- Who should make the final decision about whether the expected fraud reduction justifies the customer-service impact?
Answer: The chief operating officer, after considering input from analytics, investigations, compliance, and customer service.
When a question describes multiple stakeholders with different roles — technical experts, gatekeepers, operational owners, and affected parties — you're being tested on decision rights and governance: who has the legitimate authority to make a final business trade-off, versus who merely informs it.
The key detail here is the governance charter, which explicitly assigns the chief operating officer final authority over the business trade-off. Once you see that the proposed threshold is already legally compliant and within capacity constraints, the remaining decision is purely a business judgment: is the fraud reduction worth the customer-complaint increase? That cross-functional balancing act — weighing financial risk, operational impact, and customer experience — is precisely what the COO role exists to resolve. Answer C is correct because it respects both the formal authority structure and the principle that major trade-offs require input from all affected parties before a decision-maker rules.
Answer A tempts you into thinking technical ownership equals decision authority. The analytics team quantified the trade-off, but quantifying a trade-off is not the same as being empowered to resolve it — that's a common governance mistake. Answer B misapplies the compliance function's role: compliance holds veto power over illegal actions, but the passage tells you the threshold is already legally compliant, so compliance's gatekeeping role is satisfied and doesn't extend to pure business judgment calls. Answer D makes the opposite error of A — it elevates the most affected party to the decision seat. Customer service's concerns are valid input, but bearing consequences doesn't grant authority.
A useful exam pattern: gatekeepers constrain decisions; they don't make them. Whoever the charter designates as the final authority on business trade-offs is your answer.
- What should the director do under the stated decision rule?
Answer: Choose a supplier now because the study's expected net value after cost and delay is negative.
When evaluating whether to gather more information before a decision, the core concept is Expected Value of Information (EVI). The key rule here is simple: only delay if the net benefit of additional information is positive — meaning you must subtract all costs of delay, not just the direct study fee.
Here, the study improves decision value by $90,000 on average, but it costs $25,000 to conduct and forfeits $80,000 in savings due to the delay. The true net value is:
90,000−25,000−80,000=−15,000
Because the net value is negative, the stated decision rule requires choosing a supplier now — making A the correct answer.
B is a classic trap: it only subtracts the direct study cost (90,000−25,000=+65,000) and ignores the opportunity cost of delay. In real procurement decisions, delay costs are just as real as invoice costs. C makes a qualitatively true but analytically incomplete argument — yes, information can have strategic value, but the decision rule is quantitative, and the numbers don't support delay here. D is simply wrong in principle; analytics remain useful even under deadlines. The question is whether the numbers justify acting on them, not whether analytics apply at all.
Your strategy tip: whenever a question gives you an information-value scenario, list every cost — direct costs and opportunity costs — before computing net benefit. Exam questions are specifically designed to tempt you into ignoring one or the other. - What is the most appropriate analytical recommendation?
Answer: Report the conflict to decision owners and request an authorized revision to the capacity limit, funding level, safety-stock policy, or service-level expectation before proceeding.
When multiple constraints govern a business decision simultaneously, your role as an analyst is not to pick the most convenient one and ignore the rest — it's to surface the conflict and escalate it to the people with authority to resolve it. That's the core concept being tested here.
In this scenario, four parameters are in direct tension. The demand-based policy calls for 10,800 units (10,000+8% safety stock). Finance caps spending at $180,000÷$18=10,000 units. The warehouse caps volume at 9,500 units. And merchandising still wants a 95% service level — a target with no authorized formula linking it to any inventory quantity below policy. No single order quantity satisfies all four simultaneously. The right move is C: report the conflict to decision owners and request an authorized revision before proceeding. Choosing an order quantity unilaterally would mean quietly overriding at least one constraint without authorization.
Answer A is wrong because finance covering exactly forecast demand doesn't dissolve the safety-stock policy — it simply means you can't fund the safety stock. That's a conflict, not a resolution. Answer B is wrong because selecting the warehouse cap as "tightest" and stopping there ignores the finance ceiling and the service-level obligation; binding constraints must all be acknowledged, not rank-ordered and truncated. Answer D is wrong because no organizational authority has established that service level overrides finance or capacity limits — an analyst cannot make that prioritization call unilaterally.
Your study tip: whenever a question presents multiple constraints that cannot all be satisfied, the analytically correct answer almost always involves escalation and authorized decision-making, not picking the constraint you prefer. - Which implementation recommendation best reflects the stakeholders, constraints, and decision context?
Answer: Deploy Model Y only to consenting customers because it satisfies the return constraint and best meets finance's objective.
When a question gives you multiple stakeholders with different objectives, your job is to find the option that satisfies all hard constraints first, then optimizes among feasible choices. Think of it as a filtering problem: eliminate options that violate any constraint, then apply the objective function to what remains.
Here, three constraints govern the decision: (1) returns may increase by no more than 1 percentage point (customer experience), (2) finance wants the highest contribution-margin gain among feasible models, and (3) privacy counsel requires personalization only for consenting customers. Model X increases returns by 3 percentage points — a clear violation of the 1-point ceiling — so it is eliminated from consideration regardless of its revenue upside. Model Y increases returns by only 0.5 percentage points, staying within the constraint. Among feasible models, Model Y also delivers the superior contribution-margin improvement (6% vs. 4%), satisfying finance's objective. Deploying it only to consenting customers satisfies privacy counsel. That logic confirms B as the correct answer.
A fails because it selects Model X despite its 3-percentage-point return increase, which violates the customer-experience constraint — revenue size cannot override a binding operational limit.
C is tempting because Model Y passes the return threshold, but deploying it to all customers ignores the privacy constraint requiring consent before personalization.
D misreads the constraint. Customer experience requires returns increase by no more than 1 percentage point — it does not require zero increase. Model Y's 0.5-point increase is fully acceptable.
Study tip: On multi-stakeholder questions, always eliminate constraint violations before optimizing. A dominant metric (like revenue) never overrides a hard constraint — treat constraints as knockout criteria first. - What is the most appropriate first response from the analytics team?
Answer: Clarify each decision, time horizon, action owner, and success criterion before designing separate analytical deliverables.
When two stakeholders use the same word — "margin analysis" — it's tempting to assume they want the same thing. This question tests whether you recognize that identical terminology can hide fundamentally different analytical needs, and that designing a solution before understanding the decision is a classic analytics mistake.
The CFO wants to understand why something already happened — a diagnostic question rooted in the past. The VP of Sales wants to know who to contact next week — a predictive or prescriptive question oriented toward future action. These require different data, different models, and different outputs. Answer D is correct because clarifying the decision, time horizon, action owner, and success criterion for each request is the only way to design deliverables that actually serve each stakeholder. Skipping this step risks delivering something technically correct but practically useless.
Answer A contains a kernel of truth — diagnostic work often does precede predictive work — but it applies that logic as a rigid rule and ignores that both requests may need to run in parallel given real-world deadlines. Sequencing decisions belong after scoping, not before it.
Answer B mistakes a shared metric for a shared purpose. Gross margin on a dashboard means something different to someone explaining history versus someone prioritizing sales calls. One number, two completely different analytical frameworks.
Answer C jumps straight to building a model, skipping requirements entirely. In analytics, actionability without accuracy to the actual decision need creates waste — or worse, bad decisions made confidently.
Your study tip: whenever a question describes multiple stakeholders requesting "the same thing," treat that as a signal to question whether the underlying decisions are truly aligned before any solution is proposed.
- What recommendation is most consistent with the pre-established decision context?
Answer: Do not launch under the existing rule; require an authorized reassessment or a revised design before deployment.
When you see an A/B test question involving pre-agreed launch criteria, your first instinct should be to treat those criteria as a binding contract, not a suggestion. Steering committees establish guardrails precisely to prevent individual stakeholders from selectively interpreting results after the fact.
Here, the committee set two conditions for launch: a conversion improvement AND a support contact rate increase of no more than 10%. The test delivered a conversion win, but the support rate rose 20% — double the allowed threshold. Because both conditions must be satisfied, the launch criteria fail. The integrity of the pre-commitment process demands that the team honor the agreed rule, seek authorized reassessment, or redesign the experiment. That's exactly what D describes, making it the correct answer.
A is wrong because statistical significance only tells you the conversion result is unlikely due to chance — it says nothing about whether all business conditions for launch are met. Significance is a measurement tool, not a decision override.
B is wrong because partial deployment doesn't automatically fix the guardrail violation. A 20% support rate increase could persist or worsen at any traffic level, and "splitting the difference" was never an authorized option in the pre-agreed framework.
C is wrong because a guardrail violation doesn't invalidate the experiment's data — it simply means the overall business outcome doesn't meet launch thresholds. The experiment was still valid; the product just isn't ready to ship.
Study tip: On business analytics exams, always identify all pre-stated decision criteria before evaluating results. A single guardrail breach blocks launch regardless of how strong other metrics look. - Which threshold should be selected under the stated decision context?
Answer: Threshold B, because it has the highest net value among thresholds satisfying both binding constraints.
When a decision problem involves multiple stakeholders with hard limits, your first move is always to eliminate any option that violates a binding constraint — regardless of how attractive it looks on other dimensions. Here, Legal caps disparity at 5 points and Operations caps referrals at 8%. Both are non-negotiable, so any threshold exceeding either limit is off the table before you even look at net value.
Applying that filter: Threshold A has referrals of 10%, which exceeds the 8% operations cap — eliminated. Threshold D has disparity of 7 points, exceeding the 5-point legal cap — eliminated. That leaves Thresholds B and C as the only feasible options. Between them, B produces $2.3M in net value versus C's $2.2M. Since Finance's objective is to maximize net value among feasible options, Threshold B is the correct choice.
A is wrong because it ignores the operations constraint. Threshold A's 10% referral rate violates the 8% cap, making it infeasible even though its net value of $2.4M looks appealing. C is wrong because "most unused capacity" is not the stated objective — maximizing net value is. Threshold C is feasible but suboptimal; choosing it sacrifices $100,000 in annual value for no required reason. D is wrong because constraints don't have a compensation mechanism — exceeding a hard legal limit cannot be "offset" by higher financial performance.
The key study tip: on constrained optimization questions, always screen for feasibility first, then optimize within the feasible set. Jumping to the highest-value option without checking constraints is the most common trap on these problems. - How should these stakeholder requirements be represented in the optimization model?
Answer: Minimize labor cost subject to break and coverage constraints, with route consistency modeled as a soft preference.
When building optimization models, one of the most critical skills is correctly classifying stakeholder requirements as either hard constraints (non-negotiable) or soft constraints/preferences (desirable but flexible). The passage gives you explicit signals about which is which — your job is to read them carefully.
The finance objective is clear: minimize labor and overtime cost, making it the objective function. Break periods are mandated by a collective-bargaining agreement — legally binding, non-negotiable, so they become hard constraints. Customer contracts requiring minimum route coverage are similarly non-negotiable obligations, also hard constraints. Route consistency, however, is described as a supervisor preference that leadership explicitly permits relaxing when cost rises materially. That language tells you it belongs in the model as a soft constraint or penalty term — honored when free, sacrificed when necessary. This is exactly what C describes.
A is wrong because it collapses all requirements into equally weighted objectives. This ignores the hierarchy — some requirements are legal obligations, not tradeoffs to balance against each other.
B is wrong because it treats route consistency as a hard constraint, directly contradicting the passage's statement that leadership will relax it when costs rise. Treating a flexible preference as non-negotiable will artificially restrict your feasible solution space and raise cost unnecessarily.
D is wrong on multiple fronts: it changes the objective to maximize coverage rather than minimize cost, and it drops both break requirements and route consistency entirely — ignoring real constraints the model must respect.
Study tip: On optimization questions, always map each stakeholder requirement to one of three roles — objective function, hard constraint, or soft constraint — before evaluating the answer choices. The wording of the scenario (legal, contractual, preferred, relaxable) is your roadmap.
- Which performance-measurement approach best aligns the stakeholders with the decision context?
Answer: Use handling time with first-contact resolution and customer outcomes, subject to a disclosure-compliance guardrail.
When a question presents multiple competing stakeholder priorities — cost, customer experience, and compliance — your job is to find the measurement system that balances all of them rather than optimizing for just one. Think of it as a constraint-satisfaction problem: efficiency metrics drive cost goals, outcome metrics protect the customer, and guardrails prevent regulatory exposure.
Answer B does exactly this. Average handling time addresses the manager's labor-cost concern; pairing it with first-contact resolution captures whether rushed calls actually solve problems (directly countering the repeat-contact risk); customer outcome measures satisfy the executives' goal of not shifting costs onto customers; and the compliance guardrail ensures mandatory disclosures aren't sacrificed under time pressure. No stakeholder priority is ignored, and no single metric can be gamed without the others providing a check.
Answer A narrows the comparison group to control for call type — a reasonable statistical refinement — but still relies on handling time alone, leaving first-contact resolution, customer outcomes, and compliance entirely unmonitored. Answer C swings to the opposite extreme: customer satisfaction is valuable, but excluding efficiency metrics gives agents no incentive to manage time, which undermines the manager's cost concerns and ignores compliance tracking entirely. Answer D is creative in removing handling time to prevent rushed service, but total calls completed is essentially a volume metric that indirectly rewards speed anyway, and excluding handling time entirely discards useful operational data.
The strategic takeaway: when stakeholders have conflicting priorities, the right measurement framework is almost never a single metric — look for the answer that includes multiple complementary measures plus a mechanism that prevents one dimension from being sacrificed for another.
- Which selection rule best fits the decision context?
Answer: Filter out all ineligible customers from the ranked list, then contact the highest-ranked 600 eligible customers, drawing from below the original cutoff as needed.
When applying a predictive model to operational decisions, you must distinguish between model ranking (who is riskiest) and eligibility (who can actually be contacted). These are two separate filters, and conflating them leads to poor targeting decisions.
The right approach is to treat the ranked list as a priority queue and remove ineligible customers without shrinking your contact pool. In this scenario, the top 600 includes 70 opt-outs and 30 already-renewed customers — that's 100 ineligible records. Since the objective is to contact the 600 eligible customers with the greatest predicted churn risk, you simply drop the ineligible records and pull the next 100 highest-ranked eligible customers from below the original cutoff. This is exactly what A prescribes: filter first, then fill the 600 slots from the remaining ranked list.
B is wrong because leaving 100 slots unused wastes capacity and ignores customers who ranked just outside the original cutoff — they may still carry meaningful churn risk and are fully contactable. C commits a conceptual error: a high churn-probability score for someone who has already renewed is no longer actionable. Renewal resolves the churn risk regardless of the model score, so retaining them displaces a genuinely at-risk customer. D reverses the correct workflow. Lowering the probability cutoff first to generate a larger pool, then filtering, is unnecessarily complex and risks diluting list quality; eligibility is a constraint on the ranked list, not a reason to re-tune the model threshold.
A useful rule of thumb: rank by risk, then filter by eligibility — never let eligibility shrink your final contact count when qualified candidates remain.
