Managerial Reporting, Governance, and Ethical Accountability

Introduction

Marketing investment creates value only when managers can explain what was funded, what occurred, what was learned, and what should happen next. In contemporary brand management, this requirement extends beyond producing attractive dashboards. It involves designing an accountable information system linking strategic objectives, financial resources, customer outcomes, analytical evidence, managerial authority, and ethical safeguards.

Managerial reporting is therefore a governance activity, not merely a communication activity. A report can influence budget allocation, agency remuneration, employee performance evaluation, product decisions, and public claims about marketing effectiveness. If its definitions are unstable, its data unaudited, or its incentives poorly designed, reporting may create an illusion of control rather than control itself.

This chapter examines how marketing leaders create executive business cases, dashboards, variance reports, and post-investment reviews. It also explains decision rights, metric ownership, auditability, agency controls, privacy, bias, manipulation, transparency, and responsible data use. The central principle is that credible measurement requires both analytical quality and institutional integrity.

Learning Objectives

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

1. Construct an executive business case that connects strategic objectives, investment, evidence, risk, and decision requirements.
2. Design a decision-oriented marketing dashboard with clearly defined metrics and owners.
3. Prepare and interpret budget, performance, and forecast variance reports.
4. Conduct a post-investment review that distinguishes results, attribution, learning, and accountability.
5. Establish decision rights, data lineage, metric ownership, audit trails, and agency controls.
6. Evaluate marketing analytics practices through the principles of privacy, fairness, transparency, explainability, and responsible data use.
7. Critically assess how incentives, AI systems, and reporting conventions can produce manipulation or misleading conclusions.

1. Reporting as an Organizational Control System

A marketing report performs at least four functions:

• Descriptive: What happened?
• Diagnostic: Why did it happen?
• Predictive: What is likely to happen next?
• Prescriptive: What action is recommended?

These functions should not be confused. A dashboard may show that sales increased, but it does not by itself establish that advertising caused the increase. Similarly, a forecast is not a promise, and a model-generated recommendation is not a decision.

A useful reporting architecture connects five layers:

1. Strategy: brand objectives, commercial priorities, and customer or stakeholder outcomes.
2. Resources: budgets, personnel, media, technology, agency fees, and data assets.
3. Activities: campaigns, content, promotions, partnerships, and customer experiences.
4. Outcomes: reach, consideration, conversion, retention, revenue, margin, and brand equity.
5. Governance: definitions, permissions, controls, review forums, escalation, and accountability.

The fifth layer is frequently neglected. Without governance, the same metric may have different definitions across finance, marketing, sales, and an agency. A “conversion” might mean a click, a qualified lead, a completed order, or a first profitable purchase. A dashboard can be technically accurate while still being managerially misleading if its definitions are unclear.

1.1 The logic of managerial accountability

Accountability requires three conditions:

• Authority: a person or group can make or approve a decision.
• Information: relevant and sufficiently reliable evidence is available.
• Answerability: the decision and its consequences can be reviewed.

Responsibility without authority is unfair; authority without answerability is dangerous. A robust marketing governance system assigns both operational ownership and review independence. For example, a campaign manager may own optimization, while finance validates expenditure and an analytics function reviews measurement assumptions.

2. Executive Business Cases

An executive business case converts a proposed marketing investment into a structured decision. It is not a catalogue of campaign ideas. It should answer five questions:

1. What strategic or commercial problem is being addressed?
2. Why is the proposed intervention plausible?
3. What will it cost, and what benefits are expected?
4. How will results, uncertainty, and risks be evaluated?
5. What decision is required from the executive body?

2.1 Recommended structure

A concise but rigorous business case normally contains:

The business case should distinguish incremental outcomes from outcomes that would have occurred anyway. It should also distinguish revenue from contribution margin. If an investment generates revenue but fails to cover variable costs, reporting it as a financial success can lead to destructive scaling.

2.2 Core financial measures

Return on investment is commonly expressed as:

Here, incremental return must be defined consistently. It may be incremental contribution margin, incremental profit, or another approved return measure. Investment cost includes the relevant costs of the intervention, such as media, production, agency fees, technology, and research, subject to the organization’s accounting policy. ROI is dimensionless because both numerator and denominator are expressed in the same monetary units.

If incremental contribution margin is used, a simplified calculation is:

The contribution margin rate is the proportion of incremental revenue remaining after variable costs directly associated with the sales. This formula is appropriate only when the margin rate is applicable to the incremental sales being evaluated and fixed costs are not being incorrectly treated as variable costs.

Worked example. A campaign costs $240,000 and produces estimated incremental revenue of $900,000. The applicable contribution margin rate is 40%.

Incremental contribution margin = $900,000 × 0.40 = $360,000.

ROI = ($360,000 − $240,000) ÷ $240,000 = $120,000 ÷ $240,000 = 0.50, or 50%.

The campaign has a positive contribution-based ROI of 50%. This does not prove causality; the incremental revenue estimate depends on the evaluation design. Nor does it necessarily mean that the campaign should be repeated, because strategic fit, long-term brand effects, risk, and opportunity cost also matter.

For multi-period investments, managers should consider the time value of money. Net present value is:

where C₀ is the initial investment, CFₜ is the incremental cash flow in period t, r is the discount rate per period, and T is the number of periods. The formula assumes that cash-flow timing and the discount rate are defined consistently. NPV is positive when the present value of expected incremental cash flows exceeds the initial investment.

2.3 Scenario and sensitivity analysis

Point estimates conceal uncertainty. A stronger business case includes at least a downside, base, and upside case. Sensitivity analysis changes one or more assumptions—such as response rate, average order value, media cost, or margin—to identify which variables most affect the decision.

Managers should avoid presenting a single precise forecast when evidence is weak. Ranges, confidence intervals, scenario probabilities, or explicit qualitative uncertainty are more honest than false precision. A forecast should also state whether it is an unconditional prediction, a conditional estimate, or a target.

3. Dashboard Design for Executive Decisions

A dashboard is useful when it reduces decision latency without reducing decision quality. It should not attempt to display every available metric. The design begins with decisions, not data availability.

3.1 Metric hierarchy

A balanced marketing dashboard may contain four levels:

• Strategic outcomes: profitable growth, brand equity, customer lifetime value, retention, or market share.
• Commercial outcomes: incremental contribution, orders, qualified demand, and customer profitability.
• Leading indicators: reach quality, consideration, engagement, qualified traffic, and conversion progression.
• Operational controls: spend pace, delivery, frequency, data quality, compliance, and supplier performance.

Leading indicators should not be treated as substitutes for outcomes. High engagement can coexist with low profitability. Likewise, a short-term decline in response may accompany a valuable long-term brand-building investment.

3.2 Metric specification

Every important metric should have a formal definition record containing:

• metric name and purpose;
• numerator and denominator, where relevant;
• unit and currency;
• inclusion and exclusion rules;
• time window and attribution window;
• source systems and data refresh frequency;
• owner and approving authority;
• known limitations and permitted uses;
• version history and change date.

For example, a conversion rate can be defined as:

The definition is meaningful only when “completed target action” and “eligible opportunity” are specified. A platform-reported conversion rate and a finance-reconciled order conversion rate may both be valid for different purposes, but they must not be silently compared as if they were identical.

3.3 Dashboard visual integrity

Visual design is part of analytical ethics. Truncated axes, inconsistent time periods, selective date ranges, and color choices that exaggerate small differences can manipulate interpretation. A dashboard should show:

• actuals versus plan and prior comparable periods;
• absolute values as well as percentages;
• denominator and sample size where applicable;
• uncertainty or confidence information when relevant;
• data freshness and missing-data warnings;
• explanations for major changes;
• a clear distinction between observed, modeled, and forecast values.

A red status does not automatically mean failure, and a green status does not automatically mean success. Thresholds should reflect decision relevance rather than cosmetic optimism.

4. Variance Reporting and Management by Exception

Variance reporting compares actual performance with an approved reference point. The reference may be budget, forecast, target, prior period, or control group. These are different comparisons and should be labeled precisely.

For a simple absolute variance:

For a percentage variance, provided the reference is non-zero:

The sign convention must be declared. For expenditure, an actual cost above budget is usually unfavorable; for revenue, an actual above plan is usually favorable. A negative numerical variance is not inherently good or bad.

4.1 Types of variance

• Budget variance: actual spend compared with authorized spend.
• Volume variance: performance differs because the number of opportunities changed.
• Rate variance: unit cost, price, or response rate differs from assumption.
• Mix variance: the composition of channels, audiences, products, or regions changed.
• Timing variance: activity occurred earlier or later than planned.
• Forecast variance: actual results differ from the latest expected outcome.
• Attribution variance: reported platform credit differs from independent or finance-reconciled results.

A useful variance report does not stop at the arithmetic. It identifies cause, materiality, owner, action, and due date.

The final row should not be interpreted without considering whether the order count is truly incremental. A platform’s attributed orders may be materially higher than an experimental estimate.

4.2 Materiality and escalation

Not every variance deserves executive intervention. Organizations should define materiality thresholds based on financial exposure, strategic importance, reputational risk, legal sensitivity, and reversibility. A small numerical variance in a privacy-sensitive campaign may require faster escalation than a larger routine delivery variance.

5. Post-Investment Reviews

A post-investment review is a structured learning and accountability process conducted after a campaign, program, or funding period. It should not be an exercise in defending the original decision or assigning blame. The review asks whether the investment was sensible given the information available at the time, what actually happened, why it happened, and what should change.

5.1 Recommended review sequence

1. Reconfirm the original decision, objectives, assumptions, and approved budget.
2. Reconcile actual expenditure with finance records and supplier invoices.
3. Validate data sources, definitions, attribution windows, and exclusions.
4. Compare results with the correct baseline or counterfactual.
5. Separate causal evidence from correlation and platform-reported credit.
6. Assess financial, brand, customer, operational, and ethical outcomes.
7. Record deviations, unexpected effects, and limitations.
8. Decide whether to scale, modify, pause, repeat, or stop.
9. Capture reusable learning in an accessible knowledge repository.

5.2 Counterfactual reasoning

The key question is not simply “What happened after the campaign?” It is “What would probably have happened without the campaign?” Possible designs include randomized holdout tests, matched-market tests, difference-in-differences, interrupted time-series analysis, and carefully specified marketing-mix models. Each has assumptions.

For a basic difference-in-differences design, an estimated effect may be represented as:

where Yₜᵣ,post and Yₜᵣ,pre are outcome values for the treated group after and before the intervention, and Y𝚌,post and Y𝚌,pre are corresponding values for the comparison group. The approach depends on assumptions including parallel pre-intervention trends, no major differential shocks, and a stable treatment definition. Without these assumptions, the estimate may be biased.

Post-investment reviews should report evidence strength, not merely a headline number. “Incremental sales estimated at $X, with moderate confidence under a matched-market design” is more informative than “campaign generated $X.”

6. Governance: Decision Rights, Ownership, and Auditability

Governance converts reporting principles into operating rules. A practical model assigns four distinct roles:

• Accountable executive: owns the decision and accepts the business consequence.
• Responsible operator: performs the work and manages execution.
• Consulted specialist: provides expertise, such as legal, privacy, finance, or analytics advice.
• Independent reviewer: checks evidence, compliance, or controls without being rewarded solely for the investment’s success.

These roles can be documented through a RACI matrix, although RACI should not be used to obscure who has final authority. Every major metric should also have a metric owner who approves its definition, monitors quality, and manages changes.

6.1 Auditability and data lineage

An auditable report enables a reviewer to trace a headline result back to:

1. the decision and approved objective;
2. the metric definition;
3. source data and extraction date;
4. transformations and calculations;
5. model version and assumptions;
6. approvals and overrides;
7. published output and subsequent revisions.

Auditability does not require retaining every raw data element indefinitely. It requires a proportionate, lawful, secure record sufficient to reproduce or explain important decisions. Version control, access logs, reconciliation checks, documented transformations, and change approvals are practical controls.

6.2 Separation of duties

A person or agency that benefits from a favorable result should not be the sole authority validating that result. Separation of duties may involve finance validating spend, analytics validating methodology, procurement reviewing fees, privacy specialists reviewing data use, and an executive committee approving material changes.

7. Agency and Supplier Controls

Agencies often possess specialized data, technology, and execution capability. They may also control reporting interfaces, optimization decisions, and opaque fee structures. Effective governance treats agencies as accountable partners rather than unquestioned sources of truth.

Contracts and operating procedures should clarify:

• ownership and portability of data, audiences, creative assets, and accounts;
• permitted data uses and subcontracting;
• fee, rebate, commission, and media-markup disclosure;
• brand-safety and placement standards;
• fraud detection and invalid-traffic controls;
• service levels and reporting deadlines;
• measurement definitions and independent verification rights;
• security, breach notification, retention, and deletion;
• audit rights and rights to inspect relevant records;
• exit assistance and continuity arrangements.

Performance incentives require special care. If an agency is paid solely on attributed conversions, it may favor cheap retargeting, over-credit itself, suppress inconvenient evidence, or pursue low-quality customers. Balanced scorecards can include verified incremental outcomes, quality and profitability, compliance, transparency, experimentation, and long-term brand indicators.

8. Ethical Accountability and Responsible Data Use

Ethical accountability concerns how marketing decisions affect people, communities, markets, and the legitimacy of the organization. Compliance with law is necessary but not sufficient. A practice may be technically legal yet misleading, discriminatory, intrusive, or inconsistent with reasonable expectations.

8.1 Privacy and purpose limitation

Responsible data use begins with a defined and legitimate purpose. Organizations should collect only data necessary for that purpose, provide meaningful notice, restrict secondary use, protect data throughout its life cycle, and delete or anonymize it when retention is no longer justified. Consent, where required, should be informed, specific, freely given, and withdrawable; it should not be treated as a universal solution to unfair design.

Data governance should address:

• data minimization;
• access controls and least privilege;
• retention and deletion schedules;
• encryption and secure transfer;
• vendor and cross-border risk;
• individual rights and complaint handling;
• sensitive attributes and re-identification risk;
• privacy impact assessments for high-risk uses.

Aggregated data is not automatically anonymous. Small cells, rare attributes, location trails, and linked datasets can permit re-identification.

8.2 Bias and fairness

Bias may enter through sampling, historical behavior, labels, missing data, proxy variables, model objectives, or deployment decisions. A model can be accurate on average while producing systematically worse outcomes for a subgroup.

Fairness analysis should ask:

• Who is represented and who is absent?
• Are error rates materially different across relevant groups?
• Does the model use protected or proxy attributes?
• Could optimization concentrate exclusion or exploit vulnerability?
• Are outcomes monitored after deployment?
• Is there a meaningful appeal or correction process?

There is no single fairness criterion suitable for every marketing decision. Trade-offs may exist between equal error rates, calibration, equal opportunity, and commercial objectives. The appropriate standard depends on context, harms, legal duties, and stakeholder expectations. These choices should be made explicitly rather than hidden in technical settings.

8.3 Manipulation, dark patterns, and transparency

Marketing becomes ethically problematic when it exploits cognitive vulnerability, obscures material terms, makes refusal difficult, or presents personalized pressure that people cannot reasonably understand. Examples include deceptive countdown timers, preselected consent, confusing cancellation paths, fabricated scarcity, and targeting based on sensitive inferences.

Transparency should be proportionate and useful. A statement that “algorithms may be used” is rarely sufficient. People and decision-makers need to understand the purpose, important data categories, principal logic, material risks, and available choices. Transparency also applies internally: executives should know when a result is modeled, platform-reported, estimated, or independently validated.

8.4 AI-enabled marketing analytics

AI can improve forecasting, creative testing, anomaly detection, audience analysis, and natural-language reporting. It can also magnify privacy risks, encode historical bias, generate fabricated explanations, and create automation bias—the tendency to accept a machine recommendation because it appears objective.

Responsible deployment requires:

• documented use cases and prohibited uses;
• human accountability for consequential decisions;
• representative validation data;
• performance and fairness monitoring;
• prompt, model, and output logging where appropriate;
• protection of confidential and personal information;
• testing against adversarial or misleading inputs;
• clear disclosure when synthetic content or automated decisions materially affect people;
• a mechanism to override, appeal, or discontinue the system.

A language model may summarize a dashboard, but it should not silently invent causal explanations. Generated commentary must be checked against source data and approved by an accountable analyst or manager.

9. Illustrative Case: Northstar Home’s Reallocation Decision

Background

Northstar Home, a fictional omnichannel home-furnishings brand, allocated $2 million to a six-month customer-acquisition program. Its agency dashboard reported a 4.2 return on advertising spend, based largely on view-through and retargeting attribution. Finance, however, observed that contribution margin and repeat purchase were below plan.

Problem

The executive committee had to decide whether to release a further $1 million. The dashboard used different definitions from finance, the agency controlled several data pipelines, and the campaign used a customer list acquired from a third party. The list’s consent and provenance were not fully documented.

Analysis

An independent analytics team reconciled invoices, platform costs, orders, returns, and gross margin. It created a matched-market comparison and excluded canceled orders. The agency’s reported return was not equivalent to incremental contribution-based ROI. A privacy review found that the third-party list lacked adequate evidence for the intended use in one jurisdiction. A fairness review also found that optimization disproportionately excluded older customers because historical digital response was lower.

The independent estimate was:

• Incremental revenue: $3.1 million.
• Applicable contribution margin rate: 35%.
• Incremental contribution margin: $1.085 million.
• Total program cost: $2 million.

Thus:

ROI = ($1.085 million − $2 million) ÷ $2 million = −$915,000 ÷ $2 million = −45.75%.

The estimate was subject to uncertainty because the matched markets were not perfectly comparable. Nevertheless, the evidence did not support scaling the program in its existing form.

Approach and solution

The committee paused the third-party audience, required documented provenance and lawful purpose before reuse, and changed the agency contract so that verified incremental contribution—not platform-attributed conversions—was one component of remuneration. It approved a smaller test using randomized holdouts, introduced age-group monitoring, and required finance and privacy sign-off at each decision gate. The dashboard was redesigned to show platform attribution separately from independent incremental estimates.

Outcome

The revised program produced lower reported conversion volume but improved incremental contribution per dollar. More importantly, executives could distinguish delivery metrics from causal evidence and could identify unresolved privacy and fairness risks before scaling.

Learning points

1. A high platform return is not necessarily a high financial return.
2. Data provenance is part of performance governance, not a separate administrative concern.
3. Independent validation is especially important when suppliers control measurement.
4. Ethical controls can improve strategic decisions by preventing investments that create hidden liabilities.
5. A post-investment review should improve the next decision, not merely judge the last one.

10. Implementation Blueprint

A graduate-level marketing leader can implement the chapter’s principles through the following sequence:

First, establish a measurement charter. Define strategic outcomes, approved financial measures, attribution conventions, materiality thresholds, and prohibited reporting practices.

Second, build a metric registry. Record definitions, owners, sources, formulas, refresh cycles, limitations, and version history.

Third, map decision rights. Identify who approves budgets, changes targeting, releases data, validates results, accepts risk, and stops a campaign.

Fourth, design the reporting cadence. Use operational reports for rapid control, monthly performance reviews for resource management, and post-investment reviews for learning and accountability.

Fifth, introduce evidence gates. Require quality checks before launch, mid-course review, and scale decisions. The required evidence should be proportionate to investment size and risk.

Sixth, test incentives. Examine whether internal or agency rewards encourage over-attribution, unsafe targeting, excessive frequency, or concealment of uncertainty.

Seventh, preserve challenge. Encourage analysts, privacy specialists, finance professionals, and affected stakeholders to question assumptions without retaliation.

Challenges and Limitations

Governance systems can become bureaucratic. Excessive approvals may slow experimentation and encourage informal workarounds. The solution is proportionality: low-risk reversible tests should not face the same process as large-scale sensitive-data programs.

Measurement is also imperfect. Brand effects may emerge over long periods, customer journeys cross channels, and no single model can resolve every attribution question. Governance should therefore preserve analytical pluralism: experiments, surveys, econometric models, qualitative evidence, and financial reconciliation can inform one another.

Finally, ethical review cannot eliminate all harm. Stakeholder values conflict, data can be incomplete, and future consequences are uncertain. Responsible accountability is a continuing practice of identifying risks, documenting judgments, monitoring outcomes, and correcting course.

Chapter Summary

Managerial reporting is an organizational control system connecting strategy, resources, activity, outcomes, and governance. An executive business case should state the decision required, strategic rationale, financial assumptions, measurement design, risks, alternatives, and accountability. ROI and NPV are useful only when return, cost, timing, and assumptions are defined consistently.

Dashboards should be designed around decisions and should distinguish strategic outcomes, commercial results, leading indicators, and operational controls. Metric definitions, data lineage, uncertainty, and visual integrity are essential. Variance reports should identify not only the size of a difference but also its cause, materiality, owner, and corrective action.

Post-investment reviews should reconcile costs, evaluate counterfactual evidence, distinguish correlation from causation, and convert findings into future decisions. Governance requires clear decision rights, metric ownership, separation of duties, auditability, and agency controls. Ethical accountability requires privacy protection, fairness analysis, transparency, resistance to manipulation, and responsible AI oversight. The strongest marketing organizations treat trustworthy measurement not as a reporting preference but as a condition of sound strategic management.

Key Terms

Audit trail: A documented record linking a decision or report to its data, calculations, approvals, and revisions.

Attribution: The assignment of credit for an outcome to channels, campaigns, or touchpoints.

Counterfactual: The outcome that would likely have occurred without the intervention.

Dashboard: A structured visual display of selected information for monitoring and decision-making.

Data lineage: The traceable history of data sources, transformations, calculations, and outputs.

Decision rights: Explicit authority to approve, modify, execute, or stop an activity.

Difference-in-differences: A quasi-experimental method comparing changes over time between treated and comparison groups.

Incrementality: The portion of an outcome caused by an intervention beyond what would otherwise have occurred.

Metric owner: The person accountable for a metric’s definition, quality, interpretation, and controlled change.

Materiality: The significance of an issue or variance for financial, strategic, legal, operational, or ethical decisions.

Net present value: The present value of expected future cash flows less the initial investment.

Post-investment review: A structured assessment of an investment’s results, assumptions, evidence, risks, and learning.

Privacy by design: Integrating privacy protection into systems, processes, and decisions from the outset.

Responsible data use: Data practices that are lawful, necessary, secure, fair, transparent, and proportionate.

Return on investment: A ratio comparing net incremental return with investment cost.

Separation of duties: Dividing responsibilities so that one party does not control incompatible stages of a process.

Variance: The difference between an actual result and a defined reference value.

View-through attribution: Crediting an outcome to an advertisement viewed without a recorded click; it requires careful validation because exposure may not establish causation.

Review Questions

Short Answer Questions

1. What are the descriptive, diagnostic, predictive, and prescriptive functions of managerial reporting?
2. Why must a business case distinguish incremental revenue from incremental contribution margin?
3. What information should be included in a metric specification record?
4. Explain the difference between a budget variance and an attribution variance.
5. Identify two controls that improve agency accountability.

Long Answer Questions

1. Design an executive business-case structure for a proposed brand-building investment and explain how uncertainty should be communicated.
2. Explain how dashboard design can either support or manipulate managerial judgment.
3. Discuss the role of counterfactual reasoning in post-investment reviews, including the assumptions of difference-in-differences analysis.
4. Evaluate the relationship between decision rights, metric ownership, separation of duties, and auditability.
5. Discuss how privacy, fairness, transparency, and accountability should shape AI-enabled marketing analytics.

Analytical / Application Questions

1. A campaign costs $400,000, generates $1.2 million in incremental revenue, and has a 30% contribution margin rate. Calculate contribution-based ROI and explain two reasons why the result may still be insufficient for a scale decision.
2. An agency reports a 5.0 return on advertising spend, while an independent matched-market analysis estimates a 1.4 return based on incremental contribution. Identify possible reasons for the difference and recommend a governance response.
3. Design a post-investment review for a personalized advertising program using sensitive customer data. Specify the evidence, decision rights, privacy controls, fairness checks, and escalation conditions required before continuation.

Multiple Choice Questions with Answers

Question 1

Which feature most directly distinguishes an accountable marketing report from a descriptive dashboard?

A. It contains more visualizations

B. It links evidence to owners, decisions, and reviewable controls

C. It reports only positive results

D. It uses a real-time data refresh

Correct Answer: B. It links evidence to owners, decisions, and reviewable controls

Question 2

A campaign costs $240,000 and produces $360,000 in incremental contribution margin. What is its contribution-based ROI?

A. 20%

B. 40%

C. 50%

D. 67%

Correct Answer: C. 50%

Question 3

Which statement about a platform-reported conversion is most accurate?

A. It always proves incremental impact

B. It is necessarily equivalent to a finance-reconciled order

C. It may be useful operational evidence but should not automatically be treated as causal proof

D. It cannot be used in any marketing report

Correct Answer: C. It may be useful operational evidence but should not automatically be treated as causal proof

Question 4

What is the principal purpose of a metric owner?

A. To design all campaign creative

B. To approve a metric’s definition, quality, interpretation, and controlled changes

C. To replace the finance function

D. To maximize every metric regardless of strategy

Correct Answer: B. To approve a metric’s definition, quality, interpretation, and controlled changes

Question 5

Which comparison is represented by a budget variance?

A. Actual spend versus authorized spend

B. Treated-group change versus control-group change

C. Actual sales versus market share

D. Brand awareness versus customer satisfaction

Correct Answer: A. Actual spend versus authorized spend

Question 6

A difference-in-differences estimate depends importantly on which assumption?

A. All customers must have identical preferences

B. Treatment and comparison groups must have parallel pre-intervention trends

C. The campaign must produce a positive result

D. No data may be collected after the campaign

Correct Answer: B. Treatment and comparison groups must have parallel pre-intervention trends

Question 7

Which agency-control practice most reduces the risk of self-serving attribution?

A. Allowing the agency to define all metrics without review

B. Paying solely for platform-attributed conversions

C. Using independent validation and separating execution from verification

D. Removing audit rights from the contract

Correct Answer: C. Using independent validation and separating execution from verification

Question 8

Which practice best reflects data minimization?

A. Collecting every available customer attribute for possible future use

B. Collecting only data necessary for a defined and legitimate purpose

C. Retaining identifiable data indefinitely

D. Sharing raw customer data with every supplier

Correct Answer: B. Collecting only data necessary for a defined and legitimate purpose

Question 9

Why can a model be accurate on average yet unfair?

A. Average accuracy guarantees equal outcomes

B. Errors and impacts may differ systematically across groups

C. Fairness is unrelated to data quality

D. Models cannot use historical data

Correct Answer: B. Errors and impacts may differ systematically across groups

Question 10

What is a key requirement for responsible AI-enabled marketing analytics?

A. Removing human accountability

B. Accepting generated explanations without checking source data

C. Documenting the use case, monitoring performance and fairness, and preserving accountable human oversight

D. Optimizing only for the cheapest conversion

Correct Answer: C. Documenting the use case, monitoring performance and fairness, and preserving accountable human oversight