Operationalizing Personality-Matched Real Estate Systems: A Comprehensive Legal, Ethical, and Structural Framework

Executive Summary

The residential real estate industry stands at a pivotal juncture where technological capability intersects with increasing regulatory scrutiny. The proposed concept—a personality-matched real estate agent system that functions similarly to a “dating app” for buyers and sellers—represents a potentially disruptive innovation in a market traditionally reliant on geographic farming and sphere-of-influence marketing. By aligning the communication styles, working paces, and behavioral preferences of clients and professionals, brokerages can theoretically increase conversion rates, client satisfaction, and agent retention. However, the transition from a conceptual framework to a fully operational, revenue-generating system requires navigating a labyrinth of federal and state regulations that are currently in a state of active evolution.

The convergence of the Fair Housing Act (FHA), the Real Estate Settlement Procedures Act (RESPA), rigorous Independent Contractor (IC) labor standards, and emerging consumer data privacy statutes—specifically the Minnesota Consumer Data Privacy Act (MCDPA)—creates a high-stakes compliance environment. A system that algorithmically sorts consumers and agents carries the inherent, though often unintentional, risk of “steering,” a discriminatory practice strictly prohibited under the FHA. Furthermore, the operational mechanics of assigning leads based on behavioral profiles can threaten the tax-advantaged independent contractor status of agents if not structured as a voluntary “referral network” rather than a mandatory directive.

This report provides an exhaustive, expert-level analysis of the legal, ethical, and operational infrastructure required to launch a personality-matched real estate system immediately. It details the necessary legal safeguards against disparate impact liability, the structural requirements to maintain tax-compliant worker classification, and the technological workflows needed to automate psychometric profiling while adhering to strict data privacy standards. The analysis draws upon federal case law, Minnesota state statutes, and psychological research to provide a roadmap that is not only legally defensible but operationally robust.

Part 1: The Legal Landscape – Fair Housing, Algorithmic Bias, and the Doctrine of Steering

The most critical legal hurdle for any system that sorts, filters, or matches housing consumers is the Fair Housing Act (FHA). While “personality” is not explicitly listed as a protected class under federal law, the outcome of personality matching can create significant liability if it results in disparate impact or steering. The digitization of real estate processes has shifted the scrutiny from physical steering to algorithmic steering, where the bias is encoded in the logic of the recommendation engine rather than the overt actions of a human agent.

1.1 The Doctrine of Steering and the “Newsday” Precedent

Under the Fair Housing Act, it is illegal to “steer” prospective homebuyers toward or away from specific neighborhoods or housing opportunities based on race, color, religion, sex, handicap, familial status, or national origin.¹ Traditionally, steering was a physical act—an agent driving a client only to neighborhoods where people of their same race lived. However, recent investigations have shown that steering remains pervasive and has evolved. A landmark three-year investigation by Newsday, which utilized 145 matched-pair tests, revealed an 87% rate of racial steering, where agents provided different listings and service levels to minority testers compared to their white counterparts.³

In the context of a personality matching system, the risk is that the algorithm inadvertently replicates these patterns. If the system uses variables that correlate with protected characteristics—such as zip code preferences, language complexity, or “cultural fit”—it may automatically segregate the market. For instance, if agents who score high on “Dominance” (D-style) are predominantly white males within a specific brokerage, and the algorithm matches them exclusively with high-net-worth buyers who also test as “Dominant,” the system may be systematically excluding minority agents from lucrative leads or steering minority buyers toward a different, potentially lower tier of service.³ This form of “digital redlining” or “algorithmic steering” restricts the consumer’s choice and limits their access to the full marketplace, which is a direct violation of the FHA’s mandate to provide equal professional service.

1.2 Disparate Impact Liability and the “Inclusive Communities” Standard

The legal theory of “disparate impact” is the primary mechanism by which algorithmic bias is prosecuted. The U.S. Supreme Court’s 2015 decision in Texas Dept. of Housing & Community Affairs v. Inclusive Communities Project, Inc. affirmed that disparate impact claims are cognizable under the Fair Housing Act.¹ This ruling established that a policy or algorithm can be deemed illegal even if there is no intent to discriminate, provided it has a disproportionately adverse effect on a protected group.

For a personality matching system, this means the brokerage cannot simply claim “the algorithm is colorblind.” If the output of the matching process shows that Black buyers are disproportionately matched with junior agents, or that families with children are systematically filtered away from “quiet” agents who specialize in luxury condos, the brokerage faces liability. The burden of proof shifts to the housing provider to demonstrate that the practice is necessary to achieve a “substantial, legitimate, nondiscriminatory interest” and that there is no less discriminatory alternative available to achieve that interest.¹

1.2.1 The Danger of Proxy Variables

While personality traits like “introversion” or “conscientiousness” are not protected classes, they often function as proxies for protected characteristics due to cultural and socioeconomic factors.

• Cultural Bias in Testing: Standardized personality tests, including DISC and Myers-Briggs, can carry inherent cultural biases. Research indicates that certain ethnic groups may score differently on scales of “assertiveness” or “individualism” due to cultural norms regarding respect for authority or communal decision-making, rather than actual competency or preference.⁵ If the algorithm interprets a culturally deferential communication style as “low motivation” or “indecisive,” it may unfairly downgrade the lead quality.
• Geographic Proxies: If the matching algorithm intakes data points such as “preferred school district” or “current zip code” to infer personality or lifestyle—for example, categorizing a buyer as “The Explorer” based on their interest in rural areas—it risks redlining. Historical data confirms that real estate agents have frequently used school ratings as a proxy for race to steer clients, and codifying this into software amplifies the discrimination.³
• Name and Language Analysis: Studies have shown that applicants with “Arabic-sounding names” or African American vernacular names receive significantly fewer responses in the housing market.⁶ If the system uses Natural Language Processing (NLP) to analyze the sentiment or “personality” of a user’s typed messages, it risks detecting dialects or linguistic patterns associated with race or national origin, leading to biased routing.⁷

1.3 Safeguarding the Matching Algorithm

To legally operate a personality matching system, the algorithm must be constructed to be “blind” to protected class characteristics and rigorously tested for adverse impact prior to and during its deployment.

Compliance Protocol for Algorithmic Safety:

1. Variable Isolation: The system architecture must ensure that the matching engine only processes psychometric data points (e.g., “Do you prefer email or phone calls?”, “Do you want data sheets or narrative stories?”) and strictly excludes demographic data (race, gender, age, family status) from the matching logic. While demographic data may be collected for HUD compliance or identity verification, it must be “firewalled” from the decisioning engine.
2. Adverse Impact Simulations: Before the system goes live, the brokerage must run statistical simulations using historical or synthetic data. If the system matches Black buyers to Black agents at a rate statistically higher than random distribution, or if it routes women disproportionately to lower-performing agents, the algorithm is likely “reading” a proxy variable and must be recalibrated immediately.⁷
3. The “Business Necessity” Documentation: The brokerage must explicitly document the valid business reason for matching specific traits. For example, “Matching high-compliance (C-style) buyers with high-compliance agents is necessary to reduce contract errors and cancellation rates, which serves the consumer’s interest in a successful transaction.” This documentation creates the foundation for the legitimate business necessity defense required by Inclusive Communities.¹

Part 2: The EEOC and the Risks of Personality Testing

While the Fair Housing Act governs the relationship with the consumer, the Equal Employment Opportunity Commission (EEOC) regulations govern the relationship with the agents. Using personality tests as a gatekeeping mechanism for lead distribution introduces risks under Title VII of the Civil Rights Act of 1964.

2.1 The CVS and Best Buy Precedents

Recent enforcement actions by the EEOC against major corporations like CVS Caremark and Best Buy highlight the dangers of using personality assessments in workforce management. In these cases, the EEOC contended that the personality tests used for hiring and screening had a disparate impact based on race and national origin.⁸ The claim was not that the companies intended to discriminate, but that the tests themselves resulted in lower scores for certain minority groups, effectively barring them from employment opportunities.

Although real estate agents are typically independent contractors (a distinction discussed in Part 3), the distribution of leads is a fundamental economic benefit of the role. If a brokerage uses a personality test to determine which agents are eligible for “premium” leads, and that test disproportionately filters out minority agents, the brokerage could face claims of discrimination in the “terms, conditions, and privileges” of the contractual relationship.⁹

2.2 Validating Psychometric Tools

To mitigate this risk, the personality assessment used must be validated for the specific context of real estate sales.

• Criterion-Related Validity: The brokerage must demonstrate a statistical correlation between the test scores and job performance (e.g., closing ratio, client satisfaction). If the test cannot predict success, using it to deny leads is legally indefensible.¹¹
• Adverse Impact Analysis: The test vendor (e.g., Everything DiSC, Crystal Knows) should provide a technical manual certifying that their assessment does not produce adverse impact against protected groups. “Home-grown” quizzes created by the marketing department are highly dangerous because they lack this psychometric validation.¹²
• DISC vs. MBTI: The DISC model is generally preferred in professional sales environments over the Myers-Briggs Type Indicator (MBTI). DISC measures observable behavior and communication styles, which are mutable and trainable, whereas MBTI measures inherent cognitive processing. DISC’s focus on “behavior” rather than “identity” makes it safer for determining “business fit” without encroaching on protected personal characteristics.¹³

2.3 Medical Examinations and the ADA

Under the Americans with Disabilities Act (ADA), employers cannot require medical examinations prior to a job offer. Some personality tests, such as the Minnesota Multiphasic Personality Inventory (MMPI), are designed to diagnose mental health conditions and are classified as medical exams. The use of such clinical tests in a real estate matching system would be a violation of the ADA.⁹ The system must strictly utilize normative personality assessments (measuring normal variations in behavior) rather than clinical assessments (measuring pathology).

Part 3: Worker Classification – The Independent Contractor Paradox

Real estate agents are predominantly classified as Independent Contractors (ICs) under Section 3508 of the Internal Revenue Code. This classification is vital for the brokerage business model, as it avoids payroll taxes, unemployment insurance, and benefits. However, IC status strictly limits the amount of behavioral control a brokerage can exert over its agents. This creates a central paradox for a personality-matched system: How do you ensure the agent follows the specific “personality script” required for the match without treating them like an employee?

3.1 The IRS Common Law Control Test

The IRS determines worker status using a multi-factor “Common Law” test, which evaluates three primary categories: Behavioral Control, Financial Control, and Type of Relationship.¹⁶

• Behavioral Control: This is the most dangerous factor for the proposed system. If a brokerage gives instructions on how to do the work—for example, “You must use this specific script for D-style clients” or “You are required to attend this personality training on Tuesday”—this constitutes behavioral control and indicates an employment relationship.¹⁸
• The Risk of Reclassification: If a mandatory personality matching system forces agents to take leads and strictly dictates their interaction style, it could trigger a reclassification audit. If agents are reclassified as employees, the brokerage would be liable for back taxes, unpaid overtime, and workers’ compensation premiums, which could bankrupt the firm.¹⁹

3.2 Structuring the System for IC Compliance

To preserve IC status while ensuring the integrity of the personality match, the system must be structured as optional and results-oriented, rather than mandatory and process-oriented.

3.2.1 The “Opt-In” Referral Team Model

Instead of imposing the system on the entire agent roster, the brokerage should create a specialized, voluntary “Referral Team” or “Lead Network”.²⁰

• Voluntary Participation: Agents must actively apply to join this network. The core IC agreement remains the baseline, and the “Lead Team Agreement” acts as an addendum. Because the agent chooses to join the program to access a specific benefit (leads), the requirements of the program are viewed as contractual terms of that specific benefit, not general employment controls.
• The “Referral Agreement” Mechanism: When an agent joins the network, they sign an agreement specifically regarding the leads generated by the network. The logic is: “You are free to run your own business how you see fit (preserving IC status). However, for these specific leads that we are referring to you, you agree to service them according to our ‘Client Experience Standards’ to maintain your eligibility.” This shifts the control from the worker to the lead source.²¹

3.2.2 “Method of Compensation” vs. “Method of Work”

State laws, including Minnesota’s, often allow brokers to supervise ICs regarding compliance with laws (Fair Housing, License Law) but strictly limit supervision over the manner of sales.

• Training as a Resource: Personality training (DISC certification) should be offered as a “value-add” resource rather than a mandate. The messaging should be: “Agents who complete this certification convert 30% more leads.” This encourages compliance through self-interest rather than coercion.²³
• Outcome-Based Management: Management of the system should focus on conversion metrics (the result), not on whether the agent used a specific script (the method).¹⁸ If an agent fails to convert personality-matched leads, the consequence should be commercial (they stop receiving leads) rather than disciplinary (firing or formal reprimand), distinguishing the relationship from employment.

3.3 The Written Independent Contractor Agreement

A robust Independent Contractor Agreement (ICA) is the first line of defense. The agreement must explicitly state that the agent retains the right to reject leads and determine their own schedule.

• Drafting Language: “Agent is an independent contractor and shall determine their own methods, schedule, and manner of performing services. However, if Agent elects to participate in the Brokerage’s Voluntary Lead Program, Agent agrees to service such leads in accordance with the Program’s quality standards. Failure to meet these standards will result in removal from the Program, but will not affect Agent’s status with the Brokerage.”.²⁴

Part 4: Minnesota Regulatory Environment – Compensation and Teams

Given the specific legal context provided in the research material, the system must be designed to comply with Minnesota state statutes, which are notoriously strict regarding real estate compensation and team structures.

4.1 Minnesota Statute § 82.70: Compensation Restrictions

Minnesota Statute § 82.70 mandates that real estate licensees can only accept compensation from their licensed broker.²⁶

• Implication for Lead Routing: The personality matching system cannot pay an agent directly for handling a lead, nor can agents pay each other referral fees directly (e.g., via Venmo). All financial flows must pass through the broker’s trust or operating account.
• Commission Splitting Exception: The statute explicitly permits commission splitting and referral fees between licensed brokers and salespersons.²⁸ Therefore, the personality matching system should be legally structured as a “Referral Team” under the umbrella of the brokerage. The “cost” of the lead is deducted as a referral fee (e.g., 25-35%) at the time of closing by the broker, who then pays the net commission to the agent.²⁹

4.2 Advertising and Team Requirements (MN Stat § 82.69)

If the personality matching system is marketed as a distinct brand (e.g., “MatchHome Realty”), it must comply with Minnesota’s advertising rules.

• Brokerage Prominence: Any advertising by a team must clearly and conspicuously display the real estate brokerage name.³⁰ The brokerage name must be more prominent than the team name.
• Broker Authorization: The inclusion of a team name in advertising must be authorized by the primary broker. This means the broker is ultimately responsible for the “promise” made by the personality match. If the marketing claims “We Match You Perfectly,” the broker is liable if that claim is deemed deceptive under false advertising statutes.³²

4.3 Supervision and “Primary Broker” Responsibility

MN Statute § 82.73 places the burden of supervision squarely on the primary broker. This includes the “ongoing monitoring of listing agreements, purchase agreements… and the review of all trust account books.”

• System Oversight: The broker must have administrative access to the personality matching platform (CRM) to monitor communications. The system cannot be a “shadow IT” operation run by a team leader without broker oversight. The broker is responsible for ensuring that the personality matching data is not being used to violate Fair Housing laws.³²

Part 5: Data Privacy – The Minnesota Consumer Data Privacy Act (MCDPA)

The operational landscape for data collection will change dramatically with the full implementation of the Minnesota Consumer Data Privacy Act (MCDPA) in July 2025. This law introduces strict regulations on “profiling” that directly impact a personality matching system.³³

5.1 Defining “Profiling” and “Legal Effects”

The MCDPA defines “profiling” as automated processing of personal data to predict a person’s preferences, interests, reliability, behavior, or economic situation.³⁵

• Significant Effects Clause: The law applies heightened scrutiny when profiling produces “legal or similarly significant effects,” which explicitly includes decisions regarding housing.
• The Compliance Risk: If the algorithm determines that Consumer A (based on their profile) should be matched with a luxury agent, and Consumer B is matched with a junior agent, this decision affects their access to housing services. This falls squarely under the “significant effects” clause, triggering the highest level of compliance obligations.

5.2 Consumer Rights Under MCDPA

To operate legally, the system must build the following rights into the user interface:

1. Right to Opt-Out: Consumers must have the ability to opt-out of profiling that results in significant decisions.³⁷ The system cannot force a personality quiz as a condition of service without a clear “opt-out” mechanism. If a consumer opts out, they must still be given a path to access services (e.g., a generic directory).
2. Right to Question Profiling: Consumers have the right to ask why a decision was made. If a buyer asks, “Why was I matched with this agent?”, the brokerage must be able to provide a transparent, non-discriminatory reason (e.g., “Based on your preference for detailed data analysis, we matched you with an agent who specializes in analytical market reports”).³³ This requirement effectively bans “black box” AI models where the logic is opaque.
3. Data Minimization: The system must only collect data that is “adequate, relevant, and reasonably necessary”.³⁶ Brokerages should avoid asking invasive psychological questions that aren’t directly tied to the real estate transaction style.

5.3 Consent and Transparency Framework

• Privacy Notice: The website and quiz interface must include a clear Privacy Notice explaining that data is used for “matching purposes” to improve service quality.³⁵
• Explicit Consent: The user flow should include a checkbox: “I consent to the use of my answers to generate a personalized agent match.”
• Algorithm Audits: The MCDPA requires “data privacy protection assessments” for profiling activities. The brokerage must document the potential risks (e.g., bias) and the mitigation strategies employed.³³

Part 6: Psychometric Science – The DISC Implementation Strategy

To effectively match consumers and agents, the system relies on a valid psychological framework. DISC is the industry standard for real estate because it focuses on observable behavior rather than deep pathology or cognitive processing, making it safer and more practical for sales contexts.¹³

6.1 The Four Profiles in Real Estate Transactions

The DISC model categorizes behavior into four quadrants, each requiring a distinct service approach ¹⁴:

6.2 The Matching Logic: Compatibility vs. Mirroring

The algorithm shouldn’t simply “mirror” (match D with D). Psychological research suggests that complementary styles often yield better results.

• D-Buyer + D-Agent: Can result in power struggles. A strong D buyer might actually prefer a highly competent C-Agent who executes orders flawlessly without challenging authority.³⁸
• S-Buyer + D-Agent: High risk of friction. The D-agent may “push” the S-buyer too hard, causing them to retreat or “ghost.” This is the prime use case for the system: preventing this mismatch to save the deal.³⁹

6.3 Brand Archetypes as a Consumer-Facing Layer

To make the psychometric profiling palatable to consumers, the system should mask the clinical DISC terminology with Brand Archetypes. Using concepts from marketing psychology (e.g., Jungian Archetypes), the system can present the match in a narrative format that feels like a premium service rather than a clinical diagnosis.⁴⁰

• High D Agent -> “The Hero” or “The Ruler”: Marketed as The Executive Match for busy professionals who want results.
• High I Agent -> “The Magician” or “The Jester”: Marketed as The Lifestyle Curator for buyers who want a fun, visionary experience.
• High S Agent -> “The Caregiver” or “The Innocent”: Marketed as The Trusted Guide for first-time buyers needing safety and support.
• High C Agent -> “The Sage”: Marketed as The Analyst for investors and detail-oriented buyers.⁴²

Part 7: Operational Implementation – The Tech Stack

Implementing this system “asap” requires leveraging existing automation tools rather than building custom software from scratch. A “No-Code” stack allows for rapid deployment and flexibility.

7.1 Lead Capture: The “Personality Quiz”

Use a dedicated quiz builder such as LeadQuizzes or involve.me embedded in the brokerage website.⁴⁴

• The Hook: “Find Your Perfect Real Estate Match in 2 Minutes.”
• The Questions: Mix “Hard” criteria (Budget, Location) with “Soft” criteria (DISC proxies).
  • Question: “When making a big decision, do you prefer: A) Quick summaries and bottom lines (D), B) Hearing what others did (I), C) Taking time to think it over (S), D) Seeing all the data and spreadsheets (C)?”.⁴⁶
• Avoid: Do not use questions that ask about “safe neighborhoods” or “good schools” in the matching logic, as these are proxies for Fair Housing violations.

7.2 Automation Workflow (Zapier Integration)

Use Zapier to act as the “Lead Router” between the Quiz, the CRM, and the Agent.⁴⁷

1. Trigger: New Quiz Submission (LeadQuizzes).
2. Formatter (Logic): Convert Quiz Answers to a Score (e.g., A=1, B=2…).
3. Paths (Routing Logic):
   • Path A (High D Score): Filter for agents tagged “D-Compatible” in the CRM (e.g., Follow Up Boss or Pipedrive).
   • Path B (High S Score): Filter for agents tagged “S-Compatible.”
4. Action (Round Robin): Distribute lead to the next available agent in that specific “bucket.”
5. Action (Notification): SMS to Agent: “New Lead: John Doe. Type: High-C (The Analyst). Script: Send market data immediately, do not try to ‘sell’ aggressively.”.⁴⁹

7.3 The Human Element: Inside Sales Agents (ISAs)

For higher conversion and legal safety, introduce an Inside Sales Agent (ISA) layer.

• Role: The ISA receives the raw lead, calls to verify motivation (and confirm the personality vibe), and then introduces the agent.
• Scripting: The ISA frames the hand-off to reinforce the value: “John, based on what you told me about wanting a data-driven approach, I’m going to pair you with Sarah. She’s our top analyst-agent and loves diving into the spreadsheets.” This confirms the “consumer choice” aspect and validates the matching logic.⁵⁰

Part 8: Ethical Transparency & Consumer Choice

Consumers are increasingly wary of “black box” algorithms. To build trust and avoid ethical pitfalls, the system must operate with radical transparency regarding its financial and operational mechanics.

8.1 The “Black Box” Problem and Referral Fees

Historically, consumers were unaware that 25% of their agent’s commission might be paid to a lead generator. Recent debates within the National Association of REALTORS® (NAR) regarding Article 6 of the Code of Ethics suggest a strong industry movement toward mandatory disclosure of referral fees.⁵¹ Although a recent measure to mandate this failed by a slim margin, transparency is a powerful risk management strategy.

• eXp Realty’s “Consumer Choice” Model: Industry leaders like eXp Realty are implementing forms that explicitly disclose referral fees to consumers, ensuring they understand the financial motivations behind a referral.⁵²
• Recommendation: The system should include a “Transparency Disclosure” informing the client: “We have matched you with Agent X because they match your communication style. Agent X pays a referral fee to support the operation of this matching platform.” This disarms potential conflicts of interest.

8.2 Handling “Mismatches” and Feedback loops

The system must allow for failure. If a match does not work, the consumer needs an easy “Rematch Me” button.

• Learning Mechanism: If a D-Buyer rejects a D-Agent, the system should log this data. Perhaps that specific D-Buyer actually wanted a subservient S-Agent. The algorithm should learn from these rejections to refine its future accuracy.
• Post-Transaction Survey: Send a survey asking, “Did you feel your agent matched your working style?” This data is crucial for the “Adverse Impact Audits” required by the MCDPA.

Part 9: Strategic Roadmap for Launch

Phase 1: Preparation (Weeks 1-4)

1. Legal: Draft the Voluntary Lead Team Agreement (IC compliance) and the Consumer Profiling Privacy Notice (MCDPA compliance).
2. Vendor Selection: License a validated DISC assessment tool for agents (e.g., Everything DiSC); build the consumer-facing quiz using non-biased questions (LeadQuizzes).
3. Agent Baselining: Have all participating agents take the DISC assessment. Categorize them into “Service Buckets” (e.g., The Analyst, The Negotiator, The Concierge).

Phase 2: Tech Setup (Weeks 5-6)

1. Zapier Construction: Build the routing logic. Ensure that no demographic data (zip code, name) flows into the matching filter.
2. ISA Training: Train ISAs on the “Handoff Scripts” that explain the match rationale to consumers.
3. Beta Test: Run the algorithm on past closed data. Would the algorithm have matched the successful pairings?

Phase 3: Launch & Monitor (Week 7+)

1. Soft Launch: Roll out to a small “Alpha Team” of high-performing agents who have opted in.
2. Adverse Impact Audit: After 100 matches, review demographic data. Are we steering minority buyers to specific agents? If so, recalibrate immediately.
3. Consumer Feedback Loop: Implement the post-transaction survey to validate the “business necessity” of the match.

Conclusion

A personality-matched real estate system offers a compelling value proposition: higher trust, lower friction, and better closing rates. However, it requires a rigorous operational structure that prioritizes consent (Privacy Law), autonomy (Labor Law), and neutrality (Fair Housing Law). By building a transparent, opt-in system that uses validated psychometrics rather than demographic proxies, a brokerage can legally and ethically operationalize the “dating app” model for real estate.

Tables and Structured Data

Table 1: Risk Assessment of Personality Matching Features

Table 2: DISC Profile to Agent Matching Matrix

Table 3: Minnesota Consumer Data Privacy Act (MCDPA) Compliance Checklist

Report filed by: Senior Real Estate Operations & Compliance Consultant

Date: December 30, 2025