Lecture video · L01 From Data to Decision

[Slide 1] Welcome to the course. Today we do something slightly unusual for a first day: we are going to build a complete, working business report, end to end, before we define most of the terms for what we are doing. You will load real data, build four visuals, and publish the result to the web. The vocabulary will catch up as we go. By the end of the session you will have seen the entire life of a business intelligence project once, at speed, and the rest of this course is simply slowing down at each step to do it properly.

[Slide 2] Here is the file we will work with all session. It is one year of sales data from Lantern and Pine, a home goods retailer with thirty-eight stores and an online channel. Fifty thousand five hundred and thirty-nine rows — one row for every product on every order placed in twenty twenty-five. Open that file in a spreadsheet and scroll, and you can scroll for twenty minutes without learning one true thing about this business. That gap — between data you cannot read and a page you can — is the entire reason the field we are studying exists. By the end of today, this exact file becomes a single page that answers how much we sold, what carries the business, how the year moved, and how any region compares.

[Slide 3] Business intelligence is the practice of turning an organization's data into information people can act on. Notice that definition says nothing about any particular software. Power BI is our tool, and it is the most widely used one in the field right now, but the thing you are actually here to learn is a practice — a repeatable loop that starts with a question and ends with a decision. Software versions change every month. The loop has not changed in decades.

[Slide 4] The loop has five stations, and because this table matters more than any other slide today, let me read it in full. Station one, the question: someone needs to decide something — should we expand the Southwest stores, should we staff up for December. Station two, the data: find it, clean it, and shape it until it can be trusted — that is Module One of this course, a tool called Power Query. Station three, the model: organize the tables and define the calculations so that questions become answerable — Module Two, where you will learn a language called DAX. Station four, the visuals: choose charts that answer the question honestly — that is Module Three. And station five, the decision: share the report and someone acts on it — Module Four, in the Power BI Service. Today we ride all five stations in one session. The next fourteen weeks are those five stations done slowly and well.

[Slide 5] Before we touch anything, we need to sort out some naming, because Microsoft's names here confuse every single person who meets them for the first time. There are three, and the difference between the first two will matter to you every week of this course.

[Slide 6] Power BI Desktop is a free application that runs on Windows, and it is where reports get built: loading data, modeling it, writing calculations, arranging visuals. It will be your daily tool, and nearly every assignment in this course is built and submitted as a Desktop file. The Power BI Service is the website — app dot powerbi dot com — and it is where reports get shared: you publish from Desktop to the Service, and readers open reports in their browser. And Microsoft Fabric is the umbrella platform that the Service now lives inside — it covers data engineering tools well beyond this course. When you see the word Fabric in a corner of the screen, you are not in the wrong product. The sentence to memorize: Desktop builds, the Service shares, Fabric surrounds.

[Slide 7] Setup, and one honest caveat. Power BI Desktop installs free from the Microsoft Store — search Power BI Desktop, install, done. The Store version keeps itself updated, which is what you want this semester. You do not need any account to build files; your school account matters only at the end of today, when we publish. The caveat: Desktop runs on Windows only. There is no Mac version. If you are on a Mac you have three workable paths, in order of preference: a campus computer lab, a school-provided virtual desktop if your institution offers one, or Windows running in a virtual machine. Talk to me this week if none of those fits, not the night an assignment is due. You also need two course files: the data file, and the course theme, which we will apply near the end of the session.

[Slide 8] Let us make the naming stick with a quick prediction. Your manager wants to see the report you built this morning. Do you send her to Power BI Desktop, or to the Service? Take thirty seconds, then compare with the person next to you. The tell is in what the manager needs to do: she is not building anything — she is reading, filtering, deciding. Readers live in the Service. If your answer was Desktop, ask yourself whether you want your manager installing software and opening your working file, or clicking a link. We will confirm this live at the end of the session when we publish.

[Slide 9] Time to open the application and get oriented. There is surprisingly little to memorize — the interface has three main views, and one habit I want you to build from the very first minute.

[Slide 10] Down the left edge of Power BI Desktop sit three icons. Report view is the canvas — the blank page where visuals will live. Table view shows the actual rows of whichever table you select — think of it as a spreadsheet you cannot type into. And Model view shows your tables as boxes with relationship lines between them; with one table loaded it looks nearly empty today, and in Module Two it becomes the most important screen in the application. Now the habit, and I mean for this to follow you through your whole career: the canvas is not the data. A visual on the canvas is a question you asked. Table view is what the data actually says. When a number ever looks wrong — and I will show you numbers that look wrong all semester — you check the rows before you blame the chart.

[Slide 11] We load data from the Home ribbon: Get data, Text CSV, and choose the course file. Power BI answers with a preview dialog, and I want you to treat that preview as a checkpoint, not a formality. Look at the little icons in each column header: OrderDate should show a calendar — a date type — and Quantity and Revenue should show number types. When a date column loads as text, nothing errors, and your charts quietly stop making sense — we will see exactly how at the end of the session. The dialog offers two buttons. Load takes the file as it is. Transform Data opens Power Query, the cleaning workshop where Lessons Two and Three live. Here is a fact worth remembering about the working world: this file earns the Load button because it was prepared for teaching. Lesson Two is about how rarely that happens outside a classroom.

[Slide 12] Our first visual takes one click. In the Data pane, check the box next to Revenue. Power BI drops a chart onto the canvas — its best guess. We convert it to a card, and the card shows one number: twenty point six six million dollars. Hover it and you get the exact figure — twenty million, six hundred sixty-three thousand, two hundred fifty-two dollars and seventy-two cents. The whole year, in one number. Now notice what you did not do: you never wrote a formula. Power BI saw a numeric column, guessed you wanted a sum, and built one automatically. That automatic guess has a name — an implicit measure — and I want you to hold two ideas about it at once: today it is pure convenience, and in Module Two I will show you exactly where automatic guessing produces confident, wrong answers, and what professionals do instead.

[Slide 13] Before we chart the year, a prediction — and this one has a real payoff. Which month was Lantern and Pine's biggest of twenty twenty-five? Take a minute alone and commit to a month and a reason, then two minutes with your neighbor to argue it out. Most rooms say December, and the reasoning is sensible — holidays, gifts, retail's big season. Keep in mind what this company actually sells: furniture, fire pits, patio sets, lamps, kitchenware. Hold your answer; the chart settles it in a moment.

[Slide 14] Next question: what carries the business? Check Revenue and Category together and Power BI gives a bar chart. Furniture leads at seven point five nine million dollars. Outdoor follows at five point five five million. Then Kitchen and Dining at three point two three million, Lighting at two point six seven, and Decor trailing at one point six one million. Add the top two in your head — Furniture and Outdoor together are more than half this company's revenue. One glance at a sorted bar chart and you already know where Lantern and Pine's center of gravity is. This is why, later in the course, when you are unsure what chart to use, my standing advice will be: use a bar chart. Nobody has ever been criticized for one.

[Slide 15] Now the year's shape, and the answer to your prediction. Revenue and OrderDate, switched to a line chart. Power BI first draws a single point, because it groups dates into a hierarchy — year, quarter, month, day — so we expand down to months. And there it is. The biggest month of twenty twenty-five was May — two million, eighty-nine thousand dollars. December came second at two million, sixty-six thousand — twenty-three thousand dollars short. If you said December, your reasoning was good and the data disagrees, which is the most instructive thing data ever does. Why May? Think about the catalog: fire pits, patio sets, garden tools. Outdoor gear peaks in early summer, and it peaks hard enough to beat the holiday bump in everything else. February, for the record, is the floor — one point one four million, post-holiday, pre-summer. Nobody finds this by scrolling fifty thousand rows. Everybody sees it the moment the line renders.

[Slide 16] One more visual, and it is the one that makes a report different from a chart. Check Region and switch the visual to a slicer — a filter that lives on the page, owned by whoever is reading it. Click Online and watch: every visual on the page recalculates at once. That behavior is called cross-filtering. The card now reads nine point two two million — about forty-five cents of every revenue dollar never touches a store. Click Southwest instead and the same page tells a smaller story: one point seven six million, the quietest region in the company. Here is the shift I want you to register: a printed chart is read. This page is interrogated. The reader asks their own question — what about my region? — and the page answers instantly, with no analyst in the loop.

[Slide 17] Two finishing touches before we publish. First, the course theme. On the View ribbon, open the theme dropdown, browse for themes, and select the iris theme file from the course downloads. The entire page adopts the palette in one motion. A theme is a design decision made once instead of forty times, and this particular palette was built to be readable and safe for colorblind readers — about one in twelve of the people who will ever read your reports. Module Three turns that sentence into a skill you can defend in a design review.

[Slide 18] And now the last station of the loop. Home ribbon, Publish, sign in with your school account, choose My workspace. Desktop uploads the report to the Power BI Service, and the success dialog hands you a link. Open it and there is your report, in a browser, slicers working — anyone you share that link with reads the living report. Compare that with the old way: exporting a picture that freezes one moment in time, or emailing a file called final underscore v three underscore REAL to seven people, each of whom now has a different version. Desktop builds. The Service shares. You have now done both.

[Slide 19] One last discussion, because it is the shape of half the arguments you will referee in your careers. In this same data, Lighting sold the most units — eighteen thousand, six hundred ninety-five of them. Furniture made the most money — seven point six million dollars. So which one is our biggest category? Take a minute, pick a side, defend it — then find a pair who chose the other side and listen to them. The resolution is on the next slide, and I will tell you now: neither side is wrong.

[Slide 20] Here is the resolution. By revenue, Furniture is first. By units sold, Lighting is first with eighteen thousand six hundred ninety-five, and Furniture drops to fourth with fourteen thousand three hundred twenty-six. Both facts are simultaneously true, because inexpensive lamps sell in high volume for fewer dollars each. Neither is the answer, and that is the lesson — they are answers to different questions. Dollar rankings drive merchandising investment and sales targets. Unit rankings drive warehouse staffing, shipping contracts, and shelf space. When a manager asks which category is biggest, the honest analyst answers with both numbers and names the difference. You will do exactly this in the Hot practice this week.

[Slide 21] Four mistakes will account for most of the trouble in week one, and I want you to notice what they share: not one of them produces an error message. If OrderDate loads as text, your line chart sorts the months alphabetically — April, August, December — and it renders confidently, looking professional, and wrong. You are the only check. If a number surprises you and you keep staring at the canvas, you are asking the chart to confess; open Table view and look at rows instead. If you are in a browser and cannot find Get data, you are in the Service — building happens in Desktop. And when you submit an assignment, save the file, close it, reopen it, and confirm your work is actually in it. The grader reads what the file contains, not what you meant it to contain.

[Slide 22] Before we close, widen the lens. Everything you did today, some team does for a living. In retail it is literally today's questions — category mix, seasonality, store versus online, asked weekly by every merchandising team on the planet. In health care, the same line chart tracks patient volumes by clinic and month, and the surprising peak is a flu season instead of a fire-pit season. In finance, month-end reporting packs that once took a week of copy-and-paste now refresh themselves — which is why finance departments adopted this technology before almost anyone else. The tools change by industry less than you would think. The loop does not change at all.

[Slide 23] Here is what you did in one session. You rode the entire business intelligence loop — question, data, model, visuals, decision — on real data, ending with a published, interactive report. You learned the sentence that untangles Microsoft's naming: Desktop builds, the Service shares, Fabric surrounds. You met the implicit measure — automatic, convenient, and worth a healthy distrust you will justify in Module Two. You saw cross-filtering turn a page of charts into something a reader can interrogate. And you watched the data overturn a reasonable prediction: May beat December, because fire pits do not wait for the holidays. This week: the practice trio is open — Mild to warm up, Medium to stretch, Hot to argue with a vice president. The Lesson One quiz covers today's concepts, and Assignment A-zero-one asks you to rebuild today's page organized by region. Next session, we stop pretending data arrives clean: Power Query, and the file that fights back.