Performance, Feedback, Learning, and Career Growth

Introduction and Context

In a startup, performance management is not primarily an annual administrative exercise. It is a system for translating strategy into action, learning from uncertain conditions, developing capability, and making fair employment decisions. Startups face rapidly changing priorities, incomplete information, fluid roles, and limited managerial capacity. These conditions make rigid corporate appraisal systems particularly unsuitable. They do not, however, eliminate the need for clarity, accountability, documentation, and equitable treatment.

A lightweight performance system should answer five questions:

1. What outcomes matter most?
2. How will progress and quality be observed?
3. How will employees receive timely guidance?
4. What support is needed to improve capability?
5. What decisions follow when performance is strong, mixed, or persistently inadequate?

The central design challenge is to combine adaptability with consistency. Employees need freedom to experiment, but they also need understandable expectations. Managers need discretion, but not arbitrary standards. Founders need speed, but not decision-making that exposes the firm to bias, legal risk, or loss of trust.

Learning Objectives

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

• design a proportionate performance system for a startup;
• apply goal-setting theory and OKRs without confusing activity with impact;
• distinguish continuous feedback, coaching, reviews, and formal corrective action;
• create useful performance documentation and learning plans;
• construct early-stage career pathways that do not depend only on promotion;
• address underperformance fairly while protecting dignity and due process;
• evaluate the ethical, legal, and technological implications of performance management.

1. Foundations: What “Performance” Means

Performance is multidimensional. A narrow definition based only on revenue or output can reward harmful conduct, ignore teamwork, and penalize employees working on long-term or uncertain initiatives. A more useful model considers at least four dimensions:

• Results: what was achieved, such as qualified customers acquired, defects reduced, or milestones delivered.
• Behaviors and collaboration: how work was conducted, including reliability, communication, inclusion, judgment, and respect.
• Capabilities: knowledge and skills that enable present and future contribution.
• Adaptation and learning: how effectively a person responds to evidence, changing priorities, and failure.

These dimensions should not be treated as perfectly interchangeable. A person may deliver a target through unacceptable conduct; conversely, a responsible experiment may fail while generating strategically valuable learning. Managerial judgment must therefore interpret evidence rather than mechanically total scores.

Performance management is also distinct from performance appraisal. Performance management is an ongoing cycle of expectation-setting, observation, feedback, development, and decision-making. Performance appraisal is a periodic evaluation within that broader cycle. Treating the appraisal as the entire system creates recency bias, anxiety, and administrative overload.

The performance cycle

A practical startup cycle is:

Strategic intent → role expectations → goals and evidence → work and experimentation → feedback and coaching → review and learning → revised goals

The cycle is iterative rather than strictly annual. Goals can be revised when assumptions change, but revisions should be recorded so that employees are not judged against abandoned expectations.

2. Goal-Setting Theory and Startup Objectives

Goal-setting theory, associated especially with Edwin Locke and Gary Latham, argues that specific and challenging goals generally produce higher performance than vague exhortations such as “do your best,” provided that people accept the goals, possess or can acquire the capability to pursue them, and receive feedback. Goals direct attention, mobilize effort, increase persistence, and encourage strategy formation.

The theory has important boundary conditions. Excessive goal difficulty can produce withdrawal or unethical shortcuts. Narrow targets can reduce cooperation and exploration. Goals may also become obsolete when a startup changes its product or market assumptions. Consequently, goal quality depends on both precision and adaptability.

A strong goal normally specifies:

• the intended result;
• the relevant time horizon;
• the evidence or indicator of progress;
• constraints such as quality, safety, compliance, or customer trust;
• assumptions and dependencies;
• the employee’s degree of control.

For example, “increase activation from 28% to 35% among new self-serve accounts by the end of the quarter, while maintaining reported critical defects below the agreed threshold” is more useful than “improve activation.” It identifies an outcome and a quality constraint. The manager must still ask whether the employee controls the relevant product, traffic, and measurement conditions.

Inputs, outputs, outcomes, and impacts

Startups frequently confuse effort with value. A useful hierarchy is:

Inputs and activities are often within an employee’s control; outcomes and impacts are influenced by many factors. Fair evaluation therefore considers controllable process evidence as well as results.

OKRs in a startup context

Objectives and Key Results (OKRs) are a goal-alignment framework. An Objective describes a meaningful direction in qualitative terms. Key Results provide measurable evidence that progress toward the objective is occurring. Initiatives are the actions or projects expected to influence the key results.

Example:

• Objective: Make first-time customer onboarding simple and trustworthy.
• Key Result 1: Increase the percentage of new users completing onboarding from 52% to 70% during the quarter.
• Key Result 2: Reduce median time to first successful use from 18 minutes to 10 minutes.
• Key Result 3: Maintain onboarding-related support escalations below 3% of new accounts.
• Initiatives: usability tests, revised guidance, instrumentation, and support-content improvements.

OKRs are not automatically performance ratings. A stretch key result may be intentionally ambitious, and failure to reach it does not necessarily indicate poor performance. Conversely, achieving a target by damaging quality or trust should not be treated as success. Startups should document whether an objective is a committed operating target, a learning target, or an aspirational stretch target.

A lightweight OKR process can use quarterly or six-to-eight-week cycles:

1. connect team objectives to current strategy;
2. define two to four meaningful objectives;
3. establish two to four key results per objective;
4. identify assumptions, dependencies, and quality safeguards;
5. hold brief progress conversations every two to four weeks;
6. revise goals when strategy or evidence changes;
7. close the cycle with evidence, learning, and next actions.

3. Designing a Lightweight Performance System

A system should be proportionate to organizational maturity. A five-person startup may need a shared goal document and monthly one-to-one conversations. A fifty-person company may require role expectations, calibrated review meetings, a people-operations owner, and a secure record system. The objective is not maximum paperwork; it is reliable managerial practice.

Role expectations

Before setting individual goals, define the continuing responsibilities of the role. A product manager, for example, may be expected to frame customer problems, maintain decision quality, coordinate stakeholders, and communicate trade-offs. These expectations persist even when a particular roadmap changes.

Role expectations should distinguish:

• essential responsibilities;
• decision authority;
• collaboration requirements;
• quality and risk obligations;
• behaviors inconsistent with the organization’s values.

Evidence standards

Good evidence is relevant, sufficiently reliable, and interpreted in context. It can include customer outcomes, quality data, project artifacts, peer observations, decision records, and examples of learning. Metrics should be audited for definition changes, sampling problems, and gaming incentives.

A manager should ask:

• What exactly does the measure capture?
• What does it omit?
• Who controls the result?
• Could the target encourage harmful behavior?
• Is comparison across employees valid?
• What qualitative evidence explains the number?

Check-ins and one-to-ones

A useful check-in is a short, recurring conversation rather than a status interrogation. It may cover:

• progress against priorities;
• evidence and obstacles;
• decisions or support required;
• feedback in both directions;
• learning and energy;
• changes to goals or assumptions.

Managers should record decisions and commitments, not every private detail. Employees should have access to the goals and substantive feedback that affect their evaluation.

Reviews and calibration

A periodic review synthesizes evidence over a defined period. It should answer three questions: what contribution was made, what capabilities or behaviors were demonstrated, and what should happen next? A review should never introduce major concerns for the first time.

Calibration is a structured discussion among managers about standards and evidence. Its purpose is consistency, not forced ranking. Forced distributions are especially risky in small startups because a team may contain no poor performer—or several people may be performing well under different conditions. Calibration should examine role scope, opportunity, resources, and bias indicators.

4. Feedback and Coaching

Feedback is information intended to improve future performance or reinforce effective behavior. It is most useful when it is timely, specific, behavior-based, and connected to consequences or expectations.

A practical feedback structure is:

1. Situation: identify when and where the behavior occurred.
2. Behavior: describe observable action without mind-reading.
3. Effect: explain the consequence for customers, colleagues, quality, or strategy.
4. Next step: agree on a specific continuation, change, or experiment.

For example: “In yesterday’s launch meeting, the risk review was skipped before the release decision. That left support without a mitigation plan and increased avoidable customer exposure. For the next release, please add a documented risk review before approval; I will review the template with you today.”

Feedback should be two-way. Employees can provide information about unclear priorities, insufficient resources, flawed processes, or managerial behavior. A manager who labels every failure as an individual deficiency may be diagnosing a system problem incorrectly.

Coaching versus directing

Directing is appropriate when the task is urgent, regulated, or the employee lacks essential knowledge. Coaching is appropriate when the employee can develop judgment through questions, reflection, practice, and feedback. Coaching questions include:

• What outcome were you trying to produce?
• What evidence supports your interpretation?
• Which assumptions proved wrong?
• What alternatives did you consider?
• What will you do differently next time?
• What support or decision do you need from me?

Coaching is not avoidance. When a standard is clear and a problem is serious, the manager must state the expectation directly and establish a follow-up date.

5. Documentation, Fairness, and Decision Quality

Documentation is a memory and accountability tool, not a surveillance archive. A useful performance record includes the date, expectation, evidence, context, feedback provided, employee response, agreed support, and follow-up. It should avoid speculation, diagnostic labels, irrelevant personal information, and emotionally charged language.

A simple record might state:

“On 12 March, the agreed customer-research milestone was two usability sessions and a synthesis memo. Two sessions occurred, but the memo was not delivered by the agreed date. The employee explained that access to participants was delayed. We revised the date to 18 March and agreed that the employee would escalate recruitment constraints by day three of future cycles.”

This record is more useful than “not proactive.”

Fairness has at least three dimensions:

• Procedural fairness: consistent opportunity to understand expectations, respond, and improve.
• Distributive fairness: outcomes such as ratings, pay, or promotion are proportionate to contribution and role scope.
• Interactional fairness: people are treated with dignity and given an honest explanation.

Managers should monitor common distortions: recency bias, similarity bias, halo and horns effects, attribution error, leniency or severity, and proximity bias in hybrid work. Structured evidence prompts and review panels can reduce, but not eliminate, these problems.

Confidential records should be stored with controlled access and retained according to applicable law and organizational policy. Performance data should not be casually repurposed for unrelated monitoring. Employees should know what information is collected, how it is used, and how they can challenge factual errors.

6. Learning Plans and Career Growth

Learning is not simply attendance at courses. It is a change in capability demonstrated through application. A practical learning plan identifies:

• the capability to develop;
• why it matters to the role or strategy;
• the current evidence and gap;
• a learning method;
• an application opportunity;
• a measure or artifact of progress;
• a review date;
• the manager’s support.

The 70–20–10 idea—experience, relationships, and formal education—is a useful heuristic, not a universal law. Its proportions should not be treated as scientifically fixed. In startups, high-value learning often occurs through stretch assignments, pairing, customer exposure, incident reviews, mentoring, and deliberate reflection.

Example learning plan:

Career growth should not be reduced to promotion. It may include increased scope, deeper expertise, greater autonomy, mentoring, temporary assignments, lateral movement, or a decision to remain at the current level with meaningful work. A career pathway describes increasing contribution, not merely accumulating years.

A concise progression framework can assess:

• scope: individual task, project, team, or organization;
• complexity: routine, ambiguous, or system-level problems;
• influence: direct execution, cross-functional alignment, or strategic direction;
• autonomy: supervision required and quality of judgment;
• capability: depth, breadth, and ability to develop others.

Levels must be anchored in observable contribution. Titles should not be granted solely to retain someone if the organization cannot support the promised authority, compensation, or career expectations.

7. Addressing Underperformance Fairly

Underperformance is a gap between reasonable, communicated expectations and observed contribution. It should be distinguished from misconduct, incapacity, temporary health-related limitations, organizational failure, and legitimate disagreement about strategy. The response should fit the cause.

A fair process commonly follows these stages:

1. Clarify the gap. State the expected standard, observed evidence, impact, and relevant time period.
2. Diagnose causes. Explore capability, clarity, resources, workload, dependencies, health or accommodation needs, and managerial practices.
3. Provide support. Offer training, mentoring, tools, prioritization, or access to information.
4. Set a written improvement plan. Specify outcomes, behaviors, support, check-in dates, and consequences if improvement does not occur.
5. Review evidence. Consider progress, changed circumstances, and the employee’s response.
6. Decide proportionately. Continue, revise, reassign, or end employment in consultation with appropriate HR or legal expertise.

An improvement plan is not a punishment document. It should be achievable, time-bounded, and based on factors the employee can reasonably influence. “Improve attitude” is not a valid standard; “send a weekly risk update by Friday and raise blockers within one business day” is more observable.

Serious misconduct, safety risks, harassment, fraud, or deliberate policy violations may require immediate investigation and different procedures. Managers should not improvise legal conclusions. Employment law varies by jurisdiction, and issues involving protected characteristics, disability, family leave, whistleblowing, union rights, or retaliation require qualified advice.

8. Innovation, Adaptability, and the Risk of Over-Measurement

Performance systems can either support innovation or suppress it. If employees are rewarded only for predictable delivery, they may avoid uncertain but valuable work. If goals change without explanation, employees may experience goalpost shifting. The solution is not to abandon accountability but to evaluate both execution and learning quality.

A startup can use an experiment review that asks:

• Was the hypothesis meaningful?
• Was the method appropriate and ethical?
• What did the evidence show?
• Were risks managed?
• Was learning shared and incorporated?
• What decision followed?

A failed experiment can represent strong performance when the question was important, the method was sound, and the learning improved decisions. Conversely, repeated poorly designed experiments should prompt coaching even if the team occasionally obtains a favorable result.

Metrics should be treated as signals rather than complete representations of performance. Over-measurement produces administrative burden, gaming, and tunnel vision. A small number of balanced indicators, supplemented by qualitative judgment, is generally more effective.

9. Technology and AI in Performance Management

Digital tools can make goals visible, automate reminders, aggregate project evidence, and support learning recommendations. They can also create serious risks. Communication frequency is not equivalent to contribution; code volume is not equivalent to software quality; dashboard activity is not equivalent to customer value.

AI-generated summaries or talent recommendations require careful governance. Organizations should establish:

• a clearly stated purpose and lawful basis for data use;
• data minimization and access controls;
• validation for disparate impact and measurement error;
• human review of consequential decisions;
• notice and, where appropriate, employee opportunity to respond;
• audit trails and correction procedures;
• restrictions on sensitive or irrelevant personal data.

An algorithm should not silently convert proxies—such as meeting attendance, writing style, or online presence—into judgments about commitment or potential. Managers remain accountable for decisions even when software supplies recommendations.

10. Illustrative Case: Northstar Analytics

Background

Northstar Analytics is a fictional, 32-person business-to-business startup offering automated reporting for independent retailers. Its engineering team releases quickly, but customer-reported defects have increased. Product managers complain that engineers are judged by delivery speed, while engineers say priorities change weekly. The founder proposes a numerical quarterly rating based on tickets closed.

Problem

The proposed metric would reward volume and could encourage small, low-value tickets while discouraging careful architecture and defect prevention. Employees also report that feedback arrives only during annual compensation discussions. One engineer, Maya, has missed two integration milestones, but the milestones changed after a major customer request and the relevant API documentation was incomplete.

Analysis

The leadership team maps the work system. It finds three distinct issues:

1. expectations and dependencies are not documented;
2. the measurement system rewards output rather than customer outcomes and quality;
3. managers delay difficult conversations, so employees receive little opportunity to adjust.

Maya’s situation cannot be judged fairly from milestone completion alone. The team examines agreed scope, changes in requirements, available support, code-review evidence, and communication about risks.

Approach

Northstar adopts a lightweight system:

• each role has four to six continuing expectations;
• teams set quarterly objectives with two to four key results;
• individual priorities identify dependencies and quality safeguards;
• managers conduct monthly development check-ins;
• a shared evidence log records goals, feedback, and decisions;
• quarterly reviews summarize results, behaviors, learning, and next steps;
• managers calibrate standards without forced ranking.

For engineering, the team uses a balanced set of evidence: delivery of valuable increments, escaped-defect patterns, review quality, incident learning, and collaboration. It does not assign a single formulaic score.

Maya’s manager explains the gap, acknowledges changing requirements, and agrees on a six-week development plan. Maya will lead one integration slice, document assumptions before implementation, raise dependency risks within two working days, and pair with a senior engineer on interface design. The manager will provide stable acceptance criteria and ensure API access. Progress is reviewed weekly.

Outcome

Maya completes the revised integration, reduces rework in the next cycle, and produces a reusable interface checklist. Northstar also discovers that two other employees had experienced the same documentation problem. The company improves its planning process rather than treating the issue solely as individual failure. Ticket volume is retained as an operational signal but removed as the primary performance measure.

Learning points

• A metric can be easy to count and still be strategically misleading.
• Fair evaluation requires attention to control, context, and dependencies.
• Feedback is most effective before a formal review.
• Development plans should include reciprocal managerial commitments.
• Performance systems can reveal process defects as well as employee gaps.

11. Implementation Roadmap for Startup Leaders

A founder or people leader can implement the system in four stages.

Stage 1: Define the minimum standard. Document organizational values, role expectations, decision rights, and unacceptable conduct. Train managers in feedback and documentation.

Stage 2: Establish the operating rhythm. Introduce team objectives, individual priorities, recurring one-to-ones, and brief written records. Avoid launching complex software before the habits exist.

Stage 3: Add development and review. Create learning plans, career-level descriptions, periodic reviews, and calibration practices. Test whether employees understand how decisions are made.

Stage 4: Audit and improve. Examine completion rates, employee perceptions, promotion and pay patterns, attrition, grievance themes, and evidence quality. Disaggregate outcomes where lawful and appropriate to identify possible inequity.

The system should be reviewed whenever strategy, workforce size, work design, or legal requirements materially change.

Challenges and Limitations

No performance system can fully isolate individual contribution in interdependent work. Outcomes are affected by market timing, team composition, resource allocation, customer behavior, and luck. Excessive standardization can undermine judgment; excessive discretion can create favoritism. Quantitative metrics can clarify attention but cannot replace interpretation. Career frameworks can support transparency but may become bureaucratic or culturally narrow. Finally, feedback quality depends heavily on manager capability, psychological safety, and the organization’s willingness to hear uncomfortable information.

The appropriate response is disciplined humility: use multiple sources of evidence, disclose assumptions, invite employee voice, revise flawed measures, and distinguish confidence in a conclusion from certainty.

Chapter Summary

Effective startup performance management is a continuous system linking strategy, role expectations, evidence, feedback, development, and fair decisions. Goal-setting theory supports specific and challenging goals, but goals require acceptance, capability, feedback, and safeguards against harmful narrowing. OKRs can align teams when objectives express direction, key results provide evidence, and initiatives remain distinct from outcomes.

A lightweight system uses clear role expectations, recurring check-ins, timely feedback, coaching, periodic reviews, and proportionate documentation. Fairness requires attention to procedure, outcomes, dignity, bias, context, and employee control. Learning plans should connect capability gaps to practice and evidence, while career pathways should describe increasing scope, complexity, influence, autonomy, and capability rather than assuming promotion is the only form of growth.

Underperformance should be addressed through clarification, diagnosis, support, measurable improvement expectations, follow-up, and proportionate decisions. Innovation is protected when sound experimentation and learning are recognized, not only predictable delivery. Technology and AI may assist administration, but consequential judgments require transparency, data governance, human accountability, and safeguards against proxy discrimination. The strongest systems are simple enough to use, rigorous enough to be fair, and adaptable enough for uncertainty.

Key Terms

Performance management: An ongoing process of setting expectations, observing work, giving feedback, developing capability, and making decisions.

Performance appraisal: A periodic synthesis or evaluation of performance evidence.

Goal-setting theory: A theory proposing that specific, challenging goals can improve performance under appropriate conditions.

Objective: A qualitative statement of desired direction or achievement.

Key Result: A measurable indicator of progress toward an objective.

Initiative: An action, project, or intervention intended to influence a key result.

Outcome: A change resulting from an output or activity.

Impact: A broader strategic or stakeholder effect.

Continuous feedback: Timely information exchanged during work rather than reserved for a formal review.

Coaching: A developmental process using questions, reflection, practice, and support to improve judgment or capability.

Calibration: A structured comparison of standards and evidence across managers or teams.

Evidence standard: A criterion for judging whether information is relevant, reliable, and sufficient.

Learning plan: A documented plan connecting a capability goal to development actions and evidence of progress.

Career pathway: A description of how scope, complexity, autonomy, influence, and capability may increase over time.

Underperformance: A gap between reasonable communicated expectations and observed contribution.

Improvement plan: A time-bounded agreement specifying performance gaps, expectations, support, follow-up, and possible consequences.

Procedural fairness: Fairness in the process used to reach a decision.

Measurement error: The difference between what a metric records and the construct it is intended to represent.

Review Questions

Short Answer Questions

1. What distinguishes performance management from performance appraisal?
2. Define an Objective, Key Result, and Initiative in the OKR framework.
3. Why should ticket volume or activity counts rarely serve as a complete performance measure?
4. What information should a useful performance record contain?
5. Name two forms of career growth that do not involve promotion.

Long Answer Questions

1. Explain how goal-setting theory can improve startup performance while also creating risks of tunnel vision or unethical behavior.
2. Design a lightweight performance cycle for a 25-person startup, including goals, check-ins, reviews, documentation, and calibration.
3. Compare feedback, coaching, formal review, and an improvement plan. Explain when each is appropriate.
4. Discuss how a startup can evaluate innovation and learning without abandoning accountability for results.
5. Analyze the ethical and governance requirements for using AI-generated performance recommendations.

Analytical / Application Questions

1. A salesperson exceeds a revenue target but makes misleading claims that increase customer complaints. How should the organization evaluate the performance, and what evidence should it use?
2. An employee misses a product milestone after priorities change three times and a dependency team fails to deliver. Construct a fair diagnostic and improvement approach.
3. Design two objectives and three key results for a startup seeking faster onboarding while protecting product quality and customer trust. Explain why each measure is appropriate and what it may fail to capture.

Multiple Choice Questions with Answers

Question 1

Which statement best distinguishes an OKR key result from an initiative?

A. A key result is a measurable indicator; an initiative is an action intended to influence it.

B. A key result is always a task; an initiative is always a qualitative aspiration.

C. A key result describes employee personality; an initiative describes compensation.

D. A key result is used only for annual ratings; an initiative is used only for hiring.

Correct Answer: A. A key result is a measurable indicator; an initiative is an action intended to influence it.

Question 2

Which practice best supports fair evaluation when an employee’s target changes during a performance cycle?

A. Judge the employee against the original target without modification.

B. Ignore the target and rely only on the manager’s general impression.

C. Record the changed assumptions, revised expectations, and relevant evidence.

D. Assign the employee the lowest rating because adaptability cannot be measured.

Correct Answer: C. Record the changed assumptions, revised expectations, and relevant evidence.

Question 3

Which example is most behaviorally specific feedback?

A. You need to be more professional.

B. You are not strategic enough.

C. Your attitude is a problem.

D. In the launch meeting, you skipped the risk review, leaving support without a mitigation plan.

Correct Answer: D. In the launch meeting, you skipped the risk review, leaving support without a mitigation plan.

Question 4

What is the primary purpose of calibration?

A. To force every team into an identical ranking distribution.

B. To improve consistency in interpreting standards and evidence.

C. To replace managers’ judgment with an automated score.

D. To ensure that no employee receives developmental feedback.

Correct Answer: B. To improve consistency in interpreting standards and evidence.

Question 5

Which item is most appropriate in a learning plan?

A. A vague instruction to show more initiative.

B. A capability gap, development action, application opportunity, and evidence of progress.

C. A guaranteed promotion date regardless of performance.

D. A list of courses with no connection to role requirements.

Correct Answer: B. A capability gap, development action, application opportunity, and evidence of progress.

Question 6

Why can ticket volume be a poor standalone engineering performance metric?

A. It may reward quantity while ignoring value, quality, complexity, and collaboration.

B. It cannot be recorded in any project-management system.

C. It measures customer outcomes more accurately than any other evidence.

D. It always decreases when engineers work on important projects.

Correct Answer: A. It may reward quantity while ignoring value, quality, complexity, and collaboration.

Question 7

Which response is most appropriate as the first step in addressing possible underperformance?

A. Immediately terminate employment without discussing the issue.

B. Clarify the expected standard, evidence of the gap, and relevant context.

C. Reduce the employee’s pay informally.

D. Announce the concern to the entire organization.

Correct Answer: B. Clarify the expected standard, evidence of the gap, and relevant context.

Question 8

What does procedural fairness primarily concern?

A. Whether the decision process gives people a reasonable opportunity to understand and respond.

B. Whether every employee receives exactly the same outcome.

C. Whether performance data are collected without any human judgment.

D. Whether promotions occur at identical time intervals.

Correct Answer: A. Whether the decision process gives people a reasonable opportunity to understand and respond.

Question 9

Which statement best reflects responsible use of AI in performance management?

A. AI recommendations should automatically determine termination decisions.

B. Online presence is a reliable substitute for contribution.

C. AI outputs require validation, human review, access controls, and correction procedures.

D. Employees should not be told that their data are being analyzed.

Correct Answer: C. AI outputs require validation, human review, access controls, and correction procedures.

Question 10

A sound experiment fails to produce the expected result but uses an important hypothesis, appropriate methods, and shares useful learning. How should it generally be evaluated?

A. Automatically as poor performance because the result was negative.

B. As potentially strong performance, with attention to method, risk management, and learning.

C. As misconduct because experimentation is unpredictable.

D. Only by counting the number of meetings held during the experiment.

Correct Answer: B. As potentially strong performance, with attention to method, risk management, and learning.