1 Foundations of Business Analytics

The analytics workflow

Business analytics turns a business need into evidence that can guide action. It begins by identifying the outcome to improve, the people affected, and the decision that needs support, then carries the work through analysis, interpretation, action, and monitoring.

The workflow is a guide rather than a one-way checklist. Data limitations, unexpected results, or stakeholder feedback may require revisiting the question, measures, or assumptions. In practice, work commonly cycles among understanding the business, working with data, evaluating findings, and putting results into practice.

Frame an answerable question

A broad goal such as “increase customer satisfaction” does not yet specify an analysis. A well-framed question identifies the outcome that needs to change, the population and scope, the decision the analysis will inform, and constraints such as budget, time, or policy.

For example, a retailer could ask: “Which delivery issues should the operations team address first to reduce late deliveries for online orders next quarter, without exceeding the current service budget?” This specifies a population, a time frame, a decision, and a constraint. It points toward relevant evidence without prescribing a particular analytical technique.

Distinguish the Business problem from the Analytical problem. The business problem describes the practical need; the analytical problem describes what evidence is needed to support it. A need to reduce late deliveries might lead to an analysis comparing delay rates across carriers and regions. Analytics can inform a decision, but cannot by itself resolve an operational issue such as insufficient delivery capacity. Agreement with affected stakeholders helps ensure the work addresses the right need.

Define metrics and safeguards

A Metric is a defined measure of performance or progress. A Key performance indicator (KPI) is a metric selected as especially important to a business objective. For a measure to be interpretable and repeatable, define what it represents, how it is calculated, which cases and dates are included, where the data comes from, and who maintains the definition. Also establish the current level and the level that would count as improvement.

For a rate, specify both the numerator and the denominator: what is counted and what serves as the comparison base. For example, define the on-time delivery rate as eligible online orders delivered by the promised date divided by all eligible online orders delivered during the measurement period. State how canceled orders, partial deliveries, and missing promised dates are handled. Without such rules, teams can report different values under the same metric name.

Use measures that connect the desired outcome to possible side effects. A retailer seeking faster delivery might track on-time delivery as its primary measure and monitor shipping cost per order and customer complaints as Guardrails. Faster service is not an unqualified improvement if costs rise sharply or service quality worsens elsewhere. Establish a Baseline before an intervention so that later results can be compared with the starting point.

Interpret findings and guide action

An analysis has practical value when its findings help a person or team choose among available actions. Before analyzing, clarify who will make the decision, which options are available, when the choice is due, and what evidence could change the choice. This focus helps avoid producing a result with no clear next step.

Suppose an analysis finds a notably higher late-delivery rate in one region, while also showing that the region had unusually high order volume. A useful interpretation explains the pattern, notes relevant limitations, and identifies possible responses, such as reviewing staffing or carrier capacity there. The analysis does not automatically prove which response will work. Decision makers may need a controlled trial or further evidence, and should monitor both the main measure and the guardrails after acting.

A Decision brief can connect evidence to action by stating:

1. Decision needed: What choice is being considered, who will make it, and by when?

2. Evidence: What did the analysis find relative to the baseline?

3. Meaning and limits: What can the findings support, and what remains uncertain?

4. Recommended action: What should be done, and why is it preferable to alternatives?

5. Follow-up: Which metrics will be monitored, and when will the result be reviewed?

The workflow links a business need to an answerable question and measures that make success observable. Data preparation, analysis, interpretation, and implementation follow, with opportunities to revise earlier choices as new information emerges. Clear metric definitions, baselines, stakeholder agreement, and an explicit connection between findings and decisions help analysis support action rather than merely produce numbers.

• Guardrail