15. Standards, Ethics, Governance and Valuation Technology

Introduction and Context

Real estate valuation is not merely the production of a number. It is a professional opinion formed from evidence, assumptions, methods, judgement and communication. Because valuation conclusions influence lending, taxation, financial reporting, investment, compulsory acquisition, development decisions and public policy, the credibility of the process is as important as the conclusion itself. A technically correct calculation can still be professionally defective if the valuer lacks independence, fails to disclose a conflict, uses unsuitable data, cannot reproduce the analysis, or communicates uncertainty inadequately.

The contemporary valuer therefore works within several overlapping systems. International standards provide common concepts and reporting principles; national legislation determines enforceability and local practice; professional bodies establish ethical and competency requirements; and organisational governance controls how instructions are accepted, reviewed, approved and archived. At the same time, automated valuation models (AVMs), geographic information systems (GIS), digital twins, blockchain-based records and artificial intelligence are changing how evidence is collected and interpreted.

Technology increases speed and analytical capacity, but it does not remove professional responsibility. An algorithm may identify patterns without understanding legal title, abnormal transaction conditions, market sentiment or a neighbourhood’s social context. The central challenge is to combine innovation with explainability, data governance, cybersecurity, fairness and accountable professional judgement.

Learning Objectives

By the end of this chapter, the learner should be able to:

1. Explain the purpose and relationship of IVS, RICS standards and relevant national valuation requirements.

2. Distinguish valuation standards, professional rules, organisational controls and legal requirements.

3. Apply principles of independence, conflict management, review, documentation and auditability.

4. Evaluate AVMs, GIS, digital twins, blockchain and AI in real estate valuation.

5. Identify risks involving explainability, cybersecurity, data quality, bias and ESG information.

6. Design a governance-oriented valuation workflow for professional practice.

7. Recommend controls that preserve reliability while enabling responsible technological innovation.

The Standards Architecture of Valuation

Standards as a professional control system

A valuation standard establishes a disciplined language and a set of minimum expectations. It normally addresses the purpose of the valuation, the subject asset, the interest being valued, the basis of value, assumptions, methods, investigations, evidence and reporting. Standards do not eliminate judgement. Rather, they make judgement more structured, transparent and reviewable.

Three layers should be distinguished. First, law and regulation may prescribe who can value, which assets require valuation, how valuations are used in financial statements, or how compulsory purchase compensation is determined. Second, professional standards provide technical and ethical requirements, often with mandatory and advisory elements. Third, organisational policies specify internal approval thresholds, data controls, segregation of duties and quality assurance procedures. A valuer must comply with all three, and a professional standard cannot be used to justify non-compliance with applicable law.

International Valuation Standards

The International Valuation Standards (IVS) provide an internationally recognised framework for consistent valuation practice. Their importance lies in common principles rather than in prescribing one universal calculation. IVS generally requires the valuer to understand the assignment, identify the asset and interest, select an appropriate basis of value, use suitable methods and inputs, and communicate the conclusion clearly.

A valuation based on IVS should be anchored in the scope of work. The scope identifies matters such as the client, intended users, intended use, valuation date, currency, asset or liability, interest, basis of value, extent of investigation, assumptions, limitations and reporting format. An apparently precise value with an unclear scope is not a strong professional product.

The basis of value is particularly important. Market value, investment value, fair value and liquidation value answer different questions. Market value generally reflects an exchange between appropriate market participants under defined conditions. Investment value reflects value to a particular owner or investor. Fair value may be determined under an accounting or regulatory framework and can contain jurisdiction-specific meaning. A valuer must never substitute one basis for another merely because the resulting figure appears convenient.

RICS and national standards

The Royal Institution of Chartered Surveyors (RICS) professional and valuation standards translate international principles into a detailed professional practice environment. RICS requirements commonly emphasise competence, ethical behaviour, independence, conflicts of interest, terms of engagement, investigations, valuation approaches, reporting and monitoring. RICS members and regulated firms must also consider mandatory professional statements and rules of conduct.

National standards may adopt IVS, incorporate RICS principles, or establish an independent framework. They may contain specific requirements for land registration, taxation, mortgage lending, public acquisition, affordable housing, agricultural land or environmental restrictions. Local market evidence and legal concepts also affect the analysis. For example, the meaning of a transferable development right, leasehold interest, zoning permission or customary land right cannot be assumed from another jurisdiction.

A useful professional sequence is therefore:

1. Identify the legal and regulatory context.

2. Determine whether a national standard or sector-specific rule applies.

3. Establish whether IVS, RICS or another professional framework is required by the client or institution.

4. Resolve differences by applying the stricter applicable requirement, subject to legal advice where necessary.

5. Disclose the standards and departures used in the report.

Important terminology

A valuation date is the date to which the opinion of value relates. An effective date may be used in some reporting contexts to identify when the conclusion is considered operative. The subject asset includes the physical property and the relevant legal or economic interest. Highest and best use refers to the reasonably probable, legally permissible, physically possible and financially feasible use that results in the highest value, where that concept is relevant to the basis of value. Materiality concerns whether an omission, error or uncertainty could influence the decisions of intended users. Professional scepticism means maintaining a questioning mind and critically evaluating evidence rather than accepting representations automatically.

From Instruction to Defensible Opinion

A defensible valuation begins before inspection or modelling. The valuer should assess competence, capacity, independence, timing, data availability and the intended use. An instruction outside the valuer’s competence should be declined, or accepted only with appropriate specialist support and transparent disclosure.

The engagement letter should specify the parties, purpose, asset, interest, valuation date, basis of value, assumptions, investigations, reliance on information, limitations, fees, confidentiality, use of specialists and reporting requirements. Ambiguous instructions often produce disputes later, particularly where a lender, borrower, auditor and investor interpret the same report differently.

The work then proceeds through evidence collection and analysis. Title documents, leases, planning permissions, building information, income records, environmental reports, comparable transactions and market data should be assessed for relevance and reliability. A transaction is not automatically a valid comparable because it is geographically close. The valuer must consider date, motivation, financing, physical characteristics, legal interest, condition, planning status and abnormal circumstances.

The report should allow a knowledgeable reviewer to understand how the conclusion was reached. It need not disclose every exploratory calculation, but it should identify the principal evidence, methods, assumptions, adjustments, limitations and significant uncertainties. Where several methods produce different indications, the valuer should explain the reconciliation rather than presenting a false appearance of mathematical certainty.

Independence, Conflicts and Ethical Judgement

Independence is both an actual condition and a perception held by intended users. A valuer may be technically competent yet unsuitable if financial interests, family relationships, prior involvement, contingent fees or pressure from an instructing party could influence the work. The relevant question is not only whether the valuer believes they are unbiased, but whether a reasonable observer would question the independence of the assignment.

Conflicts may be actual, potential or perceived. Examples include valuing a property owned by the valuer, valuing an asset for a client involved in a dispute in which the valuer previously advised the opposing party, accepting a fee linked to a target valuation, or preparing a valuation that will be used to support a transaction on which the firm earns another success fee.

A practical conflict-management process includes:

• disclosure of relevant interests before accepting the instruction;
• assessment of whether informed consent is legally and professionally sufficient;
• physical and information barriers where appropriate;
• appointment of an independent valuer or reviewer;
• refusal or withdrawal where the conflict cannot be controlled; and
• documentation of the decision and communications.

Ethics also concerns competence, confidentiality, honesty, respect for users and avoidance of misleading precision. A valuer should not suppress adverse evidence, alter assumptions to achieve a requested result, or describe an automated output as an independent valuation without understanding and validating the model.

Governance, Review and the Audit Trail

Valuation governance is the system through which an organisation directs, controls and monitors valuation activity. It is especially important for banks, funds, listed companies, public authorities and large advisory practices, where valuations are produced at scale and may affect capital allocation or regulatory reporting.

A sound governance model separates roles without making them disconnected. The originating valuer gathers evidence and performs the analysis. A reviewer evaluates scope, methods, inputs, calculations, assumptions and reporting. A manager or valuation committee may approve high-risk or high-value assignments. Internal audit examines whether the overall process operates as designed, while external auditors or regulators may inspect selected files.

The three lines model is useful. Operational valuers constitute the first line and own the quality of their work. Risk, compliance or technical standards teams form the second line by setting policies, monitoring conflicts and challenging exceptions. Internal audit is the third line, independently evaluating governance and controls.

A complete audit trail should show:

• the original instruction and subsequent changes;
• identity, authority and competence of contributors;
• source documents and data extraction dates;
• versions of spreadsheets, models, code and assumptions;
• inspection notes, photographs and correspondence;
• comparable selection and adjustment rationale;
• review comments and how they were resolved;
• approvals, disclosures and final report distribution.

Version control is essential. A model that silently overwrites earlier assumptions cannot support a reliable review. Access permissions should ensure that users can edit only what they are authorised to change. Retention periods should reflect law, professional requirements, contractual obligations and the possibility of later challenge.

Review is not a mechanical re-performance of arithmetic. A reviewer should ask whether the valuation question was correctly framed, whether evidence is sufficient, whether the method is appropriate, whether assumptions are reasonable, whether uncertainty is adequately communicated and whether the conclusion is consistent with market behaviour. For automated models, review must also address data lineage, model performance, drift and override decisions.

Technology as Evidence, Not Authority

Automated valuation models

An AVM uses statistical, econometric, machine-learning or rule-based techniques to estimate value from property and market data. Typical inputs include location, area, age, condition, transaction history, rents, planning attributes and neighbourhood variables. AVMs may support portfolio screening, collateral monitoring, tax administration and the prioritisation of properties requiring full inspection.

Their value depends on the relationship between the model and the use case. A model suitable for homogeneous residential apartments may be unreliable for unique commercial assets, development land, heritage buildings or properties affected by unusual legal restrictions. Performance should be assessed through out-of-sample testing, error analysis, confidence intervals or prediction bands, and monitoring across locations, price ranges and property types.

An AVM output is not automatically a valuation conclusion. Professional oversight is required when data are sparse, the asset is atypical, market conditions have changed, or the model identifies a low-confidence case. Human overrides should be permitted, but they must be reasoned, documented and monitored; otherwise overrides become an unrecorded source of bias.

GIS and spatial analysis

GIS links property information to geography. It can display comparable transactions, accessibility, flood risk, zoning, infrastructure, demographics, environmental constraints and neighbourhood change. Spatial analysis improves the valuer’s ability to distinguish location effects from building effects and to identify spatial clusters or outliers.

However, GIS results depend on coordinate accuracy, scale, update frequency and the meaning of the underlying datasets. A map can create an impression of objectivity while concealing incomplete boundaries, incompatible definitions or outdated planning information. Spatial correlation should not be confused with causation. A higher value near a transport station may reflect several simultaneous factors, including development intensity, retail activity and socioeconomic characteristics.

Digital twins and sensor data

A digital twin is a continuously updated digital representation of a physical asset or system, potentially integrating building information modelling, maintenance records, energy use, occupancy, environmental sensors and operational data. In valuation, a digital twin may improve understanding of building condition, operating efficiency, remaining life and capital expenditure requirements.

The limitations are substantial. Sensor data may be incomplete or manipulated; data ownership may be uncertain; interoperability between systems may be poor; and operational performance may not translate directly into market value. The valuer must distinguish observed performance from assumptions about future performance.

Blockchain and smart contracts

Blockchain can create a tamper-evident shared record of transactions, approvals, ownership events or document hashes. Smart contracts may automate defined actions when stated conditions are met. These technologies could improve traceability and reduce reconciliation costs in fragmented property processes.

They do not, however, guarantee that the original information was true. A blockchain may preserve an inaccurate title record or an incorrectly entered transaction. Legal recognition, privacy, identity management, governance of permissioned networks and the correction of erroneous data remain important concerns. Immutability is not a substitute for verification.

Artificial intelligence and explainability

AI can classify documents, extract lease terms, identify comparable properties, generate market summaries, detect anomalies and estimate values. Generative AI can assist with drafting, but it may produce unsupported statements, omit qualifications or reproduce biased patterns in training data. It should not be treated as an authoritative source without verification.

Explainability is central to professional use. The user should be able to understand which variables materially influenced an output, how the model performs for the relevant asset class, what data were used, and where uncertainty is greatest. A technically sophisticated model that cannot be challenged may be unsuitable for regulated or high-consequence decisions.

Data Governance, Cybersecurity, Bias and ESG

Data governance establishes rules for ownership, quality, access, consent, retention, classification and use. Important controls include validation checks, duplicate detection, source ranking, data dictionaries, metadata, access logs and documented correction procedures. Personal information, tenant data, security plans and commercially sensitive transaction details require particular protection.

Cybersecurity risks include ransomware, credential theft, unauthorised model changes, data poisoning, interception of confidential reports and disruption of valuation platforms. Controls should include multi-factor authentication, least-privilege access, encryption, secure backups, patch management, network segregation, vendor due diligence, incident response and tested recovery procedures. A valuation firm may be small, but its data can have high financial and strategic value.

Bias can enter through historical transactions, missing data, proxy variables, geographic coverage, sampling choices or human overrides. A model may systematically underperform for certain property types, neighbourhoods or ownership groups. Fairness testing should compare error rates across relevant segments and examine whether variables act as inappropriate proxies. Removing a sensitive variable does not necessarily remove bias if correlated variables remain.

ESG information is increasingly relevant to valuation, particularly through energy costs, regulatory compliance, climate exposure, obsolescence, tenant demand, financing conditions and insurance availability. Yet ESG reporting is vulnerable to inconsistent boundaries, unverifiable claims and greenwashing. The valuer should identify the source, date, scope and reliability of ESG data and avoid treating a sustainability label as proof of superior value. Material ESG effects should be reflected through appropriate cash-flow assumptions, risk analysis, comparable selection or scenario testing, rather than through an unsupported premium.

Illustrative Case: Governance Failure in an Automated Lending Valuation

Background

A fictional national lender, Meridian Finance, used an AVM to monitor a portfolio of 18,000 residential mortgages. The model had been trained on recent sales and property characteristics and was generally reliable for standard urban apartments. The lender planned to use the outputs to determine collateral buffers and identify loans requiring revaluation.

Situation and problem

A sharp market change occurred in a suburban district where a major employer announced relocation. The AVM continued to produce values close to earlier peak prices because the training data were slow to reflect reduced demand. A portfolio manager asked analysts to adjust selected outputs upward so that the collateral ratio would remain within target limits. The adjustments were made in a spreadsheet outside the controlled model and were not recorded consistently.

A later review also found that the model treated floor area as a strong positive variable but did not adequately distinguish legally approved area from unapproved extensions. Properties in lower-income neighbourhoods had wider prediction errors because transaction data were sparse, although this was not reported to decision-makers.

Analysis

The primary failure was governance, not merely model accuracy. The intended use had changed from screening to a decision affecting lending risk, but the model had not been revalidated for that purpose. Market drift was not monitored. Overrides were permitted without documented criteria. The model’s uncertainty was not communicated, and the uncontrolled spreadsheet created an audit weakness.

The data problem also had legal and ethical dimensions. Unapproved extensions affected marketability and value, while uneven error rates could produce unfair lending outcomes. The model’s apparently objective output concealed important limitations.

Approach and solution

Meridian suspended automated use for the affected district and commissioned an independent review. It introduced a model-risk policy requiring approval of intended use, validation before deployment, periodic back-testing, monitoring of drift and documented change control. The lender created a low-confidence flag based on prediction error, data sparsity, unusual property attributes and market discontinuity.

All overrides were moved into a controlled workflow requiring reasons, evidence, authorisation and later sampling. Properties with legal-area uncertainty or unusual characteristics were referred for full professional valuation. The reporting interface displayed prediction ranges, data vintage, key variables and known exclusions rather than only a single number.

Outcome

The lender retained the AVM for portfolio triage but stopped treating it as an unconditional valuation substitute. Its risk committee received quarterly performance reports, including segment-level error analysis. The revised process was slower for exceptional cases but reduced uncontrolled adjustments and improved the credibility of collateral decisions.

Learning points

The case demonstrates that a model’s usefulness depends on purpose, data, governance and communication. Professional accountability remains with the institution and responsible valuers. An audit trail must cover not only the final output but also model versions, overrides, data limitations and decisions about escalation.

Challenges and Common Professional Errors

Several recurring errors weaken technology-enabled valuation. The first is automation bias: accepting a computer-generated result because it appears scientific. The second is false precision, in which a narrow numerical output hides uncertain evidence. The third is data determinism, assuming that available data are necessarily relevant or complete. The fourth is scope drift, where a model or report is reused for a purpose not contemplated in the original engagement.

Other risks include inadequate specialist supervision, reliance on unverified client information, failure to disclose material assumptions, weak vendor contracts, insufficient cybersecurity testing and excessive dependence on a single data provider. Ethical failures may occur when commercial pressure influences the selected method or when a valuer presents a review as independent despite participating in the original analysis.

A mature organisation responds through proportionality. Not every residential valuation requires the same level of governance as a complex infrastructure portfolio, but higher value, greater uncertainty, public importance and greater reliance require stronger controls.

Emerging Professional Requirements and Best Practice

The future valuer will require a blended competence profile: valuation theory, market analysis, law, accounting awareness, data literacy, model-risk understanding, cybersecurity awareness and communication skills. Technical expertise in software will not replace knowledge of property rights and market behaviour; instead, the professional must connect both domains.

Best practice includes the following:

1. Define the valuation question and intended use before choosing technology.

2. Match the method and model to the asset, market and decision risk.

3. Maintain independence and disclose conflicts before accepting the assignment.

4. Use reliable, current and traceable data, with documented limitations.

5. Validate models before use and monitor performance after deployment.

6. Require human review for exceptional, high-risk or low-confidence cases.

7. Preserve versions, assumptions, source documents, overrides and approvals.

8. Communicate uncertainty, sensitivity and material ESG or legal assumptions.

9. Protect confidential information through proportionate cybersecurity controls.

10. Continue professional education in standards, regulation, analytics and ethics.

The profession is moving from the image of the valuer as an isolated calculator toward that of an accountable assurance professional. The distinctive contribution of the valuer is not simply access to a formula or database. It is the ability to frame the problem correctly, evaluate imperfect evidence, recognise exceptional circumstances and explain a conclusion in a manner that users can responsibly rely upon.

Chapter Summary

Valuation standards create a common framework for defining the assignment, selecting a basis of value, analysing evidence and reporting conclusions. IVS provides international principles, while RICS and national standards add professional, legal and jurisdiction-specific requirements. Independence, conflict management, competence, confidentiality and honest communication are fundamental ethical obligations.

Governance converts professional expectations into operational controls through clear roles, reviews, approvals, version control and audit trails. AVMs, GIS, digital twins, blockchain and AI can improve efficiency and insight, but they remain tools rather than authorities. Their responsible use requires attention to data quality, model validation, explainability, cybersecurity, bias, ESG evidence and human oversight. The future professional will combine valuation judgement with digital literacy and strong ethical accountability.

Key Terms

Automated Valuation Model (AVM): A statistical, rule-based or machine-learning system that estimates property value from data.

Audit trail: A chronological record showing evidence, analysis, changes, reviews and approvals supporting a valuation.

Basis of value: The fundamental assumptions defining the type of value being estimated.

Blockchain: A distributed record system designed to preserve transactions or data entries in a tamper-evident manner.

Conflict of interest: A circumstance in which personal, financial or professional interests may affect, or appear to affect, impartiality.

Digital twin: A digital representation of a physical asset that may be updated using operational and sensor data.

ESG: Environmental, social and governance factors that may affect asset performance, risk and value.

Explainability: The extent to which the reasoning, inputs and influences behind a model output can be understood.

Fair value: A value concept defined within a particular accounting or regulatory framework.

GIS: A geographic information system used to capture, analyse and display spatial data.

Highest and best use: The use that is legally permissible, physically possible, financially feasible and maximally productive where relevant.

Independence: Freedom from interests or influences that could compromise objective professional judgement.

Investment value: The value of an asset to a particular owner or identified investor.

Market value: An opinion of exchange value based on defined market-participant assumptions.

Materiality: The significance of an omission or error to the decisions of intended users.

Model drift: Deterioration in model performance as market conditions or data relationships change.

Professional scepticism: A questioning and critical evaluation of evidence and representations.

Scope of work: The documented definition of the assignment, purpose, asset, basis, date, investigations and limitations.

Three lines model: A governance structure separating operational control, risk oversight and independent assurance.

Valuation governance: Policies, roles and controls directing, monitoring and assuring valuation activity.

Review Questions

Short Answer Questions

1. Why is the scope of work important in a standards-compliant valuation?

2. Distinguish market value from investment value.

3. What is meant by model drift in an AVM?

4. Identify two circumstances that may create a conflict of interest.

5. Why does blockchain not guarantee the truth of property information?

Long Answer Questions

1. Explain how IVS, RICS standards, national requirements and organisational policies interact in professional valuation practice.

2. Discuss the elements of an effective valuation audit trail and explain its importance to review and regulatory assurance.

3. Examine the ethical responsibilities of a valuer using AI or an AVM in a high-consequence lending decision.

4. Analyse the role of independence, competence and disclosure in maintaining confidence in valuation conclusions.

5. Evaluate the opportunities and limitations of GIS, digital twins and ESG data in real estate valuation.

Analytical / Application Questions

1. A model produces reliable estimates for standard apartments but performs poorly for development land. Design a governance process for deciding when the model may be used and when a full valuation is required.

2. A client asks a valuer to revise assumptions so that a property reaches a target loan-to-value ratio. Analyse the ethical and professional issues and recommend an appropriate response.

3. A fund wants to use building energy data to justify a value premium. Explain what evidence, adjustments, sensitivity analysis and disclosures would be needed before accepting the conclusion.

22. Multiple Choice Questions with Answers

Question 1

Which feature most directly makes a valuation conclusion defensible under professional standards?

A. A highly complex mathematical model

B. A clearly defined scope, suitable evidence, reasoned analysis and transparent reporting

C. A single comparable transaction close to the property

D. A narrow value range with no stated assumptions

Correct Answer: B. A clearly defined scope, suitable evidence, reasoned analysis and transparent reporting

Question 2

A valuer discovers that the client’s desired transaction depends on achieving a specific valuation figure. What is the most appropriate first response?

A. Adjust the comparable evidence to meet the figure

B. Accept the figure as a client assumption

C. Assess the pressure and independence implications, document them and refuse or control the conflict appropriately

D. Remove all adverse evidence from the report

Correct Answer: C. Assess the pressure and independence implications, document them and refuse or control the conflict appropriately

Question 3

Which statement best distinguishes an AVM output from a professional valuation conclusion?

A. An AVM output is always more objective

B. An AVM output requires no data validation

C. An AVM output is a tool that requires suitability assessment, validation and professional oversight

D. An AVM can be used for every property type without adjustment

Correct Answer: C. An AVM output is a tool that requires suitability assessment, validation and professional oversight

Question 4

What is the principal purpose of a valuation audit trail?

A. To make the report longer

B. To show how evidence, assumptions, changes, reviews and approvals support the conclusion

C. To replace the need for professional judgement

D. To prevent any future correction of data

Correct Answer: B. To show how evidence, assumptions, changes, reviews and approvals support the conclusion

Question 5

Which situation is most likely to indicate model drift?

A. The model is tested against a stable and representative sample

B. The model’s error increases after a major market change while its training data remain unchanged

C. A reviewer checks the model inputs

D. The model records its version number

Correct Answer: B. The model’s error increases after a major market change while its training data remain unchanged

Question 6

Why is explainability important when AI is used in valuation?

A. It guarantees that the model is correct

B. It enables users to understand influential inputs, limitations and uncertainty

C. It eliminates the need for cybersecurity

D. It ensures all properties receive identical values

Correct Answer: B. It enables users to understand influential inputs, limitations and uncertainty

Question 7

Which is a limitation of blockchain in property records?

A. It cannot preserve a record once created

B. It guarantees that the original data entered were accurate

C. It may preserve inaccurate information if the original entry was not verified

D. It prevents all privacy concerns

Correct Answer: C. It may preserve inaccurate information if the original entry was not verified

Question 8

A GIS map shows that properties near a new station have higher values. What should the valuer avoid assuming automatically?

A. That location can never affect value

B. That spatial association alone proves the station caused the entire value difference

C. That mapping can assist comparable analysis

D. That data quality and date should be checked

Correct Answer: B. That spatial association alone proves the station caused the entire value difference

Question 9

Which control best addresses unauthorised changes to a valuation model?

A. Allowing all users unrestricted editing rights

B. Deleting earlier versions after issuing the report

C. Version control, access permissions, change logs and independent review

D. Using a more attractive report design

Correct Answer: C. Version control, access permissions, change logs and independent review

Question 10

A lender’s AVM has wide errors for a particular neighbourhood because transaction data are sparse. What is the most responsible action?

A. Hide the error because the model is generally accurate

B. Use the output without qualification

C. Flag the low-confidence cases and escalate appropriate properties for further analysis or full valuation

D. Increase every value in the neighbourhood by the average error

Correct Answer: C. Flag the low-confidence cases and escalate appropriate properties for further analysis or full valuation