Platform pricing

Canadian experience

In July 2025, a document entitled¹ “Algorithmic Pricing and Competition” was discussed. Algorithmic pricing is the process of using automated algorithms to set or recommend prices for goods and services, often in real time, based on a set of input data.

The data can be obtained from consumers (online behavior, demographic information, transaction history) or may contain information about market conditions (supply and demand, competitors' prices, inventory levels). With the development of AI, it has become possible to continuously train algorithms based on data, especially if the data is constantly² updated (“reinforcement learning” ). However, the decision-making process of such algorithms is often opaque and difficult to understand (the “black box” problem). Therefore, human oversight is important to control such algorithms.

Depending on the type of data, there are different types of algorithms:

1) Dynamic pricing algorithms – setting prices based on market conditions (demand, supply, competitors' prices, etc.). The main goal of such an algorithm is to maximize the company's profit.

2) Personalized (or controlling) pricing algorithms – setting prices based on the personal characteristics of an individual or group of individuals. The main goal is to determine consumers' willingness to pay in order to maximize profit.

This leads to price discrimination, where a company charges different prices for the same product or service depending on a customer's willingness to pay. There are several degrees of price discrimination:

1) First degree: the company sets the exact price that the consumer is willing to pay (i.e., personalized pricing).

2) Second degree – setting different prices depending on the terms of sale (e.g., price reduction for bulk purchases, subscription plans).

3) Third degree – setting different prices for different consumer groups depending on age, location, and other characteristics (e.g., price reductions for students or seniors).

Algorithmic pricing leads to anticompetitive practices:

1) Cooperation between competitors for price fixing, market division, etc. For example, companies enter into a hub-and-spoke agreement — several companies use the same algorithm (identical software) that processes data provided by competitors and sets prices simultaneously for all competitors. If there is direct interaction between competitors, this is explicit collusion; if there is no interaction, tacit collusion is possible.

2) Use of anti-competitive practices involving algorithms. For example, in predatory pricing, when a dominant company deliberately lowers its price below cost in order to drive competitors out of the market (predatory phase) and raises prices after competitors leave, compensating for the price reduction (recovery phase).

Companies use algorithms to target customers: they identify customers who are most likely to switch to another seller in order to retain customers with low prices.

French experience

In July 2025, the Shein marketplace was fined €40 million for price fraud: it advertised³ “crossed-out prices” (defined as “discounts”). However, in 57% of cases, there was no price reduction, in 19% of cases, the discount offered was less than promised, and in 11% of cases, prices were found to have increased. At the same time, the French Consumer Rights Code stipulates that when posting information about price reductions, it is necessary to indicate the minimum price at which the product was sold during the 30 days preceding the promotion. This price is the reference price from which the discount should be calculated. There is no such rule in Russia.

Russia’s experience

In Russia, the use of algorithms to implement anti-competitive agreements, including for pricing, particularly in the context of an explicit collusion, is recognized as an aggravating circumstance.

From the monitoring issue No. 7 (19), July 2025. Download the full issue (PDF) · issue page at the Gaidar Institute