Chapter 1.1 — Video Recording Script

Welcome to Introduction to Data Analytics. Imagine that it is Monday morning, and your manager comes to your desk with a question: “Our retail sales have dropped by twenty percent. Can you tell me what happened, why it happened, and what we should do next?” You open a spreadsheet and see hundreds of numbers. Where do you start? This is the everyday challenge that data analytics helps us solve. In this lesson, we will move from raw numbers to meaningful business questions, learn the four types of analytics, and work through a realistic example together. You do not need to know programming to begin. You need curiosity, clear thinking, and a willingness to check the evidence.

Let us begin with a straightforward definition. Data analytics is the process of collecting, organizing, examining and interpreting data so that we can answer questions and make better decisions. Think about the difference between a number and an insight. A single transaction of two thousand rupees is data. When we combine transactions and discover that September sales were eight lakh rupees, we have information. When we compare that figure with August and realize that the online channel explains most of the decline, we begin to have an insight. That insight can help a manager decide what to investigate first. Finally, the business takes action and measures whether the decision worked. So remember this chain: data, information, insight, decision, action. The objective is not simply to build attractive charts. It is to support useful, responsible business decisions.

Data analytics is often divided into four useful approaches. The first is descriptive analytics. It answers: What happened? For example, we may report that monthly revenue fell from ten lakh to eight lakh rupees. Descriptive analytics uses totals, averages, charts and reports to summarize performance. The second is diagnostic analytics, which asks: Why might this have happened? We might compare online and in-store sales, examine product categories, or look at conversion rates. This is where we investigate possible drivers. But an important warning: a pattern or correlation is not the same as proof of cause. The third is predictive analytics. It asks: What could happen next? Using sufficient relevant data and suitable models, analysts might estimate next month's demand or forecast customer churn. Forecasts are estimates, not guarantees. The fourth is prescriptive analytics. It asks: What actions should we consider? A business might compare the costs and expected benefits of redesigning checkout, improving stock availability, or adjusting marketing campaigns. The recommendation depends on our assumptions and constraints. Here is the simple way to remember all four: descriptive is what happened, diagnostic is why it may have happened, predictive is what could happen next, and prescriptive is what we could do about it. In real projects, these approaches work together rather than operating as isolated steps.

Before we analyze anything, we must know what data we are working with. A business may collect transaction data from a sales system, customer details from a CRM, website activity from an analytics platform, or feedback from surveys. Some data is structured, like an Excel table with clearly named columns. Some is semi-structured, like JSON events. Other information is unstructured, such as free-text reviews or recorded calls. We also distinguish numerical measures, such as revenue and order count, from categories, such as region or product type. Time matters as well, because comparisons need a meaningful period. We can explore smaller datasets using Excel or Google Sheets, query larger structured data with SQL, build interactive dashboards in Power BI or Tableau, and carry out advanced analysis with Python or R. The tool is important, but choosing the right question, checking data quality and interpreting results are even more important.

Here is a practical workflow you can use in almost any entry-level analytics assignment. Start by defining the business problem. What decision are we supporting? Who needs the answer? What does success look like? Next, identify and collect the right data. Third, validate the data: look for missing values, duplicated rows, inconsistent labels, incorrect dates or unusual values. Fourth, explore the data. Compare periods, calculate important metrics, segment results and test ideas. Fifth, explain the findings with a clear story and an appropriate chart. Sixth, recommend the next action, while acknowledging what is uncertain. Seventh, measure the outcome and refine your understanding. Throughout the process, follow privacy and security rules. For example, you should not expose personal customer information simply because it is available in the dataset. One of the marks of a strong analyst is being able to say, “This is what the evidence shows, and this is what we still need to confirm.”

Now let us return to Meridian Retail, our fictional company. In August, revenue was ten lakh rupees. In September, it was eight lakh rupees. To find the percentage change, subtract the old value from the new value, divide by the old value, and multiply by one hundred. The result is minus twenty percent. That answers what happened, but it does not explain where the problem lies. Let us split revenue by channel. Online sales went from six lakh to four lakh forty thousand rupees. Store sales went from four lakh to three lakh sixty thousand rupees. The online channel lost one lakh sixty thousand rupees, while stores lost forty thousand rupees. That means online contributed eighty percent of the total two-lakh decline. Now look at order count. Across both channels, the number of orders fell from five hundred to four hundred. Average order value is revenue divided by orders. Ten lakh divided by five hundred is two thousand rupees. Eight lakh divided by four hundred is also two thousand rupees. So the average amount per order did not change. The fall in total revenue is explained arithmetically by fewer orders. Can we look more deeply? Yes. The online channel recorded six thousand visitors in both months, but its orders fell from three hundred to two hundred and twenty. That means online conversion fell from five percent to approximately three point six seven percent. This gives us a useful line of investigation. Did customers experience checkout problems? Did traffic quality change? Were popular products unavailable? Our sample also shows that online marketing spend was cut in half. That could be relevant, but we cannot say it caused the decline just from these two months. We need more evidence. A responsible analyst distinguishes what the numbers demonstrate from what still needs to be tested.

Let us summarize the key ideas. Data analytics starts with a meaningful business question and uses reliable evidence to support a decision. Descriptive analytics tells us what happened. Diagnostic analytics explores why it may have happened. Predictive analytics estimates what might happen next. Prescriptive analytics evaluates what action to consider. In our retail example, we found a twenty percent revenue decline, discovered that online sales explained most of it, and identified lower online conversion as an important area to investigate. Notice that we did not pretend to know the cause without evidence. That is an essential professional habit. Your next step is to download the practice dataset, reproduce the figures in Excel, and create a short business summary. Then take the quiz and complete your analyst briefing. Congratulations on starting your data analytics journey. In the next chapter, we will build on these foundations and become even more precise about solving business problems with data.

In this walkthrough, we will behave like an analyst who has been asked to investigate a sudden change in sales. Open the Meridian Retail dataset and start by checking its structure. We have two months, two channels, two product categories, and one row for each combination. The fields include visitors, completed orders, revenue and marketing spend. Before we start, let us check that the numbers are imported as numbers and that the labels are consistent. Now summarize revenue by month. August totals ten lakh rupees; September totals eight lakh. We calculate the percentage difference as minus twenty percent. Next, summarize by channel. Online falls by one lakh sixty thousand rupees, and Store falls by forty thousand. Online therefore accounts for eighty percent of the overall revenue decline. This is an important business finding, because it tells us where to focus our investigation first. Let us also examine completed orders and average order value. Orders decreased from five hundred to four hundred, but revenue per order stayed at two thousand rupees. That means the revenue decline is consistent with a decline in order volume. Finally, consider the online conversion rate. Divide online orders by online visitors for each month. Five percent becomes approximately three point six seven percent. We have now moved beyond a headline revenue change to a more informative pattern. But remember: we have not identified the root cause. We should check traffic sources, the checkout funnel, inventory, pricing, promotions and tracking quality. A strong analyst ends with a measured recommendation rather than an unsupported conclusion. Our recommendation is to prioritize investigating the online purchasing journey and then test improvements while monitoring orders, conversion and revenue.

Welcome to your first hands-on data analytics lab. Today, we will recreate the retail sales findings using Excel. First, download the CSV file from this chapter and import it into a new workbook using Data, From Text or CSV. Check that the headers appear in the correct columns and that revenue, visitors and orders are numbers. Save your workbook with a clear name. Next, take a moment to check for blank cells, duplicated records and inconsistent month names. In a professional environment, this data validation step helps prevent inaccurate conclusions. Now create a summary area. To calculate total August revenue, use SUMIF with the Month column as the criteria range and Revenue_INR as the sum range. Repeat for September. Your results should be ten lakh and eight lakh rupees. Calculate the percentage change by dividing the difference between the September and August figures by August revenue. Format the result as a percentage. Next, use SUMIF to total orders for each month. Divide revenue by order count to obtain average order value. You should see two thousand rupees in both months. Now use SUMIFS to filter by both month and Online channel. Calculate online orders, online visitors and the conversion rate. The rate should decrease from five percent to approximately three point six seven percent. Finally, insert a PivotTable. Put Month in rows, Channel in columns, and the sum of revenue in values. Insert a clustered column chart to compare online and store sales across the two months. Give your chart a useful title and label the revenue units. Before you finish, explain the result in four sentences: what changed, which channel mattered most, what you still need to investigate and what you recommend next. This is how a simple spreadsheet can become a professional analytics deliverable. Pause the video whenever you need to repeat a step, and use the provided answer checks to verify your calculations.