Key points

• Pricing software that draws on competitors’ data, or that many competitors use, can raise antitrust risk even when no one agrees on a price.
• Enforcers in the United States, the United Kingdom and the European Union have all said that using an algorithm does not excuse a conspiracy, and have focused on vendors and shared data pools.
• The main risks are agreements facilitated by a common vendor, exchanges of competitively sensitive information, and software that is configured to follow rivals.
• Several US states and cities have enacted laws or advanced bills aimed at algorithmic pricing, mostly in housing, and more are likely.
• Retailers should map their pricing tools, review vendor contracts, limit use of non-public competitor data and train pricing teams on what can be shared.

Why this is on enforcers’ agenda

Retailers have used dynamic pricing for years. What has changed is scale and sophistication. Software can scrape rivals’ prices in real time, recommend changes in minutes and, in some cases, apply them automatically. Machine learning tools can optimize margins using pooled data from many customers. The same features that make these tools useful create antitrust questions, and enforcement agencies have said publicly that they are paying attention.

The United States Department of Justice and the Federal Trade Commission have stated that competition law applies to pricing algorithms in the same way as to any other means of coordination. They have supported private and public cases involving software that uses non-public data from competing landlords to recommend rents, and have filed statements of interest in private cases. The theory has been tested mostly in housing, but it is not limited to it.

The legal framework

In the United States, Section 1 of the Sherman Act prohibits agreements that unreasonably restrain trade. An agreement can be inferred from circumstantial evidence, including the use of a common vendor that collects competitors’ non-public data and returns recommendations, particularly where adoption of the recommendations is high and competitors know that others are using the same tool. Courts have reached different results on how much is needed, and the cases are still developing.

In the EU and the UK, the prohibitions on anticompetitive agreements are found in Article 101 of the Treaty on the Functioning of the European Union and the Chapter I prohibition of the Competition Act 1998. Both prohibit agreements and concerted practices, and exchanges of competitively sensitive information can be an infringement in themselves, if they reduce uncertainty about rivals’ future conduct. The European Commission’s guidance on horizontal agreements addresses information exchange explicitly, and the UK CMA has published research on the risks of algorithmic pricing and on its approach to enforcement.

Three risk patterns

The first is the hub-and-spoke pattern, in which competitors each feed data to a common vendor and rely on its recommendations. The vendor is the hub, and the competitors may be treated as parties to an agreement if they know others are doing the same. The risk is highest where the tool uses non-public, current data, and where the vendor markets itself on the ability to lift prices across the market.

The second is the direct or indirect exchange of competitively sensitive information. Sharing current or future prices, promotions or margins with a competitor, through a trade association, a benchmarking service or a data platform, can be unlawful without any agreement on price. Aggregated, historical, anonymized data is far less risky than disaggregated, recent or forward-looking data.

The third is unilateral conduct that has the effect of tacit coordination. Setting up software to match every competitor price change within minutes is not itself unlawful in most settings, but it may look different where a firm has market power, where there are signals between competitors or where an internal document speaks of keeping prices high or avoiding price wars.

Statutes and state action

States and cities have begun to legislate. Measures have been enacted or proposed to restrict the use of pricing algorithms that use competitors’ non-public data, with an initial focus on residential rental markets. California has legislated on the use or distribution of common pricing algorithms in ways that may amount to coercion or an agreement, and other states have bills pending. Separately, the Federal Trade Commission has examined so-called surveillance pricing, which tailors prices to individual consumers using personal data, as a consumer protection question. Retailers using personalized pricing should therefore keep an eye on both antitrust and consumer protection law.

Compliance steps for retailers

Map the pricing and promotion tools in use across the business, including those embedded in e-commerce platforms, marketplaces and category management services. Identify what data goes in and where it comes from.

Review vendor agreements. Find out whether the vendor pools customer data, whether competing retailers are customers, and how recommendations are generated. Seek representations that the tool does not use competitors’ non-public data, and rights to audit and to turn off features.

Set a policy on competitor data. Public price monitoring is generally lawful. Use of non-public information received from a competitor or through an intermediary should be escalated to legal. Make it clear to commercial teams that conversations with rivals about future pricing are off limits, whether in person, in messages or through a shared tool.

Preserve human oversight. Document who can override the algorithm, who reviews the outputs and how objectives are set. Avoid wording in internal documents that suggests a goal of stabilizing the market or following a price leader.

Train and test. Pricing, category and data science teams should receive short, practical training, and the company should audit how the tools work in practice at least annually.

Looking ahead

Expect more enforcement and private litigation, particularly where plaintiffs can show a common vendor and a similar pattern of price moves. The legal standards will emerge slowly, which makes sound process and documentation the best protection.

Priya Raghunathan in London advises retailers, brands and technology providers on competition law, algorithmic pricing and information exchange compliance, and defends clients in investigations and litigation.