Practical Takeaways From the DOJ’s Algorithmic Pricing Settlement

• A recent settlement provides some insight into the United States Department of Justice’s (DOJ) position on the use of competitors’ information in common or shared pricing algorithms. The use of such information can raise antitrust issues depending on the type of information used and how the information is used.
• The terms of the settlement seem to suggest that, according to the DOJ:
  • The antitrust risk is highest when current or forward-looking, non-public, competitively sensitive information of competitors is used.
  • The risk is particularly acute when this information is used in the runtime operation of a pricing algorithm and the output is used in pricing decisions.
  • Another element of risk is added if pricing recommendations are automatically implemented without an opportunity for a user to choose to override the suggested price, reflecting an independent decision.
• On the other hand, at least in these circumstances, the DOJ is apparently willing to accept the use of historical information to train an algorithm, provided the information is at least 12 months old and does not include price data or geographically localized data.

Background

On November 25, 2025, the DOJ announced that it agreed to settle claims that RealPage, Inc. (RealPage) violated Section 1 of the Sherman Act by unlawfully sharing competing users’ competitively sensitive information for use in competitors’ pricing and by entering into agreements with users to align pricing. Several states are also plaintiffs in the action. The proposed settlement does not resolve the states claims against RealPage or any claims against other defendants. If approved by the court pursuant to the Tunney Act, an order of final judgment would be entered requiring several changes to RealPage’s business. This is the third settlement in this case, following prior settlements between the DOJ and two property managers and landlords. However, this settlement is the first of its kind with a pricing algorithm provider and provides the clearest indication yet of what is and is not acceptable to DOJ.

The settlement indicates that DOJ is concerned with use of non-public data for both runtime operations (including generating price recommendations) and model training, although its concerns with runtime operations are more severe. For model training, DOJ appears willing to accept some use of non-public, competitively sensitive information, but not the most sensitive categories of data (current data, price data, and geographically localized data). The settlement would also restrict product features that bypass user preferences or favor price increases over price decreases.

The Government Allegations

According to the DOJ’s complaint, “RealPage sells software to landlords that collects non-public information from competing landlords and uses that combined information to make pricing recommendations.” The complaint further alleged that:

• Users agreed with RealPage to provide RealPage with “current, forward-looking, granular, and highly competitively sensitive information . . . on effective rents, rent discounts, occupancy rates, availability, lease dates, lease terms, unit amenities, and unit layouts.”
• Users knew that these data were “pooled” and used to train RealPage’s revenue management algorithms, which used the data to “generate pricing recommendations for the landlord and its competitors.”
• RealPage monitored and attempted to enforce compliance by users to price recommendations.
• Thus, rather than setting prices independently and in competition with one another, users were setting prices based on the competitively sensitive information of competitors.

The DOJ opened a civil investigation into RealPage in November 2022. In March 2024, the DOJ reportedly opened a criminal investigation to investigate whether RealPage was facilitating price fixing, which it later closed with no action. In August 2024, the DOJ and eight states filed a civil complaint against RealPage and several landlords. The complaint alleges that the defendants’ “joining together . . . deprives the market of fully independent centers of decision-making on pricing” and thus is a “combination . . . in restraint of trade” that violates Section 1 of the Sherman Act. The complaint also alleges that RealPage violated Section 2 of the Sherman Act by monopolizing and attempting to monopolize the commercial revenue management software market. Multiple defendants moved to dismiss, but the court has not yet ruled on those motions. There is also parallel private litigation.

The DOJ’s Proposed Settlement with RealPage

Restrictions Apply Primarily to Use of Non-Public, Competitively Sensitive Information. The settlement’s restrictions generally apply only to RealPage’s combined use of unaffiliated users’ non-public, competitively sensitive information.

This is any information that is not “readily accessible to the general public” relating to users’ settings or parameters; a user’s rental pricing amount, formula, strategy, or discounts; or other current or historical information that could reasonably be used to determine current or future rental supply, demand, or pricing, either individually or in combination with other data. It does not include information available on a website or posted in a building or that a landlord would provide to any person who “reasonably presents himself as a prospective renter.”

The settlement distinguishes between non-public information from a particular user and information from unaffiliated properties. Most restrictions in the agreement apply to RealPage’s use of non-public information from unaffiliated properties, but not non-public information from the user’s own properties.

In one instance, the DOJ seems to limit use of even public data or data from other properties the user owns to make recommendations. RealPage’s “sold out mode,” which automatically increases recommended prices once a property reaches target occupancy, cannot use “anything other than the [s]ubject [p]roperty’s own information.”

Distinction Between Model Training and Runtime Operation. The settlement imposes restrictions on RealPage’s use of non-public information both for the purposes of training its pricing algorithm models and for use in runtime operations to generate recommendations for users. However, different restrictions apply to each. In general, the settlement demonstrates a higher level of concern with use in runtime operations.

Runtime Operations Restrictions. Non-public, competitively sensitive information may not be used in runtime operations of an algorithm. That is, non-public, competitively sensitive information from one user cannot be used as an input into the algorithm as it generates pricing recommendations for another user. For example, the settlement would not allow RealPage to design an algorithm that adjusts pricing recommendations for Property A based on non-public occupancy statistics of nearby competitors Property B and Property C.

Model Training Restrictions. Non-public, competitively sensitive information may not be used to train a pricing recommendations model unless it is at least 12 months old. Price data and geographically localized data are singled out as especially sensitive—RealPage is largely banned from using such information in model data, even if the information is more than 12 months old. This feature of the settlement suggests that DOJ views competitively sensitive information as falling along a spectrum of most to least sensitive, rather than viewing competitively sensitive information as a binary category, which is consistent with DOJ’s position in other recent cases.^[1]

Limitations on Product Features. The settlement limits RealPage’s use of certain product features that bypass user input or favor price increases over decreases. For example, RealPage is permitted to offer an “Auto Accept” feature that automatically implements recommended price increases or decreases within a certain range, but must require the user to manually set range rather than having it on by default. Similarly, the DOJ required changes to RealPage’s “Governor” feature, which was alleged to favor price increases over price decreases; the settlement requires RealPage to make the feature symmetrical with respect to increases and decreases.

RealPage Challenge to New York Legislation Regulating Use of Pricing Algorithms by Landlords

Separate from the ongoing litigation, there are new laws seeking to regulate the use of pricing algorithms in multiple states, including New York.

New York Governor Kathleen C. Hochul recently signed Assembly Bill 1417-B, adding provisions targeting the use of algorithms in residential rent-setting to the Donnelly Act, the state’s antitrust law (what RealPage calls the “Rent Advice Statute”). New York’s Donnelly Act prohibits “agreements for monopoly or in restraint of trade” in general. The new law, which will become effective as of December 15, 2025, makes explicit that certain conduct by landlords in setting rents and other lease terms for “residential dwelling units” could give rise to Donnelly Act violations and that third-party entities that provide certain types of algorithms or otherwise facilitate an anticompetitive agreement among landlords also risk liability under the Donnelly Act. California also recently enacted amendments to its antitrust law, the Cartwright Act, to broadly regulate the use of common pricing algorithms that affect not just prices but also “commercial terms.”

On November 25, 2025, RealPage filed a complaint in the Southern District of New York against New York State Attorney General Letitia James challenging the Rent Advice Statute as a violation of the First Amendment to the Constitution, both facially and as applied to RealPage and its customers. RealPage seeks a judgment declaring that the Rent Advice Statute violates the First Amendment and the Due Process Clause; a preliminary, and ultimately permanent, injunction barring the State from enforcing the Rent Advice Statute or causing it to be enforced; and any relief the court deems just and proper, including reasonable attorneys’ fees and the costs of this action.

* * *

[1] See, e.g., Statement of Interest of the United States 3, 20, In re Pork Antitrust Litig., No: 0:18-cv-01776 (D. Minn. Oct. 1, 2024) (“Ultimately, the antitrust laws prohibit information sharing among competitors whenever such exchanges tend to harm competition. Despite defendants’ suggestions to the contrary, courts do not apply bright-line rules in making this inquiry; they look to the full circumstances to gauge anticompetitive potential . . . . The precedents . . . make clear that the rule of reason is intended to be a flexible inquiry—one not conductive to rigid and arbitrary rules.”).