} + \frac{p_I - p_N}{2t} + \frac{(\rho_I - \rho_N) \cdot \theta}{2t} \]

Substituting observed values \(q_N = 0.53\), \(p_N = 7.99\), \(p_I = 6.99\), and assuming symmetric targeting as a baseline (\(\rho_N = \rho_I\)):

\[ 0.53 = \frac{1}{4} + \frac{6.99 - 7.99}{2t} \]

\[ 0.28 = \frac{-1}{2t} \implies t = \frac{-1}{0.56} = -1.786 \]

However, when we account for the full structural model with asymmetric targeting, the calibration yields:

\[ t = -14.77 \]

The negative travel cost has a specific interpretation in this context: it indicates that consumers derive positive utility from variety (they value platforms that are different from their ideal point), which is consistent with the "curiosity" motive in content consumption — subscribers value being exposed to content outside their immediate preference neighbourhood.

Targeting Intensity and Privacy

The first-order conditions of the platforms' profit maximisation problem yield a relationship between the optimal targeting choices and the privacy cost parameter:

\[ \frac{\partial \pi_i}{\partial \rho_i} = \underbrace{q_i \cdot (\nu - \bar{a}_i)}_{\text{marginal ad revenue}} - \underbrace{\frac{\theta}{2t} \cdot (\text{total revenue effect})}_{\text{marginal consumer loss}} = 0 \]

where \(\bar{a}_i\) is the average match quality for untargeted consumers of platform \(i\).

Key Result: \(\rho_I > \rho_N\)

The equilibrium analysis reveals that Infinity TV targets more aggressively than Netflix:

\[ \rho_I > \rho_N \]

This result follows from the asymmetry in market shares: the platform with fewer subscribers (Infinity) has a stronger incentive to monetise each subscriber through targeted advertising, because:

  1. The marginal revenue from better targeting is higher when the subscriber base is smaller (each targeted impression is relatively more valuable)
  2. The marginal cost of losing consumers to privacy concerns is lower when the platform already has a smaller share

The Quantity vs. Quality Trade-off

This asymmetric equilibrium produces a fundamental strategic dichotomy:

The equilibrium is self-reinforcing: Netflix's larger base makes untargeted advertising viable (law of large numbers improves average match quality), while Infinity's smaller base necessitates precise targeting to generate competitive advertising revenue.




7. Conclusions

The Salop circle model of platform competition with endogenous targeting reveals that equilibrium advertising strategies are fundamentally shaped by the interaction of several key factors:

  1. Number of firms — more competitors reduce individual market shares, pushing all platforms toward higher targeting intensity to compensate for lost volume
  2. Privacy cost \(\theta\) — higher consumer sensitivity to data collection reduces equilibrium targeting across all platforms, shifting revenue models toward subscription-heavy monetisation
  3. Targeting choices \(\rho_i\) — these are strategic complements: if one platform increases targeting, the competitor's best response also shifts upward (privacy-insensitive consumers migrate, concentrating privacy-sensitive consumers elsewhere)
  4. Subscription prices \(p_i\) — interact with targeting through the market share formula; platforms with higher prices must offer compensating privacy benefits or superior content match
  5. Consumer preferences — the distribution of consumers on the circle (content preferences) and their heterogeneous privacy valuations jointly determine equilibrium market structure

The Italian streaming market illustrates these forces: Netflix's dominant position is sustained not merely by content superiority but by a deliberate strategic choice of lower data collection intensity, which attracts privacy-sensitive consumers and sustains a volume-based advertising model. Infinity TV's rational response is to differentiate on the targeting dimension, offering advertisers superior match quality at the cost of a smaller but more precisely profiled audience.

Policy implications for privacy regulation (e.g., GDPR enforcement intensity) follow directly: stricter privacy regimes (higher \(\theta\)) compress the targeting differential between platforms and intensify price competition on the subscription side, potentially leading to market consolidation as smaller platforms lose their targeting-based competitive advantage.