Business Requirements, KPIs and Decision Context

Introduction

Business intelligence does not begin with a dashboard, a database, or a visualization tool. It begins with a decision that someone needs to make. If the decision is poorly understood, even technically sophisticated analysis may produce information that is irrelevant, misleading, or impossible to act upon.

This chapter examines how business analysts translate stakeholder needs into analytical requirements, critical success factors (CSFs), key performance indicators (KPIs), targets, and balanced scorecards. It also distinguishes strategic, tactical, and operational decisions. These distinctions are foundational because the same data can have different meanings at different organizational levels. A chief executive may need a quarterly view of profitability, a regional manager may need a weekly view of sales performance, and a service supervisor may need an hourly alert about unresolved customer cases.

A useful analytical deliverable therefore connects five elements:

1. a business objective;
2. a decision or action;
3. evidence required to support that decision;
4. a measurable indicator and target; and
5. a communication format appropriate to the decision-maker.

The objective is not to measure everything. It is to measure what helps an organization understand whether its objectives are being achieved and what action should follow.

Learning Objectives

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

• distinguish business requirements, analytical requirements, CSFs, KPIs, metrics, measures, targets, and thresholds;
• distinguish strategic, tactical, and operational decisions;
• translate stakeholder questions into testable analytical requirements;
• design well-defined KPIs with appropriate formulas, data sources, owners, frequencies, and targets;
• explain the logic and limitations of the balanced scorecard;
• identify problems caused by poorly designed or misused KPIs;
• apply a structured requirements-to-dashboard process to a business case; and
• evaluate ethical, governance, and implementation issues in performance measurement.

1. From Business Need to Analytical Requirement

1.1 Business requirements

A business requirement states what the organization needs to achieve or what problem it needs to solve. It is expressed in business terms rather than technical terms. For example:

The distribution division must reduce late deliveries while maintaining acceptable delivery cost and customer satisfaction.

This statement does not yet specify a database, chart, statistical method, or software feature. It identifies an outcome that matters to the organization.

A business requirement should normally clarify:

• the business problem or opportunity;
• the stakeholders affected;
• the decisions that must be made;
• the desired outcome;
• the relevant time horizon; and
• constraints such as budget, regulation, or service commitments.

1.2 Analytical requirements

An analytical requirement specifies the information, analysis, or capability needed to support a business requirement. For the distribution example, analytical requirements might include:

• calculate the percentage of deliveries arriving after the promised date;
• compare on-time performance by depot, route, carrier, and customer segment;
• identify the main operational causes of delay;
• show weekly trends for the previous twelve months; and
• provide an exception alert when performance falls below the agreed threshold.

Analytical requirements should be specific enough to test. “Provide useful delivery information” is too vague. “Provide weekly on-time delivery by depot and carrier, with drill-down to delivery order, for the previous twelve months” is more testable.

A requirements traceability matrix helps connect organizational intent to implementation.

1.3 Asking the right questions

A disciplined analyst converts broad stakeholder statements into decision-oriented questions. The following sequence is useful:

1. What objective is the stakeholder trying to achieve?
2. What decision will be made using the analysis?
3. What action could follow each possible result?
4. Which population, period, location, product, or customer group is relevant?
5. What comparison is needed: target, prior period, benchmark, forecast, or control group?
6. What level of detail and frequency are appropriate?
7. What definitions and data-quality standards apply?
8. Who owns the decision and who owns the data?

This approach prevents a common failure: collecting large volumes of data before establishing why the data is needed.

2. Decision Context: Strategic, Tactical and Operational

Decision context determines the appropriate data, time horizon, analytical method, and visualization.

2.1 Strategic decisions

Strategic decisions shape the long-term direction and competitive position of an organization. They are usually made by senior leaders and have broad consequences. Examples include entering a new market, changing the business model, investing in a new facility, or repositioning a product portfolio.

Strategic analysis generally uses aggregated and historical information, external market evidence, scenarios, forecasts, and risk analysis. The time horizon may be several years. Strategic dashboards should emphasize trends, relationships, uncertainty, and progress toward major objectives rather than minute-by-minute events.

2.2 Tactical decisions

Tactical decisions translate strategy into programmes, resource allocations, and management actions. They are often made by departmental or regional managers over a monthly, quarterly, or annual horizon. Examples include setting regional sales targets, adjusting inventory policies, selecting a marketing campaign, or scheduling staff capacity.

Tactical analysis commonly uses comparisons across teams, products, channels, or regions. Managers need variance analysis, drill-down, forecasts, and explanations of performance differences.

2.3 Operational decisions

Operational decisions concern the execution of current processes. They are often frequent, structured, and time-sensitive. Examples include assigning today’s delivery routes, responding to a stock-out, approving a service escalation, or investigating a failed transaction.

Operational analytics requires current, granular data and clear action rules. Alerts, exception reports, workflow queues, and near-real-time monitoring may be more useful than a highly designed executive dashboard.

The levels are related but not interchangeable. An operational metric may explain a tactical result, while a tactical result may contribute to a strategic outcome. However, an organization should not expect an operational dashboard to answer a strategic question without suitable aggregation and context.

3. Critical Success Factors and Performance Measures

3.1 Critical success factors

A critical success factor is a condition that must be achieved or maintained for an organization, strategy, or process to succeed. For a subscription software company, CSFs might include reliable service availability, rapid customer adoption, renewal of contracts, and sustainable acquisition cost.

CSFs are not necessarily numerical. They express what is strategically important. They guide the selection of measures but should not be confused with measures themselves.

3.2 Measures, metrics and indicators

A measure is an observed or calculated value, such as 4,820 units sold or 97.2% service availability. A metric is a defined quantitative measure used for analysis or management. An indicator is a signal that helps users interpret a condition or trend.

In practice, organizations often use these terms interchangeably. Conceptually, however, the important distinction is between a raw value and a managed measure with a clear definition, purpose, owner, and decision rule.

3.3 Key performance indicators

A KPI is a carefully selected performance indicator that is directly linked to a significant organizational objective or CSF. Not every metric is a KPI. The number of website visits may be informative, but it becomes a KPI only if it is meaningfully connected to a defined objective, such as qualified demand generation, and if a decision will be based on it.

A strong KPI has the following characteristics:

• relevance: it relates directly to an objective or CSF;
• clarity: its definition is understood consistently;
• measurability: required data can be obtained with acceptable quality;
• controllability: the responsible team can influence it, at least partly;
• timeliness: it is available at a frequency suitable for the decision;
• comparability: it can be compared across time, units, or targets;
• balance: it does not encourage harmful optimization of one dimension at the expense of another; and
• actionability: a plausible management response exists.

A KPI specification should identify the measure owner, business owner, formula, numerator and denominator, inclusion and exclusion rules, source systems, refresh frequency, target, threshold, dimensions, and known limitations.

4. KPI Formulas and Measurement Design

Quantitative definitions must be precise. Consider on-time delivery rate.

The numerator is the count of eligible deliveries completed on or before the promised date. The denominator is the count of eligible deliveries in the reporting period. “Eligible” must be defined—for example, whether cancelled orders, customer-requested postponements, or force-majeure events are excluded.

If 9,420 of 10,000 eligible deliveries were on time:

The formula is meaningful only when the numerator and denominator use the same population and period. A numerator based on delivered orders and a denominator based on all orders would create a misleading ratio.

Other frequently used formulas include:

The retention formula assumes that customer counts are defined consistently and that “new customers” are identifiable within the period. It is not the same as customer renewal rate when contracts have different renewal dates or when customers can purchase multiple products.

For an average, the analyst should state whether the measure is a simple or weighted average. A simple average of departmental rates gives each department equal weight; a weighted organization-wide rate gives greater influence to departments with more transactions. These answer different questions.

A KPI also requires a denominator discipline. “Average handling time” may be calculated as total handling minutes divided by completed cases, while “cost per order” may require allocation of fixed and variable cost. Changing the denominator changes the meaning of the KPI.

5. Targets, Thresholds and Benchmarks

A target is a desired level of performance by a specified date or period. A threshold is a boundary that triggers attention or action. A benchmark is a reference point, such as prior performance, a competitor comparison, an industry standard, or a best-performing internal unit.

These concepts should not be conflated. A target of 95% on-time delivery may be ambitious; a red threshold below 90% may indicate immediate intervention; and an internal benchmark of 97% may show what the best depot currently achieves.

Targets should be:

• linked to a clear objective;
• time-bound;
• based on historical performance, capacity, strategy, or external evidence;
• sufficiently challenging but not arbitrary;
• reviewed when business conditions change; and
• accompanied by a stated owner and action plan.

A target can create dysfunctional behaviour if it is used without context. For example, a call-centre target based only on short handling time may encourage agents to end calls prematurely, increasing repeat contacts and reducing customer satisfaction. Balanced measures are therefore essential.

6. The Balanced Scorecard

The balanced scorecard, associated with Robert Kaplan and David Norton, organizes performance management across four complementary perspectives:

1. Financial: What financial outcomes must be achieved?
2. Customer: How should customers perceive the organization?
3. Internal process: Which processes must perform well?
4. Learning and growth: What capabilities, people, technology, and culture support future performance?

The scorecard is not simply a dashboard containing four charts. It is a strategic management framework linking objectives, measures, targets, initiatives, and cause-and-effect assumptions.

The perspectives should be adapted to the organization. A public-sector agency may emphasize public value, access, compliance, and service equity. A hospital may include clinical outcomes and patient safety. The framework is a prompt for strategic balance, not a rigid universal template.

A strategy map can show hypothesized relationships: learning and growth may improve process capability; better processes may improve customer outcomes; and customer outcomes may contribute to financial or public-value results. These are hypotheses, not guaranteed causal laws. They should be tested with evidence and reviewed when assumptions change.

7. Requirements-to-KPI Design Process

A practical design process consists of the following stages.

Stage 1: Identify stakeholders and decisions

Interview sponsors, process owners, analysts, data stewards, and end users. Ask what decisions they make, how frequently they make them, and what information they currently lack. Distinguish the person requesting a dashboard from the person accountable for acting on it.

Stage 2: Define objectives and CSFs

Rewrite vague aspirations into outcome-oriented objectives. “Improve service” might become “reduce unresolved high-priority cases while maintaining customer satisfaction.” Identify the conditions that must be present for the objective to succeed.

Stage 3: Specify measures and dimensions

Select a limited set of measures. Define dimensions such as time, region, product, channel, customer segment, or process stage. A measure without a relevant dimension may be impossible to diagnose; too many dimensions may overwhelm users.

Stage 4: Define targets and decision rules

Specify target values, tolerance bands, escalation thresholds, and actions. For example, a KPI may require a manager review when performance falls below 92% for two consecutive weeks. The persistence rule prevents a single data anomaly from causing unnecessary escalation.

Stage 5: Assess data feasibility and quality

Check availability, completeness, accuracy, timeliness, consistency, uniqueness, and validity. Determine whether the source system captures the required fields and whether definitions have changed over time.

Stage 6: Prototype and validate

Use sample data and low-fidelity sketches before developing a complete dashboard. Ask users whether the displayed measures answer their questions and whether the proposed actions are realistic.

Stage 7: Govern and improve

Publish a KPI dictionary, assign ownership, document changes, monitor data quality, and review whether the KPI is still useful. A KPI can become obsolete when strategy, products, processes, or regulations change.

8. Worked Application: Illustrative Retail Case

Background

The following is a fictional illustrative case. Northstar Retail operates physical stores and an online channel. Revenue is growing, but profit is declining. Senior managers believe that discounting, stock-outs, and declining repeat purchases may be contributing factors.

Problem

The organization has many reports but no shared definition of performance. The online team reports conversion rate, stores report sales per employee, finance reports gross margin, and customer service reports complaints. Managers debate which result is most important.

Analysis

The business objective is to increase profitable customer growth. Relevant CSFs are:

• customers can find desired products;
• customers receive a satisfactory experience;
• sales convert without excessive discounting; and
• growth produces an acceptable margin.

The decision context is mixed. The executive team needs a monthly strategic scorecard. Regional managers need weekly tactical comparisons. Store supervisors need daily operational exception information.

A KPI design might include:

Suppose monthly revenue is $2,000,000 and cost of goods sold is $1,300,000. Then:

The result should not be interpreted alone. A higher margin may arise from higher prices, lower discounts, a changed product mix, or reduced sales of strategically important products. The dashboard should therefore show revenue, margin, discount rate, and product mix together.

Approach and solution

Northstar creates a scorecard with one to three measures per perspective rather than displaying every available metric. Each KPI receives a definition, source, owner, refresh schedule, target, and data-quality status. The tactical view compares regions and channels, while the operational view lists products at risk of stock-out and unresolved customer cases.

Outcome

After implementation, managers discover that online conversion is acceptable, but high discounting is concentrated in two categories with poor stock availability. The business changes replenishment priorities and revises promotions. The scorecard does not itself create the improvement; it makes a relevant pattern visible and connects that pattern to an accountable decision.

Learning points

• A KPI must be connected to an objective and decision.
• A single favourable metric can hide an unfavourable trade-off.
• Different management levels require different views of the same business.
• Definitions, ownership, and data quality are part of KPI design, not administrative afterthoughts.

9. Common Failure Modes and Limitations

Measuring activity instead of outcomes

The number of sales calls may indicate effort but not sales quality, customer value, or profitability. Activity measures can be useful leading indicators, but they should be connected to outcomes.

Too many KPIs

A dashboard containing dozens of equally prominent indicators creates cognitive overload and weakens accountability. A focused set of measures is usually more useful than an exhaustive catalogue.

Vanity metrics

A vanity metric looks impressive but does not support a meaningful decision. Total app downloads, for example, may be less useful than active users, retention, or completed transactions.

Perverse incentives

When rewards depend narrowly on one KPI, employees may optimize the measure rather than the underlying objective. This is sometimes called gaming or Goodhart’s law: when a measure becomes a target, it may cease to be a good measure.

Correlation mistaken for causation

A KPI trend can coincide with an outcome without causing it. A rise in training completion and sales may reflect seasonal demand rather than a training effect. Causal claims require suitable analysis and attention to confounding factors.

Inconsistent definitions

If one department counts a customer as active after a login and another counts only a purchase, their “active customer” KPIs cannot be compared. A KPI dictionary and data governance process reduce this risk.

Lagging indicators only

Financial results and customer churn are important but may appear after the opportunity to intervene has passed. Leading indicators such as process compliance, response time, or qualified pipeline can provide earlier signals, although they are not automatically causal.

Aggregation bias

A satisfactory organization-wide average can conceal poor performance for a particular region, demographic group, or product. Analysts should examine relevant segments while protecting privacy and avoiding unjustified comparisons.

10. Governance, Ethics and Responsible Performance Measurement

KPI systems influence behaviour, resource allocation, promotion, and sometimes employment decisions. They therefore require governance.

Important controls include:

• documented definitions and calculation logic;
• role-based access to sensitive data;
• audit trails for changes and overrides;
• validation of data completeness and accuracy;
• review of proxy measures for bias;
• appropriate aggregation and anonymization of personal data;
• clear retention and deletion policies; and
• a process for challenging an inaccurate or unfair measure.

A KPI based on customer location, demographic attributes, or employee productivity may create discriminatory effects if used without context. Privacy regulations may restrict the collection, combination, or disclosure of personal information. Analysts should apply data minimization: collect and use only what is necessary for a legitimate purpose.

Artificial intelligence can support requirement discovery, anomaly detection, forecasting, and natural-language explanations. However, AI-generated insights do not remove the need for human accountability. Users should understand the data used, the uncertainty of predictions, the possibility of bias, and the action being recommended. Automated alerts should include escalation and override procedures, especially where decisions affect employment, credit, healthcare, or access to essential services.

11. Contemporary Practice and Best-Practice Principles

Modern organizations increasingly combine descriptive, diagnostic, predictive, and prescriptive analytics. A KPI dashboard may show current performance, explain variance, forecast likely outcomes, and recommend interventions. The progression should be controlled: a recommendation is only as reliable as the data, model, assumptions, and decision rules supporting it.

Effective practice includes the following principles:

1. Start with decisions, not charts.
2. Use a small, balanced set of meaningful KPIs.
3. Define every KPI in a shared dictionary.
4. Match frequency and granularity to the decision horizon.
5. Show target, actual, variance, trend, and context together.
6. Combine leading and lagging indicators.
7. Design for action, including ownership and escalation.
8. Test data quality before interpreting performance.
9. Examine distributions and segments, not only averages.
10. Review whether KPIs still reflect current strategy.
11. Treat ethical, privacy, and fairness risks as design requirements.
12. Evaluate whether the information changes decisions and improves outcomes.

Chapter Summary

Business intelligence is valuable when it supports a real decision. Business requirements describe organizational needs; analytical requirements specify the evidence and capabilities needed to address those needs. Strategic decisions concern long-term direction, tactical decisions concern plans and resource allocation, and operational decisions concern current execution. Each requires an appropriate time horizon, level of detail, and communication format.

Critical success factors describe conditions necessary for success. KPIs are selected, defined, and governed measures linked to important objectives. Good KPI design requires precise formulas, consistent populations, appropriate targets, data-quality checks, ownership, and action rules. Targets, thresholds, and benchmarks serve different purposes and must not be confused.

The balanced scorecard connects financial, customer, internal-process, and learning-and-growth perspectives. It encourages managers to examine trade-offs and strategic relationships rather than optimize one isolated measure. Nevertheless, KPI systems can fail through excessive measurement, inconsistent definitions, perverse incentives, weak data, misleading aggregation, or unjustified causal claims. Ethical governance, privacy protection, transparency, and human accountability are therefore essential.

Key Terms

Business requirement: A statement of an organizational need, objective, or problem to be solved.

Analytical requirement: A specific information, analysis, or reporting capability needed to support a business requirement.

Decision context: The organizational level, time horizon, purpose, and constraints surrounding a decision.

Strategic decision: A long-term decision about organizational direction and competitive or public-value position.

Tactical decision: A medium-term decision about plans, programmes, and resource allocation.

Operational decision: A frequent execution-oriented decision concerning current processes or events.

Critical success factor: A condition that must be achieved or maintained for success.

Measure: An observed or calculated value representing a phenomenon or performance result.

Metric: A defined quantitative measure used for analysis or management.

Key performance indicator: A strategically or operationally important measure linked to an objective and decision.

Target: A desired level of performance to be achieved within a specified period.

Threshold: A boundary that triggers attention, escalation, or action.

Benchmark: A reference point used for comparison.

Leading indicator: A measure that may provide an early signal of future performance.

Lagging indicator: A measure of an outcome that is observed after activity or performance has occurred.

Balanced scorecard: A framework that organizes objectives and measures across complementary performance perspectives.

KPI dictionary: A controlled document defining KPI meaning, formula, source, owner, and governance.

Data quality: The degree to which data is accurate, complete, timely, valid, consistent, and fit for purpose.

Goodhart’s law: The observation that a measure used as a target may cease to be a reliable measure.

Requirements traceability: The ability to connect objectives and decisions to requirements, data, outputs, and outcomes.

Review Questions

5 Short Answer Questions

1. What is the difference between a business requirement and an analytical requirement?
2. Distinguish a KPI from a general metric.
3. What is the purpose of a threshold?
4. Name the four traditional balanced scorecard perspectives.
5. Why should a KPI definition specify its numerator and denominator?

5 Long Answer Questions

1. Explain how strategic, tactical, and operational decisions differ in time horizon, data granularity, and analytical output.
2. Describe a complete process for translating a stakeholder request into a governed KPI.
3. Discuss the advantages and limitations of the balanced scorecard as a performance-management framework.
4. Explain how poorly designed targets can create dysfunctional behaviour, using a business example.
5. Discuss the ethical and governance responsibilities involved in using employee or customer KPIs.

3 Analytical / Application Questions

1. A call centre reports that average handling time has fallen by 20%, but repeat contacts and complaints have increased. Diagnose the likely KPI-design problem and propose a balanced set of measures.
2. Design three KPIs for a university seeking to improve student success. For each, state the objective, formula or definition, target, owner, frequency, and possible limitation.
3. A retailer’s overall on-time delivery rate is 95%, but one high-value customer segment has a rate of 82%. Explain why the aggregate KPI is insufficient and recommend an analytical and management response.

Multiple Choice Questions with Answers

Question 1

Which statement best describes an analytical requirement?

A. The organization should become more customer-focused.

B. The analyst must use a particular dashboard colour scheme.

C. The system must calculate weekly on-time delivery by depot and carrier.

D. The company should increase long-term profitability.

Correct Answer: C. The system must calculate weekly on-time delivery by depot and carrier.

Question 2

Which decision is primarily strategic?

A. Assigning today’s delivery routes

B. Entering a new national market

C. Scheduling next week’s staff shifts

D. Investigating an individual failed transaction

Correct Answer: B. Entering a new national market

Question 3

A company completes 9,420 on-time deliveries out of 10,000 eligible deliveries. What is its on-time delivery rate?

A. 9.42%

B. 90.0%

C. 94.2%

D. 105.8%

Correct Answer: C. 94.2%

Question 4

Which item is a critical success factor rather than a KPI?

A. Customer retention rate of 92%

B. Reliable service availability

C. Gross margin percentage of 35%

D. Average resolution time of 18 hours

Correct Answer: B. Reliable service availability

Question 5

What is the main purpose of a KPI threshold?

A. To define the database structure

B. To trigger attention or action when performance crosses a boundary

C. To replace the business objective

D. To guarantee that performance will improve

Correct Answer: B. To trigger attention or action when performance crosses a boundary

Question 6

Which balanced scorecard perspective focuses most directly on capabilities, skills, and organizational learning?

A. Financial

B. Customer

C. Internal process

D. Learning and growth

Correct Answer: D. Learning and growth

Question 7

Why can an organization-wide average be misleading?

A. Averages always contain calculation errors.

B. Aggregation can conceal poor performance in important segments.

C. Averages cannot be displayed on dashboards.

D. An average is never a valid business measure.

Correct Answer: B. Aggregation can conceal poor performance in important segments.

Question 8

Which is the best example of a leading indicator?

A. Last year’s operating profit

B. Completed customer churn for the previous quarter

C. Percentage of staff completing required process training

D. Revenue reported at year-end

Correct Answer: C. Percentage of staff completing required process training

Question 9

Revenue is $2,000,000 and cost of goods sold is $1,300,000. What is gross margin percentage?

A. 30%

B. 35%

C. 65%

D. 153.8%

Correct Answer: B. 35%

Question 10

What is a major risk of using a single narrow KPI as an employee target?

A. The measure becomes impossible to calculate.

B. Employees may optimize the target while harming the broader objective.

C. The target automatically becomes a benchmark.

D. The KPI can no longer have an owner.

Correct Answer: B. Employees may optimize the target while harming the broader objective.