Introduction to Data Analytics

Chapter 1.1 · Introduction to Data Analytics

Every business produces information about what it does: a customer places an order, a delivery arrives late, a visitor leaves a website, or an invoice is paid. Individually, these records may seem ordinary. Together, and examined carefully, they can reveal how a business performs, where it faces difficulties, and what decisions could improve its results. This process is the foundation of data analytics.

Consider a retail business that earned ₹10,00,000 in August and ₹8,00,000 in September. Its owner knows that revenue declined, but the figure alone does not explain where the decline occurred, what might have influenced it, or what to do next. A data analyst turns a vague concern into specific questions, examines reliable evidence, and communicates findings in a form the business can use.

1.1.1 Why data analytics matters

Organizations make decisions every day. Retailers decide which products to replenish. Banks evaluate transaction patterns and operational risks. Hospitals review patient waiting times. Logistics companies examine delivery delays. Human resources teams study hiring and attrition. In each case, a decision may be improved by relevant, accurate and timely information.

Without a disciplined approach, people may rely too heavily on intuition, a few memorable experiences, or incomplete reports. Data analytics does not remove human judgement; it makes that judgement better informed. It provides a structured way to ask what changed, how large the change was, which parts of the business were affected, and which explanations are worth testing.

The value of analytics is therefore not measured solely by the number of dashboards produced. A useful analysis changes understanding or improves a decision. It might prevent unnecessary spending, identify a service problem early, improve inventory planning, or help an executive choose between competing priorities.

1.1.2 What is data analytics?

Data analytics is the systematic process of defining a question, obtaining relevant data, preparing and examining that data, interpreting the results, and communicating evidence to support a decision. It may involve simple arithmetic in a spreadsheet or advanced statistical and computational models. The underlying purpose remains the same: turn records into useful understanding.

It is helpful to distinguish several related terms. Data are recorded observations, such as the date and value of a sale. Information is data that have been organized or summarized for a purpose. An insight is an interpretation that improves understanding of a problem. A decision is the choice the organization makes; an action is the implementation of that choice.

For example, a collection of invoices is data. Adding their values by month produces a revenue report, which is information. Discovering that most of a decline is concentrated in one sales channel produces an insight. Deciding to inspect that channel's customer journey is a decision; running a diagnostic review is an action. Good analysts do not confuse these stages.

1.1.3 The four types of data analytics

Analytics is commonly described in four complementary forms. They are not rigid steps that every project must follow, but they are useful ways of organizing business questions.

Descriptive analytics: What happened?

Descriptive analytics summarizes information that already exists. It answers questions about performance over a defined period or within a defined group. Typical techniques include totals, percentages, averages, trend charts, frequency tables and dashboards.

If a sales report shows that revenue fell from ₹10 lakh to ₹8 lakh, it provides a descriptive result. It tells us the size of the change, but not its underlying cause. Most organizations begin here because it establishes a common view of what happened.

Diagnostic analytics: Why might it have happened?

Diagnostic analytics examines differences, segments and relationships to investigate an observed result. The analyst might break down sales by channel, product, region or customer type; review changes in conversion; or compare periods with and without promotions. This activity is often called drill-down analysis.

Suppose online revenue declined much more sharply than store revenue. That fact directs attention to the online channel, but it does not prove which factor caused the decline. Price changes, tracking errors, product availability and checkout problems could all be relevant. Diagnostic analysis helps form and evaluate explanations; stronger claims of cause require stronger evidence.

Predictive analytics: What could happen next?

Predictive analytics estimates future or unknown outcomes using historical and other relevant information. Examples include demand forecasts, expected delivery times, customer churn estimates and predictions of equipment failure. Methods range from simple trend analysis to statistical models and machine learning.

A prediction should always be interpreted with its assumptions and uncertainty. A forecast of next month's sales is not a promise about what will happen. A business may use several scenarios—for example, stable conversion, improved conversion, and further decline—to plan sensibly.

Prescriptive analytics: What should we consider doing?

Prescriptive analytics helps decision-makers compare possible actions in light of business objectives, costs, constraints and risks. It may use optimization, simulations, decision rules or scenario analysis.

For a retailer, possible interventions could include improving checkout performance, changing promotional targeting, or adjusting stock levels. A prescriptive analysis compares their plausible effects and resource requirements. It does not mean that software can guarantee the correct decision.

1.1.4 Understanding data: structure, meaning and sources

Before analyzing a dataset, an analyst must understand what its records represent. A row may describe one sale, one customer, one product, one day, or a monthly summary. This is called the grain of the data. Confusing the grain can produce incorrect totals or misleading conclusions.

Structured data follow a relatively fixed arrangement of fields, as in a table of orders with columns for order date, channel and amount. Semi-structured data, such as JSON event logs, contain named fields but may vary between records. Unstructured data include free-text reviews, images and audio recordings. Data can also be grouped by meaning: numerical measurements, categories and dates or timestamps.

Common business sources include point-of-sale systems, enterprise resource planning systems, customer relationship management systems, accounting ledgers, website analytics platforms, operational applications and surveys. When combining sources, it is essential to check whether fields refer to the same entity, period, currency and business definition.

For example, one system may define a customer as anyone who has registered, while another counts only customers who completed a purchase. Their figures cannot safely be treated as identical without reconciling the definitions.

1.1.5 From a vague concern to a measurable question

A strong analytics project starts with the decision that a stakeholder needs to make. “Sales are down” is an observation; “Which channels contributed most to the revenue decline between August and September, and which changes should we investigate?” is a more useful analytical question.

Before performing calculations, the analyst should agree on the reporting period, the relevant sales measure, the business entities included and the comparison baseline. Gross sales, net sales, order count and units sold are different measures. So are unique visitors, sessions and store footfall.

A metric is a defined measure, such as monthly revenue. A key performance indicator (KPI) is a metric selected because it reflects progress toward an important business objective. Every KPI should have an owner, definition, calculation method, reporting frequency and intended use.

Four introductory business metrics

These formulas are simple, but their definitions matter. AOV should use compatible revenue and order counts for the same period. Conversion requires a consistent definition of a visit and an order. A ratio may become misleading if its numerator and denominator come from different populations.

1.1.6 Worked business example: Meridian Retail

Meridian Retail is a fictional company created for this chapter. It sells electronics and home products through a website and physical stores. A manager reports that September 2026 revenue fell by 20% compared with August and asks the analyst where to focus attention.

The first step is to establish the size of the change. August revenue was ₹10,00,000 and September revenue was ₹8,00,000.

This is a descriptive finding. The analyst then breaks revenue down by sales channel.

Online sales declined by ₹1,60,000, while store sales declined by ₹40,000. Online therefore accounted for ₹1,60,000 ÷ ₹2,00,000 × 100 = 80% of the overall revenue decline. This does not mean that online sales fell by 80%. The online channel's own change was approximately 26.7%; the 80% figure describes its contribution to the total decline. Distinguishing those two percentages is an important analytical skill.

Next, examine order volume. Meridian Retail completed 500 orders in August and 400 in September. Average order value remained ₹2,000 in both months: ₹10,00,000 ÷ 500 and ₹8,00,000 ÷ 400. In other words, the lower revenue coincided with fewer completed orders rather than a change in the overall average value per order.

Online recorded 6,000 visitors in each month. It completed 300 orders in August and 220 in September. Using the exercise's consistent visitor measure, online conversion moved from 300 ÷ 6,000 × 100 = 5.00% to 220 ÷ 6,000 × 100 ≈ 3.67%. The conversion rate fell by about 1.33 percentage points, not 1.33%.

A responsible recommendation is therefore to verify the underlying tracking, review daily online funnel performance, examine inventory availability, and check whether traffic sources or customer behavior changed. Any proposed intervention should be tested and its impact measured rather than assumed.

1.1.7 The end-to-end data analytics process

Although tools and techniques differ, most useful analytics projects follow a recognizable cycle. First, the analyst defines the decision and establishes the questions and success criteria. Second, the analyst collects data from approved sources. Third, the data are validated and prepared; duplicates, missing values, impossible values and inconsistent labels are examined.

Fourth, the analyst explores and analyses the data. This may involve summaries, charts, grouped comparisons, formulas and more sophisticated methods. Fifth, findings are interpreted and communicated in a form the audience can understand. Sixth, the business decides and acts. Finally, the team measures results to learn whether the decision helped.

The process is iterative. Discovering that a definition is unclear may send the analyst back to data collection. An unexpected trend may require a revised question. Good analytical work is not a race to the final chart; it is a series of evidence-based checks.

1.1.8 Tools and professional responsibilities

Many entry-level analysts begin with Excel or Google Sheets because spreadsheets make tables, formulas, PivotTables and charts accessible. SQL is commonly used to retrieve and aggregate structured data from relational databases. Power BI and Tableau help communicate results through reports and interactive dashboards. Python and R support automation, statistical analysis and more complex workflows.

Tool proficiency matters, but it is only part of the role. A capable analyst clarifies ambiguous requirements, questions suspicious results, explains assumptions, documents calculations, protects confidential information, and speaks in language the business understands. A technically correct report that answers the wrong question is still unsuccessful.

For instance, the Meridian Retail analysis should not expose named customer records merely to demonstrate a revenue trend. Aggregate figures are sufficient for the introductory problem. In more sensitive settings, analysts must follow organizational controls, applicable laws, data minimization principles and appropriate permission procedures.

How data analysts, business analysts and data scientists differ

These titles overlap in practice, and exact responsibilities depend on the employer. However, they point to different areas of emphasis. A data analyst usually examines existing business data, validates measures and produces reports or insights. A business analyst often works most closely with stakeholders to define requirements, understand business processes and translate needs into changes. A data scientist generally applies statistical modelling, experimentation and machine learning when questions require advanced inference or prediction.

In Meridian Retail, the business analyst may clarify the commercial director's decision and reporting definitions; the data analyst may reconcile sales, compare channels and explain the observed decline; and a data scientist may help forecast demand or evaluate an experiment if the question requires it. They share responsibility for sound evidence and clear communication, rather than operating in isolated silos.

1.1.9 From calculation to a compelling business explanation

An effective analytical explanation usually begins with the business finding, supports it with the most relevant evidence, distinguishes observation from hypothesis, and ends with the next recommended step. It is not necessary to present every formula before explaining the conclusion.

For example, instead of saying “I created a PivotTable and used SUMIFS,” the analyst might tell the Sales Manager:

This version is concise, numerical and careful. It tells the manager what happened and what should be investigated without claiming that the analyst has already established a cause. The calculations remain available if the manager asks to see the evidence.

1.1.10 Chapter summary

Data analytics connects recorded activity with business understanding and decision-making. Descriptive analysis establishes what happened; diagnostic analysis investigates possible explanations; predictive analysis estimates what may happen; and prescriptive analysis evaluates potential actions. Each relies on suitable data, consistent definitions, and sound interpretation.

The Meridian Retail example demonstrated how a headline 20% revenue decline becomes a more informative account: online sales contributed most of the decline, the number of orders fell, and online conversion weakened. Those findings suggest a useful investigation, but they do not by themselves prove why the change occurred.

In the next activities, you will reproduce these calculations using Excel, test your understanding, and present an evidence-based analyst briefing. Together, these steps begin to develop the professional habits required in a data analytics role.