Update: Software-based pricing and its EU competition law boundaries

RealPage: Hub-and-spoke risk

The U.S. DOJ has submitted a proposed settlement in the RealPage proceedings, likely the most prominent dispute to date concerning software-based algorithmic pricing. The case concerns several property managers and landlords who allegedly calculated their rental prices using RealPage’s software. The software was based on non-public, competitively sensitive information (CSI). According to the DOJ, the use of the software resulted in a hub-and-spoke cartel as the shared service provider's software, that was used by competitors, coordinated prices upward for all of them. It was not necessary for the competitors to communicate directly with each other.

Key elements of the proposed settlement

The settlement does not prohibit the software itself, but imposes extremely narrow guardrails:

• Ban on the use of CSI during runtime: RealPage may not use CSI during its “runtime operation.” This applies to both current and historical data.
  • In the real estate context, CSI includes all data that can be used to determine current or future availabilities, demand (e.g., vacancy information), pricing, including pricing formulas, and pricing strategies (e.g., rental concessions).
  • Why is RealPage also prohibited from using historical data? The DOJ likely wants to prevent the algorithm from “reverse-engineering” competitors’ current strategies based on past patterns. In this sector, even older data might increase transparency.
• Strict rules for AI training: For model training, RealPage may only use CSI that is at least twelve months old and does not refer to geographic areas smaller than nationwide data.
• Ban on nudging: In addition, RealPage is prohibited from, among others, the use of market survey data, joint supply/demand curves for competitors in the same region, and features that push users toward accepting price recommendations.
• External monitor: RealPage must submit to being monitored by a court-appointed external monitor.

EU perspective: Is there a safe harbor?

Many aspects of the proposed settlement reflect the general approach the European Commission currently applies to software‑based pricing cases. However, there are subtle differences that are crucial for companies operating in Europe.

• De-Risking CSI: EU competition law demands a high level of caution when pricing software utilizes CSI from competitors - particularly regarding pricing and capacity. To mitigate antitrust risks, companies should prioritize the use of neutralized information, such as anonymized, aggregated, or sufficiently historical datasets, rather than granular, real-time inputs.
  • In the EU, data older than 12 months is often considered “historic” and thus less problematic. However, competition authorities consistently emphasize the need for a case-by-case assessment. In a fast-moving market, even six-month-old data may be viewed as historic whereas in a market with long-term contracts, even 18-month-old data may still be viewed strategically valuable and thus CSI.
  • Like the U.S., EU competition law also views the exchange of sufficiently aggregated information less critical. This applies especially when data cannot - or only with difficulty - be attributed to individual companies, or when information is aggregated across different products, particularly if they belong to different markets. This, for the real estate sector, is the case for data referring to nationwide information rather than smaller geographic units.
• Training vs. Execution: Distinguishing between the (more market distant) training of an AI and its (market proximate) deployment is essential. The logic of the settlement is compelling: collusion risks arise primarily where AI can directly influence market behaviour.
• Caution wit "Off-the-Shelf" Industry Solutions: While utilizing publicly available data in pricing algorithms is generally permissible under EU competition law, companies must remain vigilant against tacit collusion. This can occur if a common software provider implements a standardized pricing or rebate formula across a specific sector. Before adopting any "industry-standard" tool, a rigorous antitrust assessment is required to ensure it does not facilitate horizontal coordination.

Enforcement ramp up in the EU

The ongoing RealPage proceedings symbolize the increased enforcement that can also be expected on this side of the Atlantic. Evidence includes not only authorities’ statements, such as confirmation by Deputy Director General of DG Competition, Linsey McCallum, in July 2025 that several EU level investigations into algorithmic pricing are underway. Testament is also the changing regulatory environment. Two regulatory heavyweights now flank traditional competition law.

• The AI Act: Since 1 August 2024, the EU AI Act has mandated a new level of transparency between regulators. Article 74(2) requires AI market surveillance authorities across all Member States to share competition-relevant cases with antitrust bodies and the European Commission on an annual basis. For businesses, this translates to a heightened risk of discovery: an investigation into an algorithm’s safety or transparency can now seamlessly trigger a parallel antitrust inquiry. Note that the European Commission has just published guidelines on AI system definition.
• DMA: According to many policymakers, generative AI services should in the future be classified as standalone core platform services. They may also already fall under existing DMA categories (such as online search engines, virtual assistants, or cloud computing). One year ago, the European Commission clarified in its Policy Brief “Competition in Generative AI and Virtual Worlds” that it views competition law and the DMA as complementary tools for generative AI.

Checklist for companies: How to secure your pricing software

To avoid compliance risks, companies should subject their pricing tools to an “antitrust compliance audit”:

1. Caution when using CSI: A software's price and rebate recommendations should not be based on non‑public competitor data.
2. Contract design with software providers: When procuring pricing software, contractual arrangements should be carefully reviewed for antitrust risks (e.g., clear provisions on access rights, restrictions on data use, and audit rights).
3. Compliance by design: It is advisable to review the software’s functionality. Can the software technically access and feed in competitors’ CSI? What is being advertised? Insist on implementing technical restrictions if needed. If granular and/or current competitor data are used for AI training, strict separation of data pools is essential.
4. Right to intervene: Avoid auto accept features in your pricing software.
5. Monitoring Recommended Retail Prices (RRP): In case your software engages with your distributors, antitrust compliance is crucial. Ensure in particular that "recommended" prices do not become mandatory in practice. If your software automatically monitors retailers and triggers "punishments" - such as automated warning letters or lower search rankings for deviating from the RRP - it may likely amount to price-fixing. The German Federal Cartel Office has just prohibited a major online platform from applying so-called price control mechanisms.

Conclusion: The RealPage settlement and the recent EU enforcement ramp-up send a clear message: the "black box" defence ("We did not know what the software was doing") does not work. If pricing practices are illegal when implemented offline, there is a high probability that they will also be illegal when implemented online. Proactive, continuous monitoring of pricing algorithms is no longer a nice-to-have; it is a fundamental pillar of modern corporate compliance.

Authored by Elena Wiese, Julian Urban, and Kyra Harmes.